The Federal Communications Commission (FCC), which certifies every radio-emitting device (including routers) in the U.S., has announced “the addition of routers produced in a foreign country to the Covered List,” which is composed of equipment and services that “pose an unacceptable risk to the national security of the United States or the security and safety of United States persons.” While this is not technically a blanket ban on the sale of imported routers in the U.S., the FCC Public Notice (PDF) effectively has the same impact: The agency says it will not certify a foreign-made consumer router, making it illegal to sell or even import future new models into the country.

The agency said it’s doing this in response to a National Security Determination earlier this month, which says: “Recently, malicious state and non-state sponsored cyber attackers have increasingly leveraged the vulnerabilities in small and home office routers produced abroad to carry out direct attacks against American civilians in their homes. From disrupting network connectivity to enabling local networking espionage and intellectual property theft, foreign-produced routers present unacceptable risks to Americans.”

It also blamed foreign-made routers for the Volt, Flax, and Salt Typhoon cyberattacks that hit critical American infrastructure, adding that “routers in the United States must have trusted supply chains so we are not providing foreign actors with a built-in backdoor to American homes, businesses, critical infrastructure, and emergency services.”

While this might match the context of increasing instability in global geopolitics, it does not specifically address the weakness found in many consumer Wi-Fi routers. Although TP-Link was widely used in recent cyberattacks, cybersecurity experts told CNET that this was because of its ubiquity in the market, and that the exploited vulnerabilities were also present on routers made by American companies. In fact, the U.S. government itself said that the Salt Typhoon attacks often targeted Cisco hardware. Still, this did not deter the Department of Commerce from investigating TP-Link over its close ties to China.

This isn’t the first time that the FCC issued a directive that had a blanket effect on a specific type of device. In late December 2025, the agency made a similar move on foreign-made drones, effectively banning DJI and other imported brands from registering new models in the U.S. Many of the best Wi-Fi routers on sale in 2026 are from foreign companies like TP-Link. The FCC's new measure could, in theory, preclude future new products from these companies from coming to the States.

Nevertheless, this does not mean that foreign router manufacturers will forever be excluded from the U.S. market. The FCC says (PDF) that affected brands can apply for a “Conditional Approval” from the Department of War or the Department of Homeland Security to exempt them from the Covered List before they can build “trusted manufacturing capacity in the United States.” The move also does not affect any routers already on sale in the United States, or indeed any routers previously purchased, which will continue to work as normal.

“Teleporting quantum information is now a practical reality,” asserts Deutsche Telekom. The firm’s T‑Labs used commercially available Qunnect hardware to demo quantum teleportation over 30km of live, commercial Berlin fiber, running alongside classical internet traffic. In an email to Tom’s Hardware, Deutsche Telekom’s PR folks said that Cisco also ran the same hardware and demo process to connect data centers in NYC.

Many believed that readying quantum networks for a future quantum internet would require the deployment of new infrastructure. However, in both the Berlin and NYC demos, the qubits didn't travel through the existing fibers; they teleported from one end to another.

The demos

Last month, T-Labs completed the first practical test of the core components required for a future quantum internet, which would work by teleporting data. Central to the demo was Qunnect’s Carina platform, which integrates an entanglement generator, producing pairs of quantum-entangled photons for distribution over telecom fiber.

The experiment saw the recreation of an identical quantum particle at the destination “using pre-shared quantum entanglement rather than transmitting a physical particle,” explains Deutsche Telecom.

“Our fiber optic network is quantum ready,” said Abdu Mudesir, Telekom Board Member for Product and Technology. “In Berlin we have now proven that quantum information can be transmitted over 30 kilometers of commercial Telekom fiberoptics outside of a laboratory. “

There are a few wrinkles still to iron out, though. The official PR notes that the average accuracy of the teleported data is 90%. Deutsche Telecom and Qunnect also want to network quantum computers over longer distances, more locations, and multi-node teleportation configurations.

Still, it is pleasing that this milestone has been reached, with the building blocks of teleportation already operating across a real network. The teleportation wavelength used was 795nm, which is said to be a sweet spot for integration with platforms such as neutral-atom quantum computers, atomic clocks, and various quantum sensors.

A milestone

This milestone achievement shifts quantum communications from an experimental lab technology towards something telecoms providers can deploy. Key implications are expected in distributed quantum computing, quantum-secure communication, quantum sensor networks, and cloud-based quantum services.

As per the intro, Cisco used the same hardware from Qunnect to run a similar demo in NYC. We last reported on Cisco’s quantum internet efforts back in November last year, when it announced plans to jointly build a distributed quantum computing network capable of linking fault-tolerant systems over long distances, with the help of IBM.

Experts from Deutsche Telekom, Qunnect, and Technical University Dresden will be available for discussion at MWC Barcelona, on March 03 from 15:30 – 16:00 (CET). Deutsche Telekom will also have a ‘Quantum Teleportation’ showcase at its booth.

IBM and Cisco have announced plans to jointly build a distributed quantum computing network capable of linking fault-tolerant systems over long distances. In an announcement on Thursday, November 20, the companies said they aim to demonstrate a two-machine entanglement proof-of-concept by 2030, with the ultimate goal of enabling scalable quantum workloads that span multiple sites and processors. If successful, the collaboration would mark a shift in how quantum computing resources are deployed, moving beyond single-system scale to a federated architecture capable of trillions of quantum operations.

The initiative will combine IBM’s superconducting qubit hardware with new networking infrastructure from Cisco, including microwave-optical transducers, quantum network control layers, and physical and software routing protocols designed for entangled quantum state transmission.

The proposed architecture is intended to support fault-tolerant quantum computers already in IBM’s development roadmap. But it would also require the creation of new intermediary hardware — a planned ‘Quantum Networking Unit’, or QNU — to interface with IBM’s quantum processors and translate static quantum states into flying qubits suitable for transmission via photonic links.

The architecture IBM and Cisco want to build

The duo’s ambitions will be built upon a three-tier model that splits qubit modules, networking transduction interfaces, and optical entanglement layers. IBM’s Quantum Processing Unit (QPU) roadmap projects logical fault-tolerant machines with several hundred logical qubits. each requiring thousands of physical qubits, by 2030.

Cisco’s role is to link these cryogenic environments together. Entanglement between processors would be achieved using shared photon pairs or teleportation-style protocols, with photon-based carriers transmitted over optical fiber or potentially free-space links.

Because IBM’s superconducting qubits operate in the microwave scale, while long-distance transmission favors optical frequencies, a high-efficiency transducer is needed to convert quantum information from one format to another. That device — capable of preserving coherence and phase relationships between microwave and optical domains — will have to be developed and is one of the key technical hurdles of the roadmap.

The companies say the initial milestone will be to link two independent QPUs located in separate cryogenic systems. This will test both hardware entanglement and software synchronization layers. If successful, a scaled version of the architecture would allow for modular quantum computing networks, where computation is distributed across many small fault-tolerant nodes, and entanglement is dynamically allocated based on the structure of the problem being solved.

Why networks

IBM’s vision of scalable quantum computing has already shifted from single-monolithic machines toward what it calls quantum-centric supercomputing. Under that model, quantum processors function as accelerators embedded within larger high-performance compute environments, connected to CPUs, GPUs, and shared storage via classical interfaces. However, some workloads, especially those involving chemistry, material science, or cryptographic search, will require quantum circuits with hundreds of millions or billions of gates.

Running those circuits within the coherence window of a single device is infeasible, even under optimistic hardware timelines. Instead, IBM’s roadmap assumes inter-processor coordination, allowing large algorithms to be divided into subcircuits that can run on separate QPUs. This would enable workloads that exceed the qubit count or gate fidelity of any single machine.

The QNU plays a central role here, acting as the entanglement interface between QPUs. While some early experiments in microwave-to-optical transduction have been demonstrated in lab settings — including at Fermilab’s SQMS Center, where an IBM partnership is planned — the level of fidelity and error rate required for distributed fault-tolerant computing is still years away from production.

The companies are also working on software protocols that manage entanglement routing across the network. Unlike classical networks, where bits can be duplicated and retransmitted, quantum systems depend on ephemeral, one-time-use states. That means entangled links must be established just-in-time, managed through a new class of control protocols that coordinate not only logical dataflow but also the physical movement of qubit states. Cisco says it will contribute a high-speed software protocol framework to support these operations.

Quantum computing internet

The long-term vision goes well beyond inter-device communication. IBM and Cisco say their roadmap could extend into a future quantum internet where quantum processors and entangled photonic links form a planetary-scale network of physically distributed (but logically connected) resources.

The idea of a quantum internet has been proposed before. Several research groups have published designs for node-based or repeater-style architectures, but most of those are focused on specific applications such as quantum key distribution or secure messaging. IBM’s goal, however, is to make distributed compute a viable path for running quantum algorithms that can’t fit in memory on a single machine.

If achieved, it could allow new types of applications, from supply chain modelling to real-time climate simulation using quantum-enhanced sensing. IBM has suggested that such networks could support “trillions of quantum gates” across multiple QPUs, far beyond the practical limits of even a thousand-logical-qubit monolithic device.

A long road ahead

The 2030 timeline for demonstrating a basic entanglement between two QPUs is extremely ambitious. A scalable multi-node quantum network is expected to follow just a few years later, with long-distance networking only arriving in the latter half of that decade. The quantum internet vision, where processors and entangled repeaters span entire regions, is likely more than 15 years out.

Some significant engineering challenges will need to be overcome to make this timeline a reality. No existing transducer meets the required efficiency and fidelity thresholds for scalable links. Meanwhile, distributed quantum error correction is still in development, and most of the proposed network protocols are theoretical or exist only in research simulations. There is also the challenge of integrating Cisco’s photonic networking expertise with IBM’s cryogenic systems in a way that minimizes thermal interference and maximizes link yield.

After more than a decade of pushing processor design, IBM is now turning its attention to interconnects. Cisco, for its part, is betting that quantum computing will need entirely new systems thinking, where classical routing is blended with real-time entanglement management.

It is a different way to think about infrastructure, not just as a transport layer, but as a co-designed part of the computational pipeline itself. If IBM and Cisco can build it, they’ll be reshaping what it means to run a program when the processor is no longer a single machine.

The U.S. Cybersecurity & Infrastructure Security Agency (CISA) released guidelines for network engineers, defenders, and organizations with enterprise-grade networking equipment as part of its continuing investigation into the People’s Republic of China’s (PRC’s) massive cyber espionage campaign. While the agency has released general guidelines for increased visibility and hardening that apply to any network, it also had a section labeled ‘Cisco-Specific Guidance.’

According to the document, “authoring agencies have observed Cisco-specific features often being targeted by, and associated with, these PRC cyber threat actors’ activity.” Aside from the recommendations mentioned, CISA also linked to Cisco’s IOS XE Hardening Guide and Guide to Securing NX-OS Software Devices. This shows that the networking company is likely aware of its vulnerabilities and is taking steps to help protect its customers from those who need it without removing features that make it easier for others who don’t need more stringent security to use its products.

CISA acknowledges Cisco and Google Cloud Security in the guideline document, which shows how the private sector cooperates with the U.S. government to help protect its network systems. Furthermore, this warning isn’t limited to the U.S., as other cybersecurity and counter-espionage agencies from other allied countries, specifically Australia, Canada, and New Zealand, are also participating in the investigation that the U.S. is leading.

This warning was made about a month after the U.S. CISA announced that PRC-affiliated actors were targeting eight commercial telecommunications providers across the U.S., which was suspected to have started as far back as 2022. It said the attackers exfiltrated customer call records, compromised the private communications of some high-value targets in government and politics, and copied information related to U.S. court proceedings.

The good news is that these activities are seemingly bound to the existing weaknesses of the target infrastructure, which are known to the authorities and manufacturers of the affected devices, which seem to be mostly Cisco networking equipment. CISA says that you could secure your network and prevent being targeted by these threat actors by patching the affected devices and services and ensuring your environment by following its released guidelines.

AMD, Broadcom, Cisco, Google, HPE, Intel, Meta, and Microsoft have joined forces to develop Ultra Accelerator Link (UALink), a new industry standard to enable high-speed, low-latency interconnection for datacenter grade AI and HPC accelerators. UALink will allow for interconnecting up to 1,024 accelerators within one pod, which would be a major achievement. The UALink technology will essentially compete against Nvidia's NVLink, so the green company is not participating in its development.

The UALink initiative is designed to create an open standard for AI accelerators to communicate more efficiently. The first UALink specification, version 1.0, will enable the connection of up to 1,024 accelerators within an AI computing pod in a reliable, scalable, low-latency network. This specification allows for direct data transfers between the memory attached to accelerators, such as AMD's Instinct GPUs or specialized processors like Intel's Gaudi, enhancing performance and efficiency in AI compute. 

"The work being done by the companies in UALink to create an open, high performance and scalable accelerator fabric is critical for the future of AI," writes Forrest Norrod, executive vice president and general manager, Data Center Solutions Group at AMD in the press release. "Together, we bring extensive experience in creating large scale AI and high-performance computing solutions that are based on open-standards, efficiency and robust ecosystem support. AMD is committed to contributing our expertise, technologies and capabilities to the group as well as other open industry efforts to advance all aspects of AI technology and solidify an open AI ecosystem." 

AMD, Broadcom, Google, Intel, Meta, and Microsoft all develop their own AI accelerators (well, Broadcom designs them for Google), Cisco produces networking chips for AI, while HPE builds servers. These companies are interested in standardizing as much infrastructure for their chips as possible, which is why they are teaming up for this the UALink Consortium. Since Nvidia has its own infrastructure, it is naturally not interested in co-developing UALink.

By standardizing the interconnect for AI and HPC accelerators, it will be easier for system OEMs, IT professionals, and system integrators to integrate and scale AI systems in datacenters. The standard aims to promote an open ecosystem and facilitate the development of large-scale AI and HPC solutions.

"UALink is an important milestone for the advancement of Artificial Intelligence computing," said Sachin Katti, SVP & GM, Network and Edge Group, Intel. "Intel is proud to co-lead this new technology and bring our expertise in creating an open, dynamic AI ecosystem. As a founding member of this new consortium, we look forward to a new wave of industry innovation and customer value delivered though the UALink standard. This initiative extends Intel’s commitment to AI connectivity innovation that includes leadership roles in the Ultra Ethernet Consortium and other standards bodies." 

The UALink Consortium will be established to oversee the development and implementation of the UALink standard. The consortium is expected to be incorporated by the third quarter of 2024, aligning with the release of the 1.0 specification. Companies that join the consortium will have access to the specification and can contribute to its ongoing development.

Florida-based Onur Aksoy, who owns a group of companies that operated several online stores, was sentenced to six and a half years in prison after being proven guilty of selling $100 million worth of counterfeit Cisco network devices to several government, school, and military organizations. He also sold these products as new via Amazon and eBay customers in the U.S. and internationally. 

According to court documents, Aksoy, who also held dual citizenship with Turkey, imported many modified network products via China and Hong Kong. He ran the operation under the name 'Pro Network Entities,' which had nineteen companies and 25 eBay and Amazon storefronts. Upon investigation, it was found that the networking gear used components from the older models modified to look like the newer, expensive devices. 

The products were also packed in counterfeit Cisco boxes containing documentation, labels, stickers, and other materials to make the product look genuine. As one would expect, these devices had performance issues that often led to failures, affecting many critical operations usually engaged by several government agencies. The fraudulent acts were even more serious because the devices were used in many combat and non-combat operations, leading to a multi-year investigation.

"Through an elaborate, years-long scheme, Aksoy created and ran one of the largest counterfeit-trafficking operations ever," said Attorney for the United States Vikas Khanna for the District of New Jersey. "His operation introduced tens of thousands of counterfeit and low-quality devices trafficked from China into the U.S. supply chain, jeopardizing both private-sector and public-sector users, including highly sensitive U.S. military applications like the support platforms of U.S. fighter jets and other military aircraft." 

The Almost Ten-Year Investigation

The US Customs and Border Protection were able to intercept 180 shipments of counterfeit goods between 2014 and 2022, but Aksoy used fake aliases and delivery addresses to avoid scrutiny. Later, he instructed his offshore suppliers to send the goods in smaller packages to reduce suspicion. These confiscations did not deter Aksoy from stopping this operation, as the counterfeit products were 98% cheaper than their original counterparts, making it too lucrative for him to give up. Because of such tactics, the authorities took a long time to associate these goods with Aksoy's companies. The Department of Justice also said via a press release that Cisco did send him seven cease and desist letters for trafficking counterfeit goods. 

Eventually, in 2021, agents were awarded a search warranty against Akshoy's warehouse, where about 1,156 of these devices were found, valued at over $7 million. In June 2023, he pleaded guilty to conspiring with others to traffic in counterfeit goods and to commit mail fraud, wire fraud, and mail fraud. The case ended with Aksoy receiving a six-year prison sentence and being required to pay Cisco $100 million as restitution following the destruction of the counterfeit goods. 

Despite the success, we should question the government agency's verification method for whitelisting such sellers who provide devices for sensitive government and military operations. The whole ordeal is reminiscent of another story involving two weapons dealers that inspired the movie 'War Dogs.' 

The discovery and destruction of counterfeit goods are common globally; however, discovering the use of such devices will always be a cause of concern, knowing that the US government has banned goods from certain companies for national security reasons. 

Intel showcased its Gaudi3 Processor for artificial intelligence (AI) workloads alongside the formal introduction of its 14th-Gen Core Ultra "Meteor Lake" processors and 5th-Gen Xeon Scalable CPUs for datacenters. The accelerator is set to arrive in 2024 and will offer a significant performance bump compared to its predecessor, the Gaudi2.

Intel CEO Gelsinger discussed the upcoming release of Intel's Gaudi3, which is scheduled for next year, and showed off the new AI accelerator aimed at deep learning and large-scale generative AI models for the first time. The new unit looks like a huge module (OAM, we presume) with a massive ASIC and multiple HBM3 (or HBM3E) memory stacks on it. The ASIC package looks significantly larger than the Gaudi2, so we presume that it is equipped with eight HBM3E stacks on it (rather then with six in case of Gaudi2). Based on a slide Intel presented at SC23, Gaudi3 is not a monolithic processor, but rather a dual-chiplet design that fuses together two processors.

In addition, Intel announced that the Gaudi3 will offer four times higher BF16 performance, two times faster networking performance, and 1.5X higher bandwidth compared to Gaudi2.

"Our Gaudi roadmap remains on track with Gaudi3 out of the fab, now in packaging and expected to launch next year," said Pat Gelsinger, CEO of Intel, at the company's latest conference call (via SeekingAlpha). In 2025, Falcon Shores brings our GPU and Gaudi capabilities into a single product."

Intel's Habana Gaudi2 already is a quite promising product, with 24 fully programmable Tensor Processor Cores (TPCs) and 96GB of HBM2E memory, capable of challenging Nvidia's H100 GPU for AI and HPC. Intel's Habana Gaudi3, which is expected to hit the market in 2024, will offer significantly improved performance ove its predecessor, Gelsinger said earlier this year.

Intel said the Gaudi line-up has experienced significant growth — attributed to its proven performance and competitive TCO, as well as its reasonable pricing. The company is confident that the rising demand for generative AI hardware will position Intel to secure a more substantial share of the accelerator market in 2024, primarily through its range of AI accelerators, which will be spearheaded by Gaudi.

"We are pleased with the customer momentum we are seeing from our accelerator portfolio and Gaudi in particular, and we have nearly doubled our pipeline over the last 90 days," said Gelsinger on the call.

AMD chief executive Lisa Su along with two others notified Cisco that they would like to leave its board of directors. Su joined Cisco's board of directors in early 2020, but a lot of things have happened since then, which might be a reason why she decided to quit.

"On October 4, 2023, M. Michele Burns, Roderick C. McGeary, and Dr. Lisa T. Su each notified Cisco Systems, Inc. of their respective decision not to stand for re-election at Cisco's 2023 annual meeting of stockholders," Cisco's filing with SEC reads (via Don Clark). "Ms. Burns, Mr. McGeary, and Dr. Su will continue to serve as directors until the 2023 Annual Meeting." 

Neither disclosed reasons for leaving Cisco's board of directors, and we can only speculate about their intentions.

Of the three, Lisa Su is the only leader of high-technology company. When she joined Cisco's board-of-directors in 2020, AMD was beginning to recapture market share in PCs and servers from Intel and expansion prospects were rather foggy. But over the course of three years AMD got a lot bigger. 

The company purchased Pensando for $1.9 billion in 2022 to gain data processing units (DPUs), a type of processors that enables connectivity in datacenters and somewhat competes against Cisco's datacenter connectivity products. In addition, AMD recently launched its EPYC 8000-series 'Siena' processors that target communications and edge servers, thus also encroaching on Cisco's offerings. As a result, Su has at least two conflicts of interests when it comes to Cisco. 

Keep in mind that at this point that this is mere speculation. Before joining Cisco's board, Su left the board of Analog Devices.

Earlier today, the Linux Foundation shared a new press release detailing plans for the Ultra Ethernet Consortium (UEC). This effort involves several prominent industry leaders, including AMD, Arista, Broadcom, Cisco, Eviden, HPE, Intel, Meta, and Microsoft. The consortium intends to optimize Ethernet standards for high-performance networking.

As more advancements are made in artificial intelligence, machine learning and other forms of high-performance computing, so are the demands of the networks that support these systems. The goal of UEC is for the involved companies to work together in improving the standards of modern networking protocols so they might be better equipped to handle the workloads of these newer advanced systems.

Dr. J Metz is Chair of the Ultra Ethernet Consortium and stated that there is no need to overhaul Ethernet, but instead, the team hopes to make adjustments over time that will improve efficiency.

So far, a few specific technical hurdles have been identified, which the consortium intends to focus upon. This involves the development of APIs and specifications for various Ethernet communication protocols, interfaces and data structures to suit the needs of high-performance technology better.

Additional areas of interest include both link-level and end-to-end network transport protocols. The press release also mentions congestion, signaling mechanisms, and telemetry as necessary targets. What this looks like in practice will only be apparent in time, but we can expect things like new software and Ethernet security protocols developed with AI, machine learning, and high-performance computing in mind.

You can read more about the UEC on their official website, including frequently asked questions and news and updates. The latest press release also provides plenty of information about forming the new consortium and its future goals.

Intel recently held a launch event for a new China-specific version of its Gaudi 2 processor for artificial intelligence (AI) workloads, and this new version is compliant with the latest US sanctions on China-bound AI and GPU processors. The new version cuts down the card's scale-up interconnect, which somewhat limits the performance of AI servers on its base. To some degree, this will help with the massive shortages of AI GPUs, such as Nvidia's A100 and H100, as well as the cut-down A800 versions that Nvidia has also created specifically for its China customers.

Intel has two Gaudi 2 versions that come in the 600W OCP model form factor. The regular version, called Gaudi 2 HL-225H, has 24 x 100Gbps RDMA (RoCE v2) interconnects, whereas the cut-down version of the product, called Gaudi 2 HL-225B, has 21 x 100Gbps RDMA (RoCE v2) scale-up interconnects. The reduced number of scale-up interfaces reduces the performance of AI servers based on this version of Gaudi 2, but keeping in mind the huge demand for AI capabilities in China, it will likely be popular regardless.

Intel describes the Gaudi 2 accelerator's version for the Chinese market as 'The HLB-225B Processor complies with US BIS regulations' for supercomputers and semiconductors.

While the availability of Intel's Habana Gaudi 2 in China will ease shortages of GPU servers, it should be noted that it isn't likely that applications that already use Nvidia's GPUs will be ported to Gaudi. Therefore, Nvidia's existing customers will have to keep buying from Nvidia.

"On July 11, Intel held an event for customers, partners and local media in the China market," an Intel spokesperson said. "As part of the event program, Intel provided attendees with updates on our AI strategy, our unique portfolio of AI products and announced the availability of Gaudi2 for customers in China. The availability of Gaudi2 in China continues Intel’s nearly 40-year history of delivering innovative yet legally-compliant products to this key growth market."

According to DigiTimes, TSMC is accelerating orders with backend equipment suppliers as it starts an expansion plan for its chip-on-wafer-on-substrate (CoWoS) packaging capacity. The shortage of compute GPUs for artificial intelligence and high-performance computing, which Nvidia largely dominates, is mainly attributed to TSMC's limited CoWoS packaging production capabilities.

Reports suggest that TSMC has plans to increase its current CoWoS capacity from 8,000 wafers per month to 11,000 wafers per month by the end of 2023 and then to around 14,500 – 16,600 wafers by the end of 2024. Previously Nvidia was rumored to increase its CoWoS capacity to 20,000 wafers per month by the end of 2024. Remember that the information comes from unofficial sources and may be inaccurate. 

Major tech giants like Nvidia, Amazon, Broadcom, Cisco, and Xilinx have all boosted their demand for TSMC's advanced CoWoS packaging and are consuming every wafer they can get. As a result, TSMC has been forced to renew orders for necessary equipment and materials, according to DigiTimes. The production of AI servers has significantly risen, fueling the already intense demand for these advanced packaging services. 

Nvidia has already booked 40% of TSMC's available CoWoS capacity for the coming year. However, due to the severe shortage, Nvidia has started exploring options with its secondary supplier, placing orders with Amkor Technology and United Microelectronics (UMC), although these orders are relatively small, the report claims. 

To cater to its increased CoWoS packaging needs, TSMC is partnering with multiple suppliers from around the globe, including U.S.-based Rudolph Technologies, Japan's Disco, and Germany's SUSS MicroTec, along with Taiwanese experts Grand Process Technology (GPTC) and Scientech. The suppliers are under pressure to provide almost 30 sets of relevant tools by mid-2024. 

TSMC has also begun implementing strategic changes, such as redistributing some of its InFO production capacity from its northern Taiwan site in Longtan to its Southern Taiwan Science Park (STSP). It's also fast-tracking the expansion of the Longtan site. Furthermore, TSMC is increasing its in-house CoWoS production while outsourcing part of its oS manufacturing to other assembly and test (OSAT) companies. For instance, Siliconware Precision Industries (SPIL) has been one of the beneficiaries of this outsourcing initiative.

TSMC opened its Advanced Backend Fab 6 facility last week. It is set to expand its advanced packaging capacity for its frontend 3D stacking SoIC (CoW, WoW) technologies and backend 3D packaging methods (InFO, CoWoS). For now, the fab is ready for SoIC. The Advanced Backend Fab 6 can process around one million 300-mm wafers per year and carry out over 10 million hours of testing annually, with cleanroom space that is larger than the combined cleanroom spaces of all other TSMC advanced packaging facilities.

Among the most impressive features of the Advanced Backend Fab 6 is the extensive five-in-one intelligent automated material handling system. The system controls the production flow and detects defects instantly, increasing yield. This is crucial for complex multi-chiplet assemblies like AMD's MI300, as packaging defects immediately render all chiplets unusable, leading to significant losses. With data processing capabilities 500 times faster than average, the facility can maintain comprehensive production records and track every die it processes.

Nvidia uses CoWoS for its highly successful A100, A30, A800, H100, and H800 compute GPUs. AMD's Instinct MI100, Instinct MI200/MI200/MI250X, and the upcoming Instinct MI300 also use CoWoS.

Cisco was one of the first high-tech companies to halt its business in Russia and Belarus after the former started its full-scale war against Ukraine on February 24, 2022. During the company's exit, Cisco's Russian subsidiary physically destroyed unsold equipment worth 1.9 billion rubles ($23.5 million), reports CNews citing the state-run Tass agency.  

Network giant Cisco destroyed its equipment stocks worth almost 1.9 billion rubles ($23.5 million) in January 2023, months after announcing its decision to discontinue operations in Russia and Belarus in March 2022. The physically destroyed equipment included mainly spare parts for the company's devices that were not sold due to the current policy imposed by Cisco. Meanwhile, it is unclear whether those spare parts also fall under the U.S. sanctions against Russia. 

Despite the financial impact, Cisco made the decision to destroy the equipment rather than allow it to fall into Russian hands. But while $23.5 million sounds like a lot of money, the actual impact seems to be considerably higher. 

While the Russian news agency claims that the equipment was destroyed in January, Cisco's financial report for the quarter that ended January 28, 2023, does not mention this in any way. This is because the company essentially wrote off $67 million worth of assets, including spare parts already imported to Russia as well as furniture and cars, after conducting a risk assessment of its assets and potential financial exposures in Russia and Belarus as early as calendar Q1 2022. 

Based on the assessment, Cisco reserved for the non-recoverability of most of its assets in these countries. Consequently, it recorded non-recurring charges of $67 million in its cost of sales and operating expenses in the third quarter of fiscal 2022. These charges were related to the non-recoverability of certain assets and special personnel-related expenses aimed at supporting affected employees. 

It should be noted that for Cisco, which reported revenue of $13.592 billion and net income of $2.773 billion in the second quarter of fiscal 2023, $23.5 million worth of equipment and $67 million in various assets are relatively negligible sums of money. 

Here's some bad news: Cisco CEO Chuck Robbins told Yahoo Finance Live today that he thinks it could take "a couple years" for the ongoing chip shortage to finally end.

Here's some worse news: Intel CEO Pat Gelsinger agreed with Robbins. "We can't build fabs overnight," he told Yahoo, "it takes a couple of years to get built up."

This isn't the first we've heard about the shortage lasting more than a few quarters. AMD CEO Lisa Su issued a similar warning in January when she said the company simply couldn't keep pace with the higher-than-expected demand for its products.

Analysts also predicted in February that the chip shortage would continue for at least a year because, by their estimates, demand outpaces supply by about 30%. That's a significant gap that would take fabs several quarters—at least—to close.

Asus and MSI also said earlier this month that supplies of Nvidia GPUs, including some that power the best graphics cards,  fell even further in the first quarter of 2021. MSI chairman Joseph Hsu told investors that he expected graphics cards prices to rise throughout the year because of that shortage.

Many forward-looking statements have been fairly hopeful for 2022, however, as fabs make significant investments to increase production capacity. But that doesn't mean the problem will be solved; it simply means it won't be as bad as it is now.

"I think we see a few quarters of real stress in the supply chain, and we think that it will be more predictable. It may not be where we want it to be, but it will be more predictable," Robbins told Yahoo, adding "We just got to fight our way through it."

It's kinda like turning a doorknob that shocks you when you touch it. The first time is the worst because it comes as such a surprise, but eventually you can start to plan around it. Maybe you wear rubber gloves; maybe you just steel yourself for the pain.

Put another way: If you're aware, you can prepare, and now companies are very aware that the chip shortage isn't going to end when the calendars turn over. It seems that Cisco at least has embraced this reality; others will likely follow suit.

Cisco announced today that it had appointed AMD CEO Lisa Su to its Board of Directors. Su, like many other executives, has served on boards at other companies before: The move comes after Su stepped down from the Board at Analog Devices in December. 

We reached out to AMD for more information, and the company indicates that Su will stay in her position as AMD CEO and on AMD's own Board of Directors. That means we shouldn't see any major changes to the company's leadership or objectives. 

The announcement comes on the eve of AMD's financial earnings that will be released tomorrow, but Cisco's press release is light on details and doesn't include any information about Su's compensation package. Per Cisco:

"AMD President and CEO Dr. Lisa Su Appointed to Cisco Board of Directors.

Cisco today announced the appointment of Dr. Lisa T. Su, AMD president and CEO, to its board of directors effective today."

The press release includes several details of Su's previous work and notes that she continues to serve on the AMD board as well, a position she has held since 2014.  

Cisco, a networking hardware, software, and telecommunications giant, currently has a market cap of $201.38 billion. For comparison, AMD currently has a market cap of $54.68 billion. 

Notably, Cisco is a key competitor for Mellanox, which is currently in the midst of a $6.9 billion acquisition by Nvidia. That acquisition is still working its way through regulatory channels and awaits approval from China's MOFCOM organization, which might be a dicey proposition given the ongoing trade war. In either case, Su and Nvidia CEO Jensen Huang might find themselves on opposite sides of the ball once again, but this time in the networking market. 

Cisco is deeply involved in the networking field, but also has a line of Unified Computing System (UCS) servers that predominantly feature Intel's silicon, although a few models already come equipped with AMD's EPYC processors. However, Su's new position on the board shouldn't imply any increased cooperation between the two companies beyond the existing projects. 

Lisa Su also serves on the board of directors for the Global Semiconductor Alliance (chairperson) and the Semiconductor Industry Association. 

It seems like everything needs a brand these days. This can make it easy to refer to something that would otherwise be cumbersome to identify, which is why many security researchers have started naming their discoveries, but it can also lead to some confusion. That's exactly what happened earlier this week when the Microsoft Defender Advanced Threat Protection Research Team and Cisco Talos gave the same malware two different names.

Microsoft named the malware Nodersok; Cisco Talos named it Divergent. Regardless of what it's called, the new malware uses "living-off-the-land techniques" that repurpose legitimate tools for nefarious purposes. Those repurposed tools are called "living-off-the-land binaries," or LOLBins for short, and they allow this so-called fileless malware to evade the detection features employed by the vast majority of Windows security products.

Here's what Microsoft said about Nodersok's method of infection:

"Like the Astaroth campaign, every step of the infection chain only runs legitimate LOLBins, either from the machine itself (mshta.exe, powershell.exe) or downloaded third-party ones (node.exe, Windivert.dll/sys). All of the relevant functionalities reside in scripts and shellcodes that are almost always coming in encrypted, are then decrypted, and run while only in memory. No malicious executable is ever written to the disk."

In this malware's case, it installs Node.exe and WinDivert as its LOLBins. These are legitimate apps: the former is "the Windows implementation of the popular Node.js framework used by countless web applications," as Microsoft put it, while the latter is "a powerful network packet capture and manipulation utility." Both are typically harmless, but their features allowed Nodersok's creators to establish their fileless malware.

Microsoft said it saw the first indicators of Nodersok in mid-July, and that "it's been pestering thousands of machines in the last several weeks, with most targets located in the United States and Europe." Most of the affected systems are consumer devices. Cisco Talos said it believes the malware "is currently under active development" because it "has observed multiple versions of the loader being used to install" the malware with two names.

The malware's name isn't the only thing Microsoft and Cisco Talos can't agree on. While they agreed that Nodersok / Divergent spread via malicious ads that forced a download onto a system that could then install the required LOLBins, they differed on the malware's purpose. Microsoft thought it was to relay malicious traffic; Cisco Talos claimed the malware's operators wanted to use it for click fraud. More info can be found in the companies' disclosures.

The NDA has expired, our full coverage, including SKUs and pricing, is here.

AMD's CEO Dr. Lisa Su is set to take to the stage here in San Francisco to unveil the company's new 7nm EPYC Rome processors. We'll cover the event here live, but keep your eyes peeled for the full breakdown in a few hours. For now, refresh your browser frequently for updates.

Su discussed the problems facing data center administrators, such as the need to process ever-increasing amounts of data and to deal with the increasing number of security threats.

AMD's Naples chips debuted in 2017, with the goal of providing more cores, more memory bandwidth, and more I/O at every price point. Delivering on those goals required a robust ecosystem of hardware partners. Su said AMD has over 60 platforms in the market, with the number increasing daily.

AMD has also invested heavily in fostering cloud-based instances, which now number at 50 around the world. AWS introduced its first EC2 instances in November of last year.

Su played up the company's line of 7nm products, which includes Zen 2 processors and the Radeon RDNA architecture. Pairing the 7nm process with the chiplet-based Zen 2 architecture allows the company to increase density and lowers cost.

Su said the 2nd-gen EPYC processor is the highest-performance x86 processor in the world. The company will present 80 world records here for the audience.

Su said that EPYC is the new standard for the modern data center. It has the highest core count ever of 64 cores and 128 threads, the most I/O with 128 lanes of PCIe 4.0, and higher boost frequencies than the previous-gen models. 

Su pointed out that the data center has had incremental improvements over the last decade, but AMD looks to change that paradigm.

Su said the company has almost double the performance of Intel's Cascade Lake processors.

AMD claims to have 97% higher performance than Intel's 8280, it's $10,000 flagship, at a lower cost.

Su said EPYC provides from 2x more performance per dollar than Xeon, and up to 4x.

That equates to lower operational and capital expenditures.

Su said it takes a robust ecosystem to bring these platforms to market, and its partners are here to tell us about their new systems.

Su invited an HPE representative to the stage to talk about the company's new Rome systems. HPE has three systems available today, the DL25, the DL35, and the Apollo 35. HPE will have 12 Rome systems in its portfolio by this time next year.

HPE set 37 world records with its EPYC Rome servers. The company also discussed its new security initiatives and how AMD's EPYC Rome has an in-built security co-processor that helps secure servers and data at rest.

Jen Frasier, Senior Engineer at Twitter, came to the stage to talk about how the company uses AMD's EPYC processors. Twitter processes hundreds of millions of tweets per day, which requires scaling up to more powerful processors.

Frasier said the Rome processors have reduced the amount of power the company consumes per core, which reduces cost. Rome's increased density allows Twitter to pack in 40% more cores per rack, and 25% lower TCO. Twitter will roll out Rome in its data centers in 2019.

Su and Frasier took a selfie with the audience before Frasier left the stage.

Mark Papermaster came to the stage to talk about the move to the 7nm process node. AMD needed to work closely with fab, design, and EDA partners to bring the designs to market. Rome provides double the number of cores in a similar TDP envelope as its previous-gen EPYC Naples processors.

Papermaster said that spreading Zen 2's 15% IPC improvement across 32 cores increases it to 23%. That's an odd measurement. Papermaster also outlined the improvements to the Zen 2 microarchitecture, but all of these advances are known. We'll cover these details in our full breakdown that will post shortly at NDA lift.

AMD moved to a second generation of the Infinity Fabric, just like we see with the Ryzen 3000-series processors. This link is also used for socket-to-socket communication in servers. Papermaster displayed the new chip layout, with a 12nm I/O die serving as the lynchpin for eight 7nm eight-core compute dies.

Papermaster talked about the benefit of faster I/O throughput borne of the new PCIe 4.0 interface.

Papermaster claims that EPYC Rome performance scales well based on core counts and memory throughput, even out to two-socket configurations, which indicates the cores aren't memory-bandwidth-starved.

A Dell representative came to stage and talked about the company's first-gen EPYC Naples products, and now is moving forward to Rome processors. The parade of partners continues. VMWare's Krish Prasad came to stage to talk about how the company is working with AMD.

AMD has completed the design phase of Zen 3, and already has Zen 4 in the design phase.

Forest Norrod came out to talk about Rome improvements. The biggest Rome processors come with up to 32 billion transistors spread across 1000mm² of silicon.

Norrod also pointed out that Rome works wonderfully with Java code, and showed some benchmarks to back it up.

Cray came to stage and announced that it has sold over a billion dollars of Cray supercomputers powered by EPYC processors. One of those big sales includes the money make to power the Frontier supercomputer. Cray announced a new Cray supercomputer for the US Air Force for weather forecasting. Cray also announced that Indiana University will deploy a new "Big Red 200" Cray supercomputer powered by EPYC.

Norrod said that a single-socket Rome server can outperform almost Intel's entire product stack -- but when they are in dual-socket servers. That's quite the claim.

Finally, Su returned to stage and invited Google's Bart Sano, the Vice President of engineering, out to stage. He announced that Google already has Rome silicon working in its data centers in a production environment. Google plans to widen the deployment over time.

Google will make Rome processors available via VMs on Google Cloud.

Su wrapped up the event and closed the show by thanking her employees, the industry partners and customers, but she left us with one point.

AMD has completed Zen 3 that is going into Milan, and the company is already working on Zen 4 to power Genoa.

The NDA has expired, our full coverage, including SKUs and pricing, is here.

Cisco has settled a lawsuit over claims that it sold video surveillance technology that it knew was vulnerable to a four-year-old flaw. The vulnerability could have allowed malicious parties to hack into cameras that Cisco had been selling to U.S. hospitals, airports, schools, police departments, state governments and federal agencies.

According to a settlement unsealed Wednesday with the U.S. Justice Department, 15 states and the District of Columbia, Cisco learned about the vulnerability for the first time back in 2008, when whistleblower James Glenn came forward and revealed the flaw. However, Cisco waited four years before doing anything about it. In the meantime, the company kept promoting its vulnerable product.

Cisco’s surveillance technology was also connected to door locks and alarms, and those could have also been bypassed due to this flaw.

Michael Ronickher, one of Glenn’s attorneys, said that the flaw was easy to exploit:

"It was like the moment in the heist movies when a person types on a laptop for 30 seconds and says 'I'm in.'"

Cisco said that there was no evidence that the flaw has been abused. Ronicker agreed with that statement but also noted that it’s possible hackers abused the flaw without being detected.

For its first time, Cisco had to settle under the whistleblower law for not having adequate security protections. The Justice Department learned about the flaw as it was reviewing many of the multi-billion dollar contracts that may not have prioritized cyber security. With the rise of ransomware and it disabling and holding hostage hospitals and police departments, cybersecurity issues have become a much more pressing issue for the U.S. government.

The federal government and the state governments that joined the settlement with Glenn will get 80% of the $8.6 million, while Glenn and his attorneys will get 20%. This should leave Glenn with more than $1 million for his whistleblowing act after fees and expenses, which is still significantly more than what most bug bounties would pay.

Cisco and Acacia Communications announced that they have entered into a definitive agreement whereby Cisco will acquire Acacia for $2.6 billion. Acacia is Cisco's existing supplier of optical interconnect technologies.

Cisco intends to continue supporting Acacia’s existing customers and any new customers, the company said. However, the acquisition fits in the company’s strategy of “reinventing every domain of the network” with its intent-based architectures and enrich the company’s optical systems offerings. Acacia, for its part, thinks it can “accelerate the trend toward coherent technology and pluggable solutions while accommodating a larger footprint of customers worldwide.”

David Goeckeler, executive vice president of Cisco’s networking and security business further notes: “With the explosion of bandwidth in the multi-cloud era, optical interconnect technologies are becoming increasingly strategic. The acquisition of Acacia will allow us to build on the strength of our switching, routing, and optical networking portfolio to address our customers’ most demanding requirements.”

EETimes notes that Cisco views coherent optical networks as a multi-billion dollar market in its early stage. Coherent optical networks use modulation of the amplitude and phase of light, with transmission across two polarization and DSP used at both transmitter and receiver. About a quarter of Acacia’s sales are DSPs and photonic integrated circuits, with the rest being photonic modules. Acacia also sells 600G transceivers.

The deal is subject to customary closing conditions and regulatory approvals. Cisco expects to close the acquisition in the first half of 2020.

Facebook is a place to catch up with friends, stay in touch with family and find hundreds of thousands of cyber criminals for hire. That last one is according to Cisco Talos, the networking giant's threat intelligence group, which today detailed how the social network serves as a forum for digital ne’er-do-wells.

In a blog post, Talos company said it found 74 groups devoted to cyber crime with a collective total of 385,000 members. Some of the groups were active for up to eight years, Talos said, and many sported obvious names like “Facebook hack (phishing)” or “Spammer and Hacker Professional.” Welcome to the cyber criminal not-so-underground.

There is a certain degree of inevitability when it comes to Facebook’s platform being misused. It turns out that criminals need to socialize, too, and in some cases advertise their services. Having more than 2 billion users makes Facebook the ideal place to do both.

Talos said that it tried to get the groups it found taken down by using the social network’s “report abuse” tool. Some groups were closed; others merely had specific comments removed. When Talos reached out to Facebook’s security team, though, all of the groups were finally taken down. Not that bringing down these groups accomplished a whole lot in the long run. Similar groups were formed after the initial batch were shuttered, and we suspect that pattern would continue as long as Facebook’s approach resembles Whac-a-Mole. Bopping them just makes ‘em pop back up later.

This problem isn’t exclusive to cyber crime. Facebook’s platforms (Facebook, Messenger, Instagram and WhatsApp) have long been used to sell drugs, guns and illegal services. The company moderates them, sure, but more of them continue appearing. It makes sense for that problem to extend from selling pot and .22s to phishing services and stolen credit cards.

Here’s what Talos thinks we should do about that:

“To combat these motivated adversaries, we need to work together. Social media platforms should continue their efforts, both manual and automated, aimed at identifying and removing malicious groups. Security teams and vendors must work together to actively share information, take action and inform our customers. Businesses need to be diligent about their protection and cyber hygiene efforts. And finally, consumers need to become as informed and skeptical as possible. Attacks like spam prey on the individual as an entry point.”

The company also advised individual users to continue using the “report” button on Facebook. Because that worked out so well when Talos did it, right?

Google made waves today at the Game Developers Conference (GDC) in San Francisco with a gaming announcement of its own: Stadia, a streaming game service. Stadia will launch in 2019 in the U.S., Canada, UK and Europe. Google did not share pricing information.

Stadia's architecture uses Google's data center network, which consists of fiber optic links and under-sea cables that have supported Google searches for years with 7,500 edge nodes. At launch, Stadia will support resolutions up to 4K at 60 frames per second (fps) with surround sound. In the future, Google is promising 8K up to 120 fps. It will allow for cross play between systems, run on a data center with custom GPUs and work with a controller that Google is making for the service.

Besides the stream you get, a separate one will be made at 4K and 60 fps to stream to YouTube.

Google worked with AMD to custom design a GPU with more than 10 teraflops of power to handle the service. It will run on Linux and Vulkan and work on the Unreal Engine and Unity. Google said this is more powerful than the Xbox One and PlayStation 4 (PS4) combined.

Google's Phil Harrison, formerly of Xbox, took to stage to detail the service.

Gaming and Streaming on Almost Any Platform

"Our vision is to bring those worlds closer together," Harrison said of gamers, streamers and people who make games. But the ambition of the platform goes beyond streaming Assassin's Creed Odyssey, as the Project Stream test did.

Harrison showed a demo where a user watched a YouTube trailer for Assassin's Creed Odyssey and clicked a play button in the trailer to get the game running immediately in the browser. All you have to do is click on the link to instantly start a game. He also took a dig at games installing slowly on hard drives.

"A single creative vision and single code base can be enjoyed instantly on any screen," he said, pointing to laptops, desktops, TVs and phones. There is no additional box required.

The company then showed a seamless transition from a Chromebook to a phone to a desktop and then a tablet and a TV over a Chromecast, all at 1080p and 60 fps.

More than 100 development kits have been sent to studios around the world. Additionally, Google is forming its own first-party game studio, Stadia Games and Entertainment, headed up by Jade Raymond, formerly of EA and Ubisoft. That arm of the company will also partner with third-party studios on games and tech, Raymond said on stage.

Developer Tools

Google also announced an initiative to bring split screen to Stadia, each screen powered by separate Stadia instances. This could also be used to pull up the views of your teammates in squad-based games.

Luz Sancho, chairwoman of Tequila Works, came on stage to discuss machine learning for developers. For example, Style Transfer can apply patterns to a game based on images fed to the machine learning tool. Additionally, Q-Games' Dylan Cuthbert was on stage to show off State Share. It shows states like time, health or other conditions that let you set up moments to share and challenge others.

YouTube Streaming

YouTube's head of gaming, Ryan Wyatt, detailed how streamers can use the service. Taking on Twitch, he said "YouTube is where people come to watch gaming."

Streamers will be able to capture directly from Stadia. Additionally, Stadia's new Crowd Play button will let YouTube creators generate lobbies for games to play with YouTubers.

Also with YouTube, players will be able to ask the Google Assistant (with a button on the Stadia Controller) for help, which will bring up YouTube clips of how to beat certain puzzles.

The Stadia Controller

You can use your current fpUSB controller, but Google will have its own: The Stadia Controller. It looks like a mix between a PS4 and Xbox One controller. But it connects through Wi-Fi to the game in the Google data center to ensure the best performance. It also has a share button to stream via YouTube and a Google Assistant Button to use the controller's microphones to get help with features built by developers. 

Id Software took to the stage to show off Doom Eternal on Stadia. The game will run at 4K with HDR at 60 fps. Harrison showed how multiple GPUs are more flexible and powerful using the 3DMark Fire Strike benchmark.

Additionally, Harrison said Stadia will embrace platform play, including game saves.

Google CEO Sundar Pichai started the keynote, discussing how gaming has been an important part of Google's research, including using AI, like Google's Deepmind Go, to play games, and using game-like environments in now-sister company WayMo, a self-driving technology firm.

But now, Pichai said they're building a game platform for everyone. Pichai said that Chrome was originally designed for web application and that now it is approaching those goals. He recalled last October's Project Stream, which was a test of Google's ability to stream games at high fidelity and low latency. Games will be on Chrome, Chromebooks, Chromecast and even other browsers in time. 

"It was probably the worst kept secret in the industry," Pichai said.

"Beautiful graphics really need high-end consoles or PCs," he said, pointing out that they don't have instant access. He wants to combine game players, people who watch streams and those who make games themselves.

Want to comment on this story? Let us know what you think in the Tom's Hardware Forums.

Image Credits: Google

Google is hitting the stage at Game Developers Conference (GDC) 2019 in San Francisco for a keynote, under the slogan "gather around," where it will detail its plans for the future of gaming. That's very likely to involve streaming.

The keynote will take place at 10 a.m. PT / 1 p.m. ET. Watch it live below:

Late last year, Google ran a test of its Project Stream technology, in which participants could stream Assassin's Creed Odyssey for free through the Chrome browser on laptops or desktops, as long as they had a 25-megabit-per-second internet connection. 

There have been rumors that Google may fully unveil a streaming service, while the idea of hardware -- or at least a controller -- has also been floating around. We'll find out for sure when Google makes its announcements today.

Over the past few months, not one, not two, but five different backdoors joined the list of security flaws in Cisco routers.

Cisco Architecture for Lawful Intercept

Way back in 2004, Cisco wrote an IETF proposal for a “lawful intercept” backdoor for routers, which law enforcement could use to remotely log in to routers. Years later, in 2010, an IBM security researcher showed how this protocol could be abused by malicious attackers to take over Cisco IOS routers, which are typically sold to ISPs and other large enterprises.

Attackers could exploit these backdoors and not leave any audit trail. That’s how the lawful intercept protocol was designed so that ISP employees can’t tell when a law enforcement agent logs to the ISP’s routers (even though law enforcement is supposed to gain this access with a court order or other legal access request).

Furthermore, this protocol could be abused by ISP employees because no one else working for the ISP could then tell when someone gained access to the routers via Cisco’s Architecture for Lawful Intercept.

New “Undocumented Backdoors” Appear

In 2013, revelations made by German paper Der Spiegel showed that the NSA was taking advantage of certain backdoors in Cisco’s routers. Cisco denied accusations that it was working with the NSA to implement these backdoors.

In 2014, a new undocumented backdoor was found in Cisco’s routers for small businesses, which could allow attackers to access user credentials and issue arbitrary commands with escalated privileges.

In 2015, a group of state-sponsored attackers started installing a malicious backdoor in Cisco’s routers by taking advantage of many of the routers that kept the default administrative credentials, instead of changing them to something else.

In 2017, Cisco, with help from a Wikileaks data leak, discovered a vulnerability in its own routers that allowed the CIA to remotely command over 300 of Cisco’s switch models via a hardware vulnerability.

Five New Backdoors In Five Months

This year has brought five undocumented backdoors in Cisco’s routers so far, and it isn't over yet. In March, a hardcoded account with the username “cisco” was revealed. The backdoor would have allowed attackers to access over 8.5 million Cisco routers and switches remotely.

That same month, another hardcoded password was found for Cisco's Prime Collaboration Provisioning (PCP) software, which is used for remote installation of Cisco’s video and voice products.

Later this May, Cisco found another undocumented backdoor account in Cisco’s Digital Network Architecture (DNA) Center, used by enterprises for the provisioning of devices across a network.

In June, yet another backdoor account was found in Cisco’s Cisco’s Wide Area Application Services (WAAS), a software tool for Wide Area Network (WAN) traffic optimization.

The most recent backdoor was found in the Cisco Policy Suite, a software suite for ISPs and large companies that can manage a network’s bandwidth policies. The backdoor gives an attacker root access to the network and there are no mitigations against it, other than patching the software with Cisco’s update.

Whether or not the backdoor accounts were created in error, Cisco will need to put an end to them before this lack of care for security starts to affect its business.

Cisco's revelation of the VPNFilter malware-slash-botnet in May was scary enough. The company said at the time that VPNFilter affected more than 500,000 devices that could be used to gather information, misdirect investigations of cyber attacks, or cut off hundreds of thousands of people's internet access. Now the company has released new research showing that VPNFilter is more widespread--and dangerous--than expected.

The first discovery showed that VPNFilter wasn't limited to select products from Linksys, Netgear, TP-Link, and MikroTik. Instead, Cisco said in a blog post that devices from Asus, D-Link, Huawei, Ubiquiti, UPVEL, and ZTE have also fallen victim to VPNFilter. More products from the original batch of manufacturers were also found to be affected by the malware. These discoveries mean VPNFilter has more potential victims than originally thought.

Cisco also discovered that VPNFilter has additional capabilities beyond the ones it outlined in the announcement of the malware. The company said:

"We have also discovered a new stage 3 module that injects malicious content into web traffic as it passes through a network device. At the time of our initial posting, we did not have all of the information regarding the suspected stage 3 modules. The new module allows the actor to deliver exploits to endpoints via a man-in-the-middle capability (e.g. they can intercept network traffic and inject malicious code into it without the user's knowledge). With this new finding, we can confirm that the threat goes beyond what the actor could do on the network device itself, and extends the threat into the networks that a compromised network device supports."

Another module is designed to allow other modules that didn't previously include a "kill" command to completely disable the device they've infected. Cisco said this module also removes all traces of VPNFilter before rendering the compromised device unusable, which means someone could activate the malware and nobody would be any the wiser. They'd probably just assume the device in question "died" of natural causes.

The U.S. Department of Justice announced shortly after VPNFilter was revealed that the Department of Homeland Security and FBI had crippled the malware by seizing a domain used by its command-and-control infrastructure. Yet the new discoveries from Cisco drive home the fact that VPNFilter was wounded, not defeated, and that it poses a greater risk to an unknown number of devices from who-knows-how-many manufacturers.

The U.S. Department of Justice announced that it has disrupted the VPNFilter botnet revealed by Cisco's Talos Intelligence Group. After obtaining permission from Pennsylvania courts, the FBI seized a domain used by the botnet's command-and-control infrastructure, effectively crippling its ability to act on infected devices. Yet the Justice Department was careful to note that VPNFilter has been crippled, not slain outright.

Cisco said when it revealed VPNFilter that more than 500,000 devices in 54 countries--with a particular focus on Ukraine--had been compromised by the botnet. The malware targeted small and home office (SOHO) products from Linksys, Netgear, TP-Link, and MikroTik, as well as unidentified NAS devices. These products make particularly good targets because they're rarely protected by antivirus solutions and other security tools.

Targeting SOHO products and NAS devices also gives VPNFilter's operators plenty of options. Cisco said the malware could be used to collect information that passes through the infected devices, to conduct attacks that appeared to be conducted by their victims, and even to render the devices completely inoperable, which could in turn potentially disrupt the internet access of hundreds of thousands of people.

Cisco also said that it had noticed two spikes in VPNFilter activity in May (one on May 8, one on May 17) which is why the company decided to reveal the botnet's existence before finishing its research. It's no wonder, then, why the Justice Department announced that it was taking action to disrupt VPNFilter the same day it was revealed. VPNFilter isn't a potential problem; it's considered a real threat to national security.

Here's what the Justice Department said about its efforts to disrupt VPNFilter:

In order to identify infected devices and facilitate their remediation, the U.S. Attorney’s Office for the Western District of Pennsylvania applied for and obtained court orders, authorizing the FBI to seize a domain that is part of the malware’s command-and-control infrastructure. This will redirect attempts by stage one of the malware to reinfect the device to an FBI-controlled server, which will capture the Internet Protocol (IP) address of infected devices, pursuant to legal process. A non-profit partner organization, The Shadowserver Foundation, will disseminate the IP addresses to those who can assist with remediating the VPNFilter botnet, including foreign CERTs and internet service providers (ISPs).

The Justice Department also advised anyone who owns SOHO or NAS products that may have been infected by VPNFilter to restart their devices. That should temporarily remove the second stage of the malware from the device, and even though the first stage will linger and attempt to reinstall the second stage, the FBI's seizure of the domain used by VPNFilter's command-and-control infrastructure should block those efforts.

VPNFilter Appears To Be The Work Of Russian Hackers

Cisco didn't attribute VPNFilter to any particular organization, but it did talk about what nation-state actors could do with the botnet, prompting fears that it was controlled by a government organization. The Justice Department went a step further and outright attributed VPNFilter to the Sofacy Group, which also goes by APT28, Fancy Bear, Pawn Storm, and other aliases and has been active since at least 2007.

The FBI and Department of Homeland Security said in December 2016 that the Sofacy Group was connected to Russian intelligence services and government officials. Combine that with VPNFilter's apparent focus on Ukraine, where Russia currently holds a military presence, and the connection between this botnet and an organization with at least some loose ties to the Russian government becomes even clearer.

That doesn't mean other countries have nothing to fear from the botnet, however. Cisco said it had already found infected devices in 53 countries outside Ukraine, and the apparent targets (Linksys, Netgear, etc.) ship products around the world. Hence the Justice Department's swift response.

Cisco's Talos Intelligence Group revealed that new malware, which it dubbed VPNFilter, has infected at least 500,000 devices in 54 countries. The malware is said to target Linksys, Netgear, TP-Link, and MikroTik small and home office (SOHO) products as well as unidentified NAS devices. Activating the malware could render affected devices inoperable, which could, in turn, cut off hundreds of thousands of people's internet access.

VPNFilter is said to have steadily infected more and more devices since at least 2016. Cisco said the malware doesn't rely on any specific exploit--instead, it spreads by taking advantage of known vulnerabilities in each individual product. That's made possible at least partly because people neglect to update these devices' firmware, and because they're rarely covered by antivirus solutions and other consumer security tools.

Cisco said VPNFilter could be used for three major purposes: conducting attacks that are mistakenly attributed to the malware's victims; collecting information from devices connected to the affected products; and cutting off victims' access to the internet via the built-in "kill" command. None of these possibilities are particularly welcoming, but the last one, in particular, could be devastating if it's used on many devices.

Unfortunately, knowing about VPNFilter doesn't make it all that much easier to defend against it. Cisco explained in its blog post:

Defending against this threat is extremely difficult due to the nature of the affected devices. The majority of them are connected directly to the internet, with no security devices or services between them and the potential attackers. This challenge is augmented by the fact that most of the affected devices have publicly known vulnerabilities which are not convenient for the average user to patch. Additionally, most have no built-in anti-malware capabilities. These three facts together make this threat extremely hard to counter, resulting in extremely limited opportunities to interdict malware, remove vulnerabilities, or block threats.

It's important to note that Cisco published this report before it finished its research into VPNFilter. That's because the company detected a spike in the rate with which the malware was infecting new devices on May 8, with "almost all" of the newly infected devices being located in Ukraine. Another spike occurred on May 17. Cisco decided to reveal VPNFilter's existence before finishing its research because of these spikes.

A Big Problem Borne Of Many Small Ones

Remember that VPNFilter doesn't rely on new vulnerabilities in networking or NAS products. Instead, the malware spread by taking advantage of a bunch of known flaws that simply haven't been fixed, either because the product makers didn't fix them, or device owners didn't install them. The reason why doesn't matter--what matters is that VPNFilter provides another example of how small vulnerabilities can grow in importance.

This is why experts keep advising companies to stay on top of their products' security, telling consumers to stay up-to-date with security patches, and pleading with regulators to force action on these issues. VPNFilter poses a very real threat to hundreds of thousands of people, many of them in the already embattled Ukraine, and there isn't anything just one company will be able to do to address this threat. It takes a village.

Cisco said in its blog post:

While the threat to IoT devices is nothing new, the fact that these devices are being used by advanced nation-state actors to conduct cyber operations, which could potentially result in the destruction of the device, has greatly increased the urgency of dealing with this issue. We call on the entire security community to join us in aggressively countering this threat.

Apple, Cisco, Aon, and Allianz announced a first-of-its-kind cyber insurance plan that could lower cyber insurance deductibles for companies that use the security-focused Apple and Cisco products. This deal could ultimately allow more companies to pay for cyber insurance as well as improve their security in order to lower the deductibles for that insurance.

Apple, Cisco To Lower Cyber Insurance Costs

Apple tends to offer some of the most secure devices on the market, even though its track record hasn't been flawless. It continuously implements modern security solutions ahead of other industry players. These solutions include default storage encryption, Secure Enclaves, advanced biometric authentication, enforcing TLS encryption for most iOS App Store applications, and so on.

Cisco is also one of the better security solution providers in the industry, offering solutions such as advanced email security, endpoint protection, internet site blocking, and other solutions that can help companies protect themselves against ransomware and malware.

Companies that have already been breached that sign-up for this insurance plan can benefit from Cisco and Aon’s incident response teams, too.

Allianz, the insurance company in the group, can take lower deductibles because it trusts the other three companies in the partnership to have implemented strong security for their customers, thus reducing the number or impact of potential data breaches. 

An End To The Stream Of Data Breaches?

Over the past few years, we’ve seen massive data breach after massive data breach. The data of hundreds of millions of people, both customers and employees, was exposed due to the careless security practices of many of the companies being hacked. Most of the impacted companies didn’t seem to take security too seriously, and as such didn’t sufficiently invest in protecting themselves against these attacks.

As Maersk’s chair recently said, it’s not enough for companies to strive to be only average in terms of security. Otherwise, they’ll eventually suffer the consequences, and it could cost them significant amounts of money. Maersk ended up paying up to $300 million to overhaul the company’s infrastructure after the devastating NotPetya attack.

Cyber insurance can help companies recover money lost though damages in case of a data breach. Insurers will want to see strong security being implemented before they sign the companies up for their plans. Therefore, cyber insurance plans should create a positive feedback loop where more companies want to hedge against data breaches.

At the same time, the insurers will promote and enforce stronger security for their customers in order to reduce their own costs. The lower they can make the data breach risk for their customers, the lower they can make the deductibles. Therefore, insurers, security companies, their enterprise customers, as well as the end-users should all benefit from such plans in the long term, as security is improved across the board in the industry.

A new report from Cisco showed that two-thirds of companies are currently losing sales because of their customers’ growing privacy concerns, which prolong the sales process. Similarly, bad cybersecurity practices are costing companies millions to hundreds of millions of dollars in repairing data breach damage, as well as lost revenue.

2018 - The Year Of Strong Data Protection?

Privacy is quickly becoming a bigger factor in companies’ sales, according to Cisco’s 2018 Privacy Benchmark Maturity (PDF). Privacy policies are longer just legalese meant to keep companies out of trouble, because if done wrong, many businesses risk losing sales from potential customers.

Over 65% of the companies questioned by Cisco have already admitted that data privacy concerns are delaying their sales process. Over 90% of the companies reported delays up to 20 weeks, while the average delay was 7.8 weeks.

A significant number reported delays between 50 to 100 weeks. This means the sales process can take at least 1-2 years longer simply because their customers don’t feel satisfied with the companies’ existing focus on privacy.

The delays don’t just mean that the customers will buy the products later than they would otherwise, because some of them end up either not buying the product at all or buying it from a competitor. Therefore, bad privacy policies can lead not just to lost sales but also to a loss of market share to competitors’ benefit.

The study shows that the longest sales delays happened in Latin America (an average delay of 15.4 weeks), Mexico (13 weeks), and Japan (12.1 weeks). The shortest delays were reported in China (2.8 weeks) and Russia (3.3 weeks).

In terms of industry, government and healthcare sales saw the biggest delays due to cybersecurity and privacy concerns.

Privacy-Immature Companies Most At Risk

The report also found that companies that didn’t take privacy too seriously were the most impacted by these delays. Cisco benchmarked the privacy-maturity of companies based on standards defined by the  American Institute of Certified Public Accountants (AICPA) and the Canadian Institute of Chartered Accountants (CICA). These standards are defined as follows:

Ad hoc — Privacy procedures or processes are generally informal, incomplete, and inconsistently applied. Repeatable — Privacy procedures or processes exist; however, they are not fully documented and do not cover all relevant aspects. Defined — Privacy procedures and processes are fully documented and implemented, and cover all relevant aspects. Managed — Reviews are conducted to assess the effectiveness of the privacy controls in place. Optimized — Regular review and feedback are used to ensure continuous improvement towards optimization of privacy processes

Companies that had a “defined” privacy procedures, saw a 70% improvement in sale processes compared to the companies that had “ad hoc” or informal and incomplete privacy procedures.

Privacy-Mature Companies Are More Secure

Companies that are privacy-mature are not only seeing much shorter sales processes, but are also more protected against data breaches. Only 39% of the privacy-mature companies saw losses of over $500,000 compared to the 74% of the privacy-immature companies. According to Cisco, the lower damages that privacy-mature companies see may also have something to do with them gathering less data than the immature companies.

As hackers become more sophisticated in breaking into large organizations, it may be a good idea for companies to treat customer data as more of a liability than an asset, at least data that isn’t required for the functioning of the product or service. Then, if a data breach does happen, at least the damage will be minimized and the companies won’t have to suffer as large of a hit to their public image.

Maersk Chair: Companies Need To Stop Being Naive About Cybersecurity

Recently, the giant shipping company Maersk suffered a devastating cyber attack through the NotPetya malware. After the attack, the company had to essentially replace its whole infrastructure and reinstall 45,000 PCs and 4,000 servers. This task, which the chair Jim Hagemann Snabe said would have normally taken six months, was performed in a record 10 days. However, even though getting rid of NotPetya from its large infrastructure took a relatively small amount of time, this single attack still ended-up costing the company $250-$300 million, part of which was due to losing 20% of the sales in that period.

At the World Economic Forum in Davos, Snabe said that this incident should be a significant “wake-up call” for every company out there, because it could be them next. He also talked about three important lessons that the company learned during this whole incident:

1. Cybersecurity needs to become their competitive advantage, and being mediocre like everyone else is no longer enough. This is a lesson he argued more companies should learn sooner rather than later.
2. Companies need to stop being naive about cybersecurity. Many companies will experience their own similar data breaches in the future if they don’t treat cybersecurity in a more proactive, rather than reactive way.
3. There is a need for a radical new and more secure infrastructure for the internet, as everything we do becomes more digitized, and thus more at risk of suffering cyber attacks.

This year may be the year when do get this wake-up call that cyber security and data privacy and protection is important not just for the customers who give their data away to the companies, but also for the companies themselves if they don’t want to lose sales of suffer major disruptions because of poor data protection procedures.

Following the recent CCleaner malware incident, Avast (the new owner of CCleaner) and Cisco’s Talos Intelligence security research group have continued to analyze the attack. The two found that the malware was more sophisticated than originally thought and was targeting large companies to steal their intellectual property.

Advanced Persistent Threat

Avast concluded that the malware was in fact a type of attack called an advanced persistent threat (APT), which is a sophisticated attack usually launched by nation states. The APT that infected CCleaner was supposed to deliver a second-stage payload only to select victims.

Of the 2.27 million users who installed the infected CCleaner version, Avast believes that only a few hundred were also infected by the second-stage payload. Avast had previously said that it believes the second stage payload was never deployed, but the company has now walked back that statement.

Large Companies Targeted

The antivirus maker stated that the APT targeted large technology and telecommunications companies from Japan, Taiwan, UK, Germany, and the U.S. via a watering hole attack, which is an attack that targets popular websites or tools used by millions, only to infect a few targets that may also use those same tools. The name is taken from the real world where some predators wait for an opportunity to pick their prey from the animals that come to drink water at a watering hole.

Avast didn’t want to say who were the targets of this attack, but Cisco revealed a list of targets the attackers were attempting to hack. Besides Cisco itself, the list includes Intel, HTC, Samsung, Sony, VMware, Microsoft, Vodafone, Epson, Linksys, MSI, Akamai, and a few others.

Complex Obfuscated Code

According to Avast, the second-stage payload contains complex and obfuscated code and includes two DLL components. The first component comes with anti-debugging and anti-emulation mechanisms, and its purpose seems to be finding another command and control (C2) server.

The C2 server’s address could be modified in the future, which means that it may not be enough that law enforcement shut down the original C2 servers. The attackers may be able to regain control of the infected machines and continue to control them remotely through a new another server.

The second part of the second-stage payload is responsible for persistence on the operating system, and they seem to be piggybacking on other vendors’ applications to avoid detection and maintain persistence. The 32-bit version of the code embeds itself into a Winzip package, whereas the 64-bit one uses a Symantec dll. Most of the malicious code is delivered from the registry. Avast noted that all of these techniques demonstrate a high level of sophistication from the attacker.

Attribution

Attribution is difficult for cyber attacks, as sophisticated attackers can often make it look as if someone else did it by re-using other attackers’ code or hacking styles. They may effectively hide behind reused IP addresses by launching their attacks from computers they hack in a given country. This is also why Avast is reluctant to say for sure who the attackers were right now, but it promised to continue to work with law enforcement to find out who was responsible.

However, Cisco and Kaspersky were both able to confirm that the malware uses code that overlaps with malware code used by “Group 72,” also called “Deep Panda,” “Axiom,” and “Shell Crew.” Group 72 is believed to be a cyber espionage group funded by the Chinese government, and it’s also believed to be responsible for stealing 80 million U.S. social security numbers from health insurance company Anthem.

Avast initially suggested that it should be enough to update to the clean version of CCleaner, but Cisco recommended that it would be safer to restore from backups and reimage the systems. Avast also recommends updating to CCleaner 5.35, as the company has now also revoked the Symantec certificate it was using to sign the infected 5.33 version as well as the cleaned-up 5.34 version.

The VR Health Institute developed a rating system called the VR Exercise Rating that compares the calorie burn of VR games to common exercise activities such as walking, running, and swimming. It should come as no surprise that room-scale VR experiences can burn calories, but you may be surprised by how much exercise can you really get from playing a VR game.

The VR Health Institute partnered with the San Francisco State University’s Department of Kinesiology to measure the oxygen consumption of 40 test subjects while they played a variety of games to determine the how many calories the average person would burn in one minute of gameplay. The results were surprising.

“Many of the games we’ve assessed have turned out to be the equivalent of very vigorous exercise,” said Dr. Marialice Kern, lead VR researcher and the Chair of the Kinesiology department at San Francisco State University. “At the same time, we’ve observed that participants often don’t realize how much effort they’re actually exerting. They’re expending energy with less awareness of the pain of doing it.”

Dr. Kern’s observations are interesting, and they suggest that we could see a future where gamers are more fit than they’ve ever been before. Playing video games is traditionally a sedentary activity, and many people spend several hours every day playing their favorites games. In 2014, Nielson published a report that suggested that the average gamer spends 5.6 hours per week playing games. As VR gains more popularity, more and more gamers will, at least occasionally, get up and start moving around.

Dr. Kern’s observations also suggest that VR could have a positive impact on fitness engagement in general. Physical activity is paramount for a healthy lifestyle, but going to the gym can be a chore for many people. If you can get the same amount of exercise while playing a VR game at home that you can from using the equipment at your local gym, why would you go to the gym?

The VR Health Institute’s VR Exercise Rating features eight levels that represent different workout intensities that you can compare to, including Resting, Walking, Elliptical, Tennis, Rowing, Biking, Swimming, and Sprinting. Games with the Resting rating don’t offer physical activity and would burn less than three calories per minute. Games with the Tennis rating offer a brisk workout, with 6-8 calories burned per minute. And you can expect a significant workout with more than 15 calories burned per minute from games with the Sprinting rating.

“I believe that virtual reality is going to change what it means to go to the gym,” said Aaron Stanton, director and founder of the VR Health Institute. “Our goal is to help quantify which games may be the most useful as part of a healthy lifestyle. For those of us that simply suck at traditional exercise, like me, the diversity and engagement of VR is a game changer. I now believe that I will live longer and be fitter because I’m a gamer, not in spite it.”

The VR Health Institute and San Francisco State University have tested 25 games so far, including Space Pirate Trainer (Elliptical), Tilt Brush (Walking), Audioshield (Biking), and Thrill of the Fight (Sprinting). If you're a developer and you wish to have a rating for your game, you can submit your game for testing. The VR Institute said that it’s not necessary for developers to submit games, though, as it's actively looking for titles to include. Submitting a game may accelerate the process, though.

You can read more about the VR Institute and the VR Exercise Rating at the VR Health Institute's website.

Cisco released its 2017 mid-year cybersecurity report showing some important trends in corporate cybersecurity, including a decline in use of exploit kits, a rise in IoT botnets and destruction of services campaigns, and the return of email spam as an effective attack vector.

Exploit Kits Are Fading Into The Background

According to Cisco’s cybersecurity report, three of the main exploit kits--Angler, Nuclear, and Neutrino--abruptly vanished from the landscape last year. Neutrino returned eventually, but in the form of a subscription service for attackers, rather than as an exploit kit that’s available as a one-time purchase.

What this “service” means for the landscape is that the attacks are more controlled, so it’s harder to detect, while at the same time giving smaller crime groups a chance to use it, too, for a lower subscription cost. However, Cisco said that the overall exploit kit activity has been declining dramatically since January 2016.

According to the company, reasons for the decline include the arrest of the Angler exploit kit author, as well a faster update cycle for previously more vulnerable software platforms such as Flash, web browsers, and Windows, which gained automatic updates with the launch of Windows 10.

Network security research firm Qualys found that it took companies, on average, 308 days to patch 80% of Flash vulnerabilities in 2014, whereas it took them only 62 days in 2016. Exploit kit developers now find that they have to chain together multiple exploits in order for their attacks to be successful.

Cisco’s own research also found that the time it takes to detect a data breach inside a company’s network has been reduced from 39 hours in 2015 to about 3.5 hours for the period from November 2016 to May 2017.

However, Cisco warned that companies should remain vigilant. Although the exploit kit landscape may remain dormant for now, the tools have already been created, and they could be re-activated with the appearance of new major vulnerabilities in popular platforms. If companies allow too much time to pass between the discovery of a major vulnerability and fixing it, then the exploit kit market could once again become lucrative.

Email Spam Is Back

As exploit kit activity has decreased, email spam and malware seem to have made a comeback. Malicious hackers use email as a delivery method for their ransomware and other types of malware. The Cisco researchers anticipate that the volume of email spam will continue to rise as the exploit kit landscape remains in flux. Through email, attackers can also gain entrance to more privileged areas of an organization’s network, from which they can do more damage.

According to Cisco, business email compromise (BEC) is the #1 threat in organizations, even though this type of attack is not as high-profile as ransomware attacks tend to be these days. A BEC campaign involves having an attacker request funds from a financial employee using spoofing to appear as if the request is a legitimate one.

Attackers have siphoned $5.3 billion from companies between 2013 and 2016 using this tactic, compared to only $1 billion gained through ransomware in 2016, according to the Internet Crime Complaint Center.

Ransom Denial Of Service (RdOS)

Some cybercriminal groups, such as the Armada Collective, leverage the threat of a DDoS attack to demand a ransom from companies. A “demo” attack usually follows the request to show that they mean business.

According to Cisco, nearly half (49%) of the researched companies received a ransom note in 2016, a number that seems striking. The Armada Collective typically asks for 20 Bitcoins as ransom, although other groups may ask for less or more.

Destruction Of Services

Destruction Of Services (DeOS), a term coined by Cisco, seems to be on the rise, as well. Over the past year, Cisco has observed attackers building IoT botnet capabilities that can later be used to disrupt services. We’ve also seen with the recent NotPetya attack that some actors, especially some nation-states, may not be interested in obtaining ransom money, but in destroying or disabling certain infrastructure.

IoT Botnets

As proven last year by the DynDNS attack, as well as other major DDoS attacks, we now seem to be in the age of terabit-per-second DDoS attacks enabled by hundreds of thousands or millions of infected Internet of Things (IoT) devices.

The open source Mirai botnet, as well as the severe lack of security or a proper automatic update mechanism for the vast majority of IoT devices, has led to attack campaigns that can be setup and launched with over 100,000 infected devices within a day.

Cisco recommended the following actions for companies looking to defend themselves against attacks targeting their IoT networks:

Keep older signatures activeSurround IoT devices with IPS defensesClosely monitor network traffic (this is especially important to do in IIoT environments, where network traffic patterns are very predictable)Track how IoT devices are touching the network and interacting with other devices (for example, if an IoT device is scanning another device, that is likely a red flag signaling malicious activity) Implement patches in a timely manner Work with vendors that have a product security baseline and issue security advisories

Cybersecurity Has Never Been More Important

Cisco warned that attackers are building tools that can completely disrupt or destroy the operations of a company, because they know the vast majority of firms don’t have a contingency plan for rebuilding their IT infrastructure from scratch in case of a devastating cyber attack. The attackers plan to use this to their advantage, either to obtain larger sums of money as ransom or to further some other objective.

Cisco said that this means companies need to start taking cybersecurity much more seriously and to invest in tools and infrastructure that can keep their networks and data secure. Cisco also argued that security vendors need to work together to develop more interoperable tools that can together meet the challenge of more potent attacks.

"Complexity continues to hinder many organizations' security efforts," said David Ulevitch, Senior Vice President and General Manager, Security Business Group, Cisco, in an email to Tom's Hardware."It's obvious that the years of investing in point products that can't integrate is creating huge opportunities for attackers who can easily identify overlooked vulnerabilities or gaps in security efforts. To effectively reduce Time to Detection and limit the impact of an attack, the industry must move to a more integrated, architectural approach that increases visibility and manageability, empowering security teams to close gaps," he recommended.

Western Digital announced that it’s taking its fight with Toshiba to the International Court of Arbitration in San Francisco to stop the sale of its jointly operated memory-chip business without its consent.

For those of you not up to speed on the current situation, Western Digital acquired control of half of Toshiba's main semiconductor plant when it bought SanDisk almost a year ago today. Since then, Toshiba suffered significant financial hardship after its U.S. subsidiary, Westinghouse Electric, filed for bankruptcy earlier this year. As a result, Toshiba transferred ownership of the joint venture assets held with Western Digital into the newly formed Toshiba Memory with the intent of selling the unit to the highest bidder.

Western Digital's chief executive, Steve Milligan, said:

Toshiba's attempt to spin out its joint venture interests into an affiliate and then sell that affiliate is explicitly prohibited without SanDisk's consent. Seeking relief through mandatory arbitration was not our first choice in trying to resolve this matter. However, all of our other efforts to achieve a resolution to date have been unsuccessful, and so we believe legal action is now a necessary next step.

Toshiba CEO Satoshi Tsunakawa responded:

The majority stake sale of the semiconductor business poses no conflicts with the joint venture contract with Western Digital. Western Digital has no basis for stopping the procedure. We will communicate to the candidates the legitimacy of our argument to wipe out their concerns.

Given Toshiba’s current financial state, sale of the newly formed memory unit is extremely important to the company's financial recovery. In a statement, Toshiba's CEO stated that Western Digital’s complaint was groundless and that his company would push on with the sale, with a second round of bidding to commence on Friday. Just last week, Toshiba sent two letters to Western Digital asserting that the company had the right to sell its part of the semiconductor joint venture and that Toshiba would block Western Digital employees from its facilities and networks unless it complies by May 15.

A three-person panel in San Francisco will hear the arbitration under the rules of the International Chamber of Commerce. Toshiba has 30 days to respond to the request for arbitration.  

HTC announced the second batch of companies joining its Vive X accelerator program. The companies hail from San Francisco, Shenzhen, Beijing, and Taipei and focus on different aspects of hardware and software development for VR and AR. HTC also said that the Vive X program will expand to Israel.

To say that HTC is raring to fund the next big thing in VR and AR would be an understatement. Vive X debuted in August 2016 with $100 million to invest in companies experimenting with VR and AR. Almost three dozen companies signed on between August 2016 and October 2016, when HTC announced that it was looking to support even more companies, and just a few months later the accelerator program has already signed on 60 startups.

Here's how the company described its goal in today's announcement:

We work with the most promising VR/AR companies to advance innovation and move the whole industry forward. We’re continuing to invest in and support the development of foundational platform services and hardware advancements, as well as expanding areas like enterprise, commerce, education, health, social, and eSports.

The companies joining Vive X run the gamut. Construct Studio is creating a VR experience called The Price of Freedom based on the CIA's mind-control-focused Project MK Ultra. Realiteer is working on cognitive behavior therapy programs to help treat depression, anxiety, substance abuse, and other mental health problems. And Limitless is developing the Limitless VR Creative Environment to let you create VR experiences while you're in VR.

Those are just some of the companies joining Vive X from San Francisco. Aurora AR in Shenzhen is working on glass optics and device design for AR, PlusOne in Beijing is developing artificial intelligence for AR that will help companies train employees to interact with customers, and TEGway is making flexible thermoelectric devices that are supposed to let you feel temperatures and pain within VR and AR experiences. (Don't sign us up for that one.)

You can find the full list of startups joining Vive X in this batch in HTC's announcement. Many others are focused on VR gaming, helping enterprise companies adopt XR, and the like. We expect to learn more about them as they go through the accelerator program. And, as if that weren't enough, we also expect to see more announcements from HTC regarding the Vive VR HMD and all the peripherals and content the company is making for it.

Remember the story about three little pigs, their questionable choice in building materials, and the big bad wolf that blows down two of their houses? Well, the Internet Engineering Task Force (IETF) is the one that makes sure the internet isn't made from the technical equivalent of straw and wood. As of next week, the IETF will have a new leader: Alissa Cooper, a Fellow at Cisco who's worked with the IETF in various capacities since 2008.

Cooper got her start at the IETF while working at the Center for Democracy & Technology (CDT) policy organization in Washington, D.C. At CDT, she focused on privacy, free speech, and how technical standards could help support those ideals. Those interests continued at the IETF and Cisco, where she helped the company protect its customers' privacy as it transitioned from on-premise to cloud-based services, and she'll keep pursuing them as the IETF chair.

"What’s interesting about the IETF is that the essence of what we do is build the building blocks that then get incorporated into network designs, corporate products, and larger software systems that get used on the internet," she said in an interview with Tom's Hardware. "So some of the design choices that get made when you’re designing those protocols can have a fundamental impact on privacy.” 

She told us that one of her favorite examples is the possibility of designing a standard with a stable identification that follows you around the internet. Anybody who uses that standard, Cooper said, would be effectively building a permanent identifier into their products. This would make it easier for companies--they'd always know if you visited their site before, what you've done on other sites, and the like--but would also be a privacy disaster.

"For me, it was always an interesting challenge to think through what those considerations are in IETF standards," she said, "Because while they are these fundamental building blocks, you don’t always know how they’re going to be used.” That's not to say that IETF has any say in what people build on top of these standards; many sites use cookies and other identifiers to create something pretty close to the system Cooper described in her example above.

The difference is that someone has to build those things on top of standards that, at least in theory, were designed to protect internet users. It's kind of like IETF making sure a house's foundation is made of brick; it's not the organization's fault if someone decides to build the rest of the house from straw.

Cooper told us that IETF will continue to focus on privacy and security under her stewardship. The organization will also have a particular focus on the Internet of Things (IoT) and how that can be secured. You might be familiar with some of the IoT's security failings--we've covered massive attacks on popular websites, vulnerabilities in connected devices, and some of the industry's attempts to secure the IoT before even more damage can be done.

Cooper will also focus on the administrative side of things. "This is not really the sexiest topic," she said, but the IETF's administrative structure was set up 10 years ago. She wants to look at how the organization can get new people involved and how it can work more closely with the open source style of development and with the people in charge of deploying tech based on its standards through every step of its process.

"I think the nature of what goes on at the IETF and how remarkable it is is kind of underappreciated," she said. "If you think about the internet, it is embedded into basically every aspect of modern life at this point. It’s this huge, distributed system that we all rely on so significantly. At every moment of the day now, if you have a smartphone or a reasonable broadband connection. You're using the internet for everything right now."

Still, for all its influence, Cooper said that the organization hasn't really felt the need to toot its own horn. Other companies are the ones deploying tech based on its standards, introducing new services, and the like, so the IETF has largely focused more on developing new standards than drumming up PR for the ones other people are deploying. That won't change too much under Cooper, but she said she will try to make people appreciate the IETF more.

“I personally think there’s more we could do to emphasize the fact that this thing you use or that people rely on wouldn’t exist if this community of volunteers hadn’t come together to work on it,” she said. You like having a house that some big bad wolf can't just stroll up to and blow over, right? Well, it might be time to thank the architects who carefully decided how everything should be built instead of just throwing a bunch of straw around.

Cisco learned of a vulnerability in its software from the CIA documents published by WikiLeaks on March 7. But the security flaw wasn't included in the problems highlighted by WikiLeaks--Cisco's security team discovered the problem themselves while digging through the "Vault 7" document trove.

The company said in a security advisory that the vulnerability could "allow an unauthenticated, remote attacker to cause a reload of an affected device or remotely execute code with elevated privileges." The problem was in the Cisco Cluster Management Protocol (CMP) processing code used by the Cisco IOS and Cisco IOS XE software. Cisco provided a list of 318 products affected by the vulnerability; you can find the full list in the company's advisory.

The vulnerability resulted from two problems:

• The failure to restrict the use of CMP-specific Telnet options to only internal, local communications between cluster members and instead accept and process such options over any Telnet connection to an affected device, and
• The incorrect processing of malformed CMP-specific Telnet options.

Cisco said it plans to address the vulnerability in future software updates and that no workarounds can mitigate the problem in the meantime. But it did advise customers to switch from the Telnet protocol to SSH because "disabling the Telnet protocol as an allowed protocol for incoming connections would eliminate the exploit vector." Anyone who can't do that can still "reduce the attack surface by implementing infrastructure access control lists (iACLs)."

The vulnerability was publicly disclosed on March 17. Cisco said at the time that "the Cisco Product Security Incident Response Team (PSIRT) is not aware of any public announcements or malicious use of the vulnerability that is described in this advisory." That's good news, considering 10 days passed between WikiLeaks' publication of the Vault 7 documents and Cisco's advisory about a critical vulnerability that affects hundreds of products.

Other vulnerabilities have been found in the Vault 7 trove. WikiLeaks revealed that the CIA targets smartphones to work around end-to-end encrypted messaging apps, that the spy agency circumvented major antivirus software, and that the agency has shown interest in remotely hacking cars. (Three antivirus vendors named in the docs--F-Secure, Avira, and AVG--later told us that the problems have been addressed, or they downplayed their impact.)

Now it's clear that more problems are likely to be found in the Vault 7 documents--and that's just within the files WikiLeaks decided to publish. Here's what the organization said about some of the things it decided not to release:

Wikileaks has carefully reviewed the "Year Zero" disclosure and published substantive CIA documentation while avoiding the distribution of 'armed' cyberweapons until a consensus emerges on the technical and political nature of the CIA's program and how such 'weapons' should analyzed, disarmed and published.

WikiLeaks also said that it "has intentionally not written up hundreds of impactful stories to encourage others to find them and so create expertise in the area for subsequent parts in the series" and that there are "very considerably more stories than there are journalists or academics who are in a position to write them." Expect more companies to find vulnerabilities (or hear about them directly from WikiLeaks) well into the foreseeable future.

Nvidia GeForce GTX 1080 Ti Review

Nobody was surprised when Nvidia introduced its GeForce GTX 1080 Ti at this year’s Game Developer Conference. What really got gamers buzzing was the card's $700 price tag.

Based on its specifications, GeForce GTX 1080 Ti should be every bit as fast as Titan X (Pascal), or even a bit quicker. So why shave off so much of the flagship’s premium? We don’t really have a great answer, except that Nvidia must be anticipating AMD’s Radeon RX Vega and laying the groundwork for a battle at the high-end.

Why now? Because GeForce GTX 1080 Ti is ready today, Nvidia tells us. And because Vega is not, we’d snarkily add.

Turning A Zero Into A Hero

There are currently two graphics cards based on Nvidia’s GP102 processor: Titan X and Quadro P6000. The former uses a version of the GPU with two of its Streaming Multiprocessors disabled, while the latter employs a pristine GP102, without any defects at all.

We’re talking about a 12 billion transistor chip, though. Surely yields aren’t so good that they all bin into one of those two categories, right? Enter GeForce GTX 1080 Ti.

The 1080 Ti employs a similar Streaming Multiprocessor configuration as Titan X—28 of its 30 SMs are enabled, yielding 3584 CUDA cores and 224 texture units. Nvidia pushes the processor’s base clock rate up to 1480 MHz and cites a typical GPU Boost frequency of 1582 MHz. In comparison, Titan X runs at 1417 MHz and has a Boost spec of 1531 MHz. 

Where the new GeForce differs is its back-end. Both Titan X and Quadro P6000 utilize all 12 of GP102’s 32-bit memory controllers, ROP clusters, and slices of L2 cache. This leaves no room for the foundry to make a mistake. Rather than tossing the imperfect GPUs, then, Nvidia turns them into 1080 Tis by disabling one memory controller, one ROP partition, and 256KB of L2. The result looks a little wonky on a spec sheet, but it’s perfectly viable nonetheless. As such, we get a card with an aggregate 352-bit memory interface, 88 ROPs, and 2816KB of L2 cache, down from Titan X’s 384-bit path, 96 ROPs, and 3MB L2.

Left alone, that’d put GeForce GTX 1080 Ti at a slight disadvantage. But in the months between 1080’s launch and now, Micron introduced 11 Gb/s (and 12 Gb/s, according to its datasheet) GDDR5X memories. The higher data rate more than compensates for the narrower memory bus: on paper, GeForce GTX 1080 Ti offers a theoretical 484 GB/s to Titan X’s 480 GB/s.

Of course, eliminating one memory channel affects the card’s capacity. Stepping down from 12GB to 11GB isn’t particularly alarming when we’re testing against a 4GB Radeon R9 Fury X that works just fine at 4K, though. Losing capacity is also preferable to repeating the problem Nvidia had with GeForce GTX 970, where it removed an ROP/L2 partition, but kept the memory, causing slower access to the orphaned 512MB segment. In this case, all 11GB of GDDR5X communicates at full speed.

Specifications

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Meet The GeForce GTX 1080 Ti Founders Edition

During its presentation, Nvidia announced that its Founders Edition cooler was improved compared to Titan X's. Looking at this card head-on though, you wouldn't know it. It looks identical except for the model name. The combination of materials (namely cast aluminum and acrylic) is also the same, as is the commanding presence of its 62mm radial cooler.

Nvidia's reference GeForce GTX 1080 Ti is the same size as its Titan X (Pascal). The distance from the slot cover to end of the cooler spans 26.9cm. From the top of the motherboard slot to the top of the cooler, the card stands 10.5cm tall. And with a depth of 3.5cm, it fits nicely in a dual-slot form factor. We weighed the 1080 Ti and found that it's a little heavier than the Titan X at 1039g.

The top of the card looks just as familiar as its front, sporting a green back-lit logo, along with one eight- and one six-pin power connector. The bottom is even less interesting; there’s really nothing to say about its plain cover.

Some hot air may escape the back of the card through its open vent. But the way Nvidia designed its thermal solution ensures most of the waste heat exhausts out the back.

Nvidia improves airflow through the cooler by removing its DVI output. You get three DisplayPort connectors and one HDMI 2.0 port, while a bundled DP-to-DVI dongle covers anyone still using that interface.

Cooler Design

We had to dig deep in our tool box because Nvidia primarily uses thin 0.5mm screws, which fit into the mating threads of special piggyback screws that sit below the backplate. These uncommon M2.5 hex bolts also attach the card’s cover to its circuit board.

One improvement to the GeForce GTX 1080 Ti became apparent as we started taking the card apart: Nvidia mated the PWM controller on the back of its PCA with part of the backplate using thick thermal fleece. This is a material we don't see used very often, and it's meant to augment heat dissipation. The move would have been even more effective if Nvidia cut a hole into the plastic sheet covering the backplate in this area.

The exposed board reveals two areas on the left labeled THERMAL PAD 1 and THERMAL PAD 2. However, these do not actually host any thermal pads. We don’t know if Nvidia’s engineers deemed them unnecessary or if its accountants decided they were too expensive. Our measurements will tell.

The cooler’s massive bottom plate sports thermal pads for the voltage converters and memory modules, as well as several of the thermal fleece strips we mentioned previously. Those strips connect other on-board components to the cooler's bottom plate.

Similar to its other high-end Founders Edition cards, Nvidia uses a vapor chamber for cooling the GPU. It’s attached to the board with four spring bolts.

Board Design & Components

Physically, the first thing you might notice about the PCA is its full complement of voltage regulators. Nvidia’s Titan X (Pascal) had the same layout, but not all of its emplacements were populated. The Quadro P6000, on the other hand, uses this board design. That card's eight-pin power connector points toward the back, and you can see the holes for it on the 1080 Ti's PCB.

The opposite holds true for memory: compared to Titan X (Pascal), one of GeForce GTX 1080 Ti's modules is missing.

A total of 11 Micron MT58K256M321JA-110 GDDR5X are organized around the GP102 processor. They operate at 11 Gb/s data rates, which helps compensate for the missing 32-bit memory controller compared to Titan X. We asked Micron to speculate why Nvidia didn't use the 12 Gb/s MT58K256M321JA-120 modules advertised in its datasheet, and the company mentioned they aren't widely available yet, despite appearing in its catalog.

Nvidia sticks with the uP9511 we've seen several times now, which makes sense because this PWM controller allows for the concurrent operation of seven phases, as opposed to just 6(+2). The same hardware is used for all seven of the GPU's power phases, and they're found on the back of the board.

The voltage converters’ design is interesting in that it’s quite simple: one buck converter, the LM53603, is responsible for the high side, and two (instead of one) Fairchild D424 N-Channel MOSFETs operate on the low side. This setup spreads waste heat over twice the surface area, minimizing hot-spots.

For coils, Nvidia went with encapsulated ferrite chokes (roughly the same quality as Foxconn’s Magic coils). They can be installed by machines and aren’t push-through. Thermally, the back of the board is a good place for them, though we find it interesting that Nvidia doesn't do more to help cool these components. Stranger still, the capacitors right next to them receive cooling consideration.

The memory gets two power phases run in parallel by a single uP1685. The high side uses the FD424 mentioned above, whereas the low side sports two dual N-Channel Logic Level PowerTrench E6930 MOSFETs in a parallel configuration. Because the two phases are simpler, their coils are correspondingly smaller.

So, what’s the verdict on Nvidia's improved thermal solution? Based on what we found under the hood, it'd be safer to call this a cooling reconceptualization. Switching out active components and using additional thermal pads to more efficiently move waste heat are the most readily apparent updates. The cooler itself should perform identically to cards we've seen in the past.

MORE: Nvidia GeForce GTX 1080 Roundup

MORE: Nvidia GeForce GTX 1070 Roundup

How We Tested Nvidia's GeForce GTX 1080 Ti

Nvidia’s latest and greatest will no doubt be found in high-end platforms. Some of these may include Broadwell-E-based systems. However, we’re sticking with our MSI Z170 Gaming M7 motherboard, which was recently upgraded to host a Core i7-7700K CPU. The new processor is complemented by G.Skill’s F4-3000C15Q-16GRR memory kit. Intel’s Skylake architecture remains the company’s most effective per clock cycle, and a stock 4.2 GHz frequency is higher than the models with more cores. Crucial’s MX200 SSD remains, as does the Noctua NH-12S cooler and be quiet! Dark Power Pro 10 850W power supply.

As far as competition goes, the GeForce GTX 1080 Ti is rivaled only by the $1200 Titan X (Pascal). The only other comparisons that make sense are Nvidia’s GeForce GTX 1080, the lower-end 1070, and AMD’s flagship Radeon R9 Fury X. We add a GeForce GTX 980 Ti to the mix for showing what 1080 Ti can do versus its predecessor.

Our benchmark selection now includes Ashes of the Singularity, Battlefield 1, Civilization VI, Doom, Grand Theft Auto V, Hitman, Metro: Last Light, Rise of the Tomb Raider, Tom Clancy’s The Division, Tom Clancy’s Ghost Recon Wildlands, and The Witcher 3. That substantial list drops Battlefield 4 and Project CARS, but adds several others.

The testing methodology we're using comes from PresentMon: Performance In DirectX, OpenGL, And Vulkan. In short, all of these games are evaluated using a combination of OCAT and our own in-house GUI for PresentMon, with logging via AIDA64. If you want to know more about our charts (particularly the unevenness index), we recommend reading that story.

All of the numbers you see in today’s piece are fresh, using updated drivers. For Nvidia, we’re using build 378.78. AMD’s card utilizes Crimson ReLive Edition 17.2.1, which was the latest at test time.

Ashes, Battlefield 1 & Civilization VI

Ashes of the Singularity (DX12)

In the past, we tested Ashes of the Singularity using the game’s Extreme preset. This time around we bumped up to Crazy, given the GeForce GTX 1080 Ti’s forgone position in the performance hierarchy.

It comes as no surprise that the 1080 Ti performs a lot like Nvidia’s Titan X (Pascal). True to Jen-Hsun’s word, the new flagship is even a little faster at 2560x1440.

AMD’s Radeon R9 Fury X notably beats the GeForce GTX 980 Ti and 1070 in our average frame rate measurements. Paging through the charts, though, you’ll notice its frame times spiking more often than the other cards. The implication of this is reflected in our unevenness metric, where almost 80% of the run is perfectly smooth, but that last 5% or so starts breaking down into less consistent frame delivery.

The same finishing order carries over to 4K, where Nvidia’s GeForce GTX 1080 Ti nearly averages 60 FPS and doesn’t drop below 40 FPS.

As before, the Radeon R9 Fury X lands in the middle of our field. While it still demonstrates worrisome frame time spikes, Nvidia’s GeForce GTX 980 Ti seems more prone to stutter according to our unevenness index.

Then again, given the minimum frame rates achieved by the GeForce GTX 1080 and lower, you’d either want to dial quality back in Ashes or upgrade to a faster card for 4K.

Battlefield 1 (DX12)

Our Battlefield 1 benchmark comes from the Avanti Savoia! story, where we run an 80-second sequence at the beginning of O La Vittoria. Naturally, all of our test subjects utilize DirectX 12 and the Ultra quality preset.

The GeForce GTX 1080 Ti carves out an almost-24% advantage over the GeForce GTX 1080 at 2560x1440, pushing it over 100 FPS on average. That’s a 68% speed-up over last generation’s GeForce GTX 980 Ti. AMD’s fastest single-GPU solution is desperately looking for relief from the Radeon RX Vega, expected next quarter.

4K is more interesting, if only because the higher resolution puts more of a strain on these high-end graphics cards.

GeForce GTX 1080 Ti maintains a small average frame rate advantage over the Titan X (Pascal), and its lead over GeForce GTX 1080 grows to almost 29%. That’s significant insofar as the 1080 Ti keeps its nose above 50 FPS through our test sequence, while 1080 drops closer to 40 FPS.

Our unevenness index illustrates how the GeForce GTX 1080 Ti and Titan X deliver mostly-smooth experiences, while the GeForce GTX 980 Ti and Radeon R9 Fury X spend about half of the run in an “average” playability range.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Civilization VI (DX12)

Let’s be real: Civilization VI isn’t great as a graphics benchmark. It’s largely CPU-bound, even on our Core i7-7700K. But we know this is a popular title, so we crank all of the Advanced detail settings up as high as they’ll go and apply 8x MSAA to increase the graphics workload as much as possible.

Nevertheless, at 2560x1440, all of the GeForce cards land within a few FPS of each other. The 1080 Ti even appears at the bottom of this pile. Jumping up to 4K should give the GP102-based boards an opportunity to stretch their legs a little more…

The finishing order straightens out a bit, but the fact that four cards all bunch up around 60 FPS (yes, with v-sync and the game’s internal frame-limiter disabled) tells us more graphics horsepower won’t help. Expect Civ VI to show up more often in our CPU reviews and perhaps mainstream graphics stories. It’s just not a good indicator of high-end GPU performance.

Doom, GTA V & Hitman

Doom (Vulkan)

Use of the same benchmark sequence and quality settings allows us to compare the new results in today’s story and older data from our Nvidia Titan X Pascal 12GB Review. Why does this matter? At a recent workshop for members of the press, Nvidia made it a point to mention performance improvements tied to DirectX 12- and Vulkan-based games.

Nvidia’s GeForce GTX 1080 Ti slips past the Titan X (Pascal), turning in a result that’s 30% faster than GeForce GTX 1080 and more than 76% quicker than the 980 Ti.

More to the point of the preceding paragraph, Titan X is almost 12% faster than when we first reviewed it. In the meantime, we did upgrade our test bed to a Core i7-7700K (from a -6700K), so let’s look at the 4K results, which further deemphasize platform performance.

The Titan X is now 14% faster than it was last August, so it does appear Nvidia exacted positive changes, either through its driver or via cooperation with id. GeForce GTX 1080 Ti similarly benefits from these improvements and is once again faster than Titan X (Pascal).

A more resource-laden GPU allows the 1080 Ti to outperform Nvidia’s vanilla 1080 by 34%. It’s a full 88% faster than last generation’s GeForce GTX 980 Ti. And although a higher resolution helps AMD’s Radeon R9 Fury X beat both the GeForce GTX 980 Ti and 1070, 1080 Ti is still 68% faster than the Fiji-based flagship.

Fortunately, all of the cards are fast enough at 4K to deliver what our unevenness index considers to be acceptable playability.

Grand Theft Auto V (DX11)

Our Grand Theft Auto V benchmark is made more demanding by the use of 4xMSAA (instead of 2x) and 4x Reflection MSAA (rather than Off). The rest of the settings are still maximized to whatever extent is possible.

This time around, Nvidia’s Titan X (Pascal) lands in the lead by a few frames per second. Still, the GeForce GTX 1080 Ti is more than 26% faster than the 1080, and 65% quicker than an older 980 Ti.

As we’ve seen over the past year, DirectX 12 tends to paint AMD’s Graphics Core Next architecture in a favorable light. Radeon cards typically don’t fare as well in DirectX 11-based titles, though. Consequently, GeForce GTX 1080 Ti establishes its largest lead yet (89%) over the Radeon R9 Fury X in GTA V, the first DX11 game we’ve encountered.

The new GeForce’s advantage over AMD’s flagship grows to 95% at 3840x2160, where Nvidia’s top two cards are playable and the Radeon is not.

A higher resolution helps the 1080 Ti hop out ahead of Titan X (Pascal), extending the card’s lead over GeForce GTX 1080 to almost 36%.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Hitman (DX12)

The GeForce GTX 1080 Ti matches the Titan X in our Hitman benchmark, and turns in an average frame rate just 14% faster than 1080.

Our unevenness index shows all six contenders delivering fairly smooth performance, even under the Ultra level of detail setting. So let’s move on to 4K testing for a more taxing workload.

A more graphics-bound task allows the 1080 Ti’s higher clock rate to exert an advantage over Titan X. At the same time, the 1080 Ti’s lead over GeForce GTX 1080 grows to 30%.

All of the cards in today’s line-up serve up playable frame rates, more or less. But the top-end 1080 Ti is more than 81% faster than last generation’s 980 Ti. Particularly at 4K, that’s a big difference.

Metro, Tomb Raider & Ghost Recon

Metro: Last Light (DX11)

Despite its age, Metro: Last Light is a staple in our test suite for its ability to tax modern graphics cards. We use the Very High Quality preset with 16x anisotropic filtering, Normal Motion blur, and Normal Tessellation.

Everything from a Radeon R9 Fury X and up cuts through Metro fairly easily at 2560x1440. The Titan X (Pascal) and GeForce GTX 1080 Ti serve up comparable performance, both beating Nvidia’s GeForce GTX 1080 by about 19%.

The results at 4K are marginally more interesting, since last generation’s flagships can’t quite deliver smooth performance at Metro’s top detail settings, yet the fastest Pascal-based cards do. Not even a GeForce GTX 1070 cuts it.

As we saw in GTA V, another DX11 title, the GeForce GTX 1080 Ti takes a slim loss at 2560x1440 and turns it into a slimmer victory at 3840x2160. The resulting average frame rate is 26% higher than Nvidia’s vanilla 1080, and 60% faster than the aging 980 Ti.

Rise of the Tomb Raider (DX12)

In an effort to make our Rise of the Tomb Raider more demanding than in past reviews, we enabled 2x SSAA (rather than SMAA, a post-processing effect that has been shown to leave many surfaces aliased).

The performance impact of increasing the internal rendering resolution is understandably immense.  Whereas we saw 110+ FPS from the Titan X back in August using SMAA, we’re under 80 FPS at 2560x1440. That’s a good target for ultra-high-end GPUs, though.

Averaging 79.5 FPS, the GTX 1080 Ti is 34% faster than a 1080 and 70% faster than a 980 Ti.

Although this a DX12-enabled title, AMD’s Radeon R9 Fury X struggles mightily. A couple of big frame time spikes translate into problematic sequences our unevenness index flags as unplayable. It doesn’t look good for the bottom half of this chart as we shift to 4K.

Even the GP102-based boards suffer under super-sampling. Our unevenness index suggests we’re still in the playable range on Titan X (Pascal) and GeForce GTX 1080 Ti. However, the other cards clearly stutter. Given the pixel density of a 27” 4K monitor, you’d probably be fine turning anti-aliasing off entirely at this resolution and enjoying the extra performance.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Tom Clancy’s Ghost Recon Wildlands (DX11)

We received our keys for Ghost Recon Wildlands right as we were finishing up benchmarking and decided to add this title for its newness factor. The game’s Ultra preset hits performance unbelievably hard, given graphics we consider good but not groundbreaking. So we dialed back to Very High, which has the side benefit of turning off Nvidia’s Turf Effects feature. Until we can quantify its effect on AMD’s hardware, that only seems fair.

Ghost Recon is a DirectX 11-based title. As such, we’re hardly surprised to see Nvidia’s cards fare so much better than AMD’s in this TWIMTBP game right out of the gate. Perhaps the better comparison is 1080 Ti to Titan (the 1080 Ti is already a little faster), 1080 Ti to 1080 (1080 Ti is 23% quicker), and 1080 Ti to 980 Ti (big Pascal trumps big Maxwell by almost 56%).

The burden of Very High settings weighs heavy on even the fastest single-GPU cards. Interestingly, Titan X takes the lead, while 1080 Ti is just 18% faster than 1080. Our unevenness index isn’t kind to any card—they all fall short of perfect. So perhaps we need to revisit Ghost Recon once both AMD and Nvidia have a chance to do a little optimization work.

The Division & The Witcher 3

Tom Clancy’s The Division (DX12)

While we’re using the same Ultra quality preset to test The Division, DirectX 12 support was added after our Titan X review, so these are our first numbers under the lower-level API.

Compared to the data published in our Nvidia Titan X Pascal 12GB Review, the Titan X is up almost 10% through some combination of our upgrade to a Core i7-7700K, the move to DirectX 12, and updated drivers. In comparison, AMD’s Radeon R9 Fury X is up a little more than 10%.

The GeForce GTX 1080 Ti ties the Titan X in average frame rate, but delivers a higher minimum.

Only the GeForce GTX 980 Ti registers frame time spikes large enough to appear on our unevenness index as less-than-ideal.

Titan X (Pascal)’s gains shrink to ~7% compared to our review last August, while the Radeon’s drop to ~3%. Whereas the Titan X is playable, though, AMD’s Fiji-based card isn’t.

GeForce GTX 1080 Ti lands right underneath the other GP102-based board, besting the 1080 by almost 33%.

The Witcher 3 (DX11)

We re-ran these numbers using the latest drivers (because that’s the right thing to do), but performance looks really similar to what we saw last August. All of the cards fare well, serving up playable frame rates at 2560x1440.

For those of you keeping score, GeForce GTX 1080 Ti is 28% faster than the 1080 at this resolution.

Then, it’s almost 38% faster than the 1080 at 3840x2160. Previously, respectable single-GPU performance at 4K and The Witcher 3’s highest detail settings would have cost you $1200. Now you’re looking at a $700 outlay.

Slower cards can handle this resolution, but you’ll probably want to dial the quality sliders back a notch or two. Or better yet, run at 2560x1440 with the settings max’ed out—The Witcher 3 is one of those games that’ll have you marveling at the developer’s attention to detail.

Power Consumption

Our graphics card test setup and methodology are detailed in How We Test Graphics Cards.

At idle, the GeForce GTX 1080 Ti Founders Edition’s power consumption lands exactly where we'd expect: right around 13W.

One result that really stands out is the >30W difference in our gaming workload between the cold and fully-exercised card. We see power drop a bunch once the 1080 Ti reaches its full operating temperature, which means it's hitting a thermal limit and being slowed down by GPU Boost.

To back our claim, we recorded the temperature and clock rate during warm-up and plotted them in the same graph:

Our stress test yields the same drop in frequency, though it's less pronounced due to the much higher load.

How do these results compare to Nvidia's Titan X (Pascal)? A look at our library of data yields some interesting findings.

The GeForce GTX 1080 Ti's average clock rate is a tad higher than the Titan X's once both cards reach their peak operating temperature. However, the 1080 Ti gets there faster, giving the Titan X a boost early on.

Also interesting is that the Titan X's curve is a lot smoother than the 1080 Ti's, which behaves somewhat frenetically.

The power consumption curves as graphs are provided below, and we start by comparing the cold and warmed-up cards:

Notice the same large difference in power consumption between the cards' temperature levels.

We documented the GeForce GTX 1080 Ti's power consumption and current exclusively at its operating temperature because the two maximum values ended up being almost identical. This is due to the power target of 250W imposing a hard limit, which Nvidia manages to hit very well.

Adherence To Specifications

Ever since the launch of AMD's Radeon RX 480, we look closely at every new card's adherence to the PCI-SIG’s specifications. Nvidia's GeForce 1080 Ti Founders Edition is no exception, and it passes our test with flying colors. It doesn't use the 3.3V rail at all; only the 12V rail.

Our readings put the motherboard slot's 12V rail at approximately 4.4A. Given a ceiling of 5.5A, the card has plenty of room to spare.

Heat & Noise

Temperature Under Load

The advantages of Nvidia's direct heat exhaust design are plain to see. The average temperatures are the same whether you're working with a closed PC case or on an open bench table. In fact, due to an ample supply of fresh air, temperatures in the closed PC case rise more slowly than those on the open bench table. Otherwise, the curves are almost identical, which means that Nvidia's GeForce GTX 1080 Ti Founders Edition is a good choice for closed PC cases and small enclosures.

The card does hit a temperature limit, though. We know this due to the power consumption and clock rate results at different temperatures. Then again, Nvidia's Titan X (Pascal) did the same thing when we reviewed it.

Of course, it goes without saying that we’re using our high-resolution infrared technology to record temperatures around the card, since the GPU's thermal reading isn't the only one that matters. Other components play an important role too, and some of them get even hotter!

The Warm-Up Process

We ran the gaming and stress test loops on an open bench table and in a closed PC case, so we ended up with four datasets corresponding to one video each. Since the original videos would have gotten boring for you to watch in real-time, we sped them up by a factor of 10. The videos display only the GPU temperatures as their main measurement focus.

If the videos are either too fast or slow for you, we're also including the final results as stills for every scenario. In these, we added the voltage converter and memory module temperatures.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

Gaming Loop

At idle or under a light load (browsing the Internet or watching a video), the fan spins leisurely at 1000 RPM or less. This is enough to keep temperatures nice and cool.

The temperature results under load demonstrate once again that it really doesn’t matter where you install the GeForce GTX 1080 Ti Founders Edition, thanks to its cooling solution.

Stress Test Loop

The load might be different, but our infrared pictures appear unchanged. The memory modules do rise 1°C when switching from the open bench table to a closed PC case during both the gaming and stress test loops. However, with Micron’s GDDR5X memory modules rated for a maximum temperature of 95°C, the result isn’t even close to worrisome. Also, there aren't many enthusiasts who run FurMark recreationally.

The Founders Edition card's cooling performance is generally very good, especially when it comes to the voltage converters. However, Nvidia's reference design does reach a noise level that is quite noticeable.

MORE: Nvidia GeForce GTX 1080 Roundup

MORE: Nvidia GeForce GTX 1070 Roundup

Noise

The GeForce GTX 1080 Ti's operating noise after warm-up is obvious, but (just so) bearable. This is certainly not what we'd call a quiet graphics card. Let's have a look at the relationship between fan speed and temperature.

In a closed PC case, the fan has to make a noticeable push to to keep the temperature stable at Nvidia's configured ceiling once the card has been running for a while.

A direct comparison between Nvidia’s GeForce GTX 1080 Ti Founders Edition and Titan X (Pascal) yields almost no difference. The temperatures and fan speeds are practically identical during our gaming loop.

The only physical change to the Founders Edition card involves its rear exhaust, which is more free-flowing. Perhaps as a result, the 1080 Ti's noise profile is a little different compared to Titan X. Subjectively, however, there’s no difference between the two graphics cards, and their noise levels are almost the same as well. At full operating temperature, the 1080 Ti is slightly higher, despite a slower fan speed. Then again, its clock rate is a bit more aggressive as well, and there could always be slight manufacturing-related differences between the boards.

The noise level increases during our stress test. This is due to the temperature target the card must adhere to, no matter what.

Bottom line: there’s really not much difference between Nvidia’s Titan X (Pascal) and its GeForce GTX 1080 Ti Founders Edition. This might be a good or a bad thing, depending on your point of view.

Conclusion

Prior to Jen-Hsun’s pricing announcement on-stage in San Francisco, the Tom’s Hardware team debated where GeForce GTX 1080 Ti might land. $800 sounded cheap. $900 would have even been reasonable, given the card’s theoretical similarities to Titan X (Pascal) at $1200. Nobody suggested that Nvidia might replace GTX 1080 altogether at $700.

Why not? For a couple of reasons.

A $700 GeForce GTX 1080 Ti makes a $1200 Titan X look pretty ridiculous. The Titan’s only advantage is an extra 1GB of GDDR5X, after all. So it’ll either need to go away or drop in price significantly. Because Nvidia only sells the card though its own site, though, controlling Titan X’s fate seems like a simple matter. There’s always the 3840-core GP100 processor, if there ever came a need for a beefier Titan card…

Then there’s the fact that 1080 Ti effectively replaces 1080 at the same price point. If you’re a gamer who didn’t upgrade when the 1080/1070 launched last year, this is great news. But if you bought a 1080 Founders Edition card one month before GDC, then you’re probably reading up on return policies. Shoot, even Oculus knew it’d get backlash after announcing its $600 Rift/Touch package. The company dropped $50 into our Store account, since we had registered our Touch controllers within the 30 days leading up to GDC. But Nvidia says there are no plans to do something similar.

Gamers who were previously on the fence about a new high-end graphics card are looking at an entirely new class of performance at a price point we simply weren’t expecting. In essence, it’s a Titan X at the 1080 Founders Edition’s price. Does Nvidia know something about AMD’s upcoming flagship that nobody else does and is pricing accordingly? Perhaps. Truth be told, secrets don’t last very long in this business. But the beginnings of GeForce GTX 1080 Ti likely started with Nvidia figuring out what it could harvest out of GP102s saddled by a bad ROP cluster or memory controller.

What results is a card capable of smooth performance at 3840x2160, in many games with the quality settings max’ed out. It tears through 1440p too, if you prefer gaming on a fast-refresh display, or want to give G-Sync a spin. We have VR titles in the lab that cause a GTX 1070 to trip up. In time, they’ll start overwhelming 1080s as well, necessitating something like the 1080 Ti to maintain a constant 90 Hz.

Of course, Nvidia is handling this launch differently than Titan X. Board partners will have their own custom designs, to start. The $700 MSRP applies equally to the third-party and Founders Edition models, we’re happy to report. And although some SKUs will undoubtedly command premium pricing, we expect Nvidia’s implementation to anchor the rest, preventing the rampant inflation that plagued Pascal last summer.

When a room full of experienced reviewers hears a price that makes them all look at each other, muttering “wow,” then you know the product’s going to do well. GeForce GTX 1080 Ti delivers frame rates we could have predicted at a price that caught us by surprise.

Equally relevant is the pressure that 1080 Ti puts on the old guard. GeForce GTX 1080 cards are already selling for $500 online, and the Founders Edition board dropped to $550 on Nvidia’s own site. GeForce GTX 1070s aren’t any cheaper yet, but that move could be waiting in the wings once Radeon RX Vega makes itself known.

Oh, right. Vega. We already know so much about what AMD’s next-generation flagship can do. And yet we’re still waiting for some indication of what it can do. It’s hard for us to imagine a scenario where Radeon RX Vega delivers notably better performance per dollar than what we just tested. But we invite AMD to prove us wrong. Really. AMD, please.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content

We've seen a long slow trickle of information about AMD's Ryzen processors spill out over the last several months. AMD has fed us enough information to keep us listening, and combined with a mudslide of mostly-false leaks, it’s fair to say the excitement has reached a fever pitch. We are finally on the cusp of the official March 2 launch date, and as such, AMD held a Tech Day event in San Francisco to lay out the pricing, specifications and its own internally generated benchmarks for its three leading SKUs. The Ryzen 7 1800X ($499), 1700X ($399), and 1700 ($329) all pack 8 cores and 16 threads at an impressive price point. 

AMD CEO Lisa Su also presented three demos pitting Ryzen against competing Intel processors, including a Cinebench multi-threaded test, HandBrake video transcoding test, and 4K gaming session. 

The company also surprisingly announced that Ryzen processors are available for pre-order at 1:30pm ET today (Feb 22) from 180 global e-tailers and boutique OEMs, which is somewhat odd timing considering that product reviews aren't out yet.

AMD began the blank-sheet Zen processor core design phase four years ago and invested two million engineering hours optimizing the architecture and process technology to strike the right blend of power and performance. The end result comes in the form of the Ryzen processors, which come packing 4.8 billion 14nm transistors. AMD finally shared a naked image of the die, and we can clearly spot the two separate CPU complexes, which come with four cores each. We've already covered the Zen microarchitecture in our Everything Zen article, but we'll revisit the topic with new details in our review.

AMD originally set out to increase IPC 40% over its own previous-generation chips, but the company revealed that it actually surpassed that goal and measured a 52% increase. Every processor design is an engineering marvel--for instance, the 8-core, 16-thread Ryzen processors come with 2,000 meters of internal signal wiring, but perhaps AMD's greatest feat comes in the form of Ryzen's low price tag. 

AMD is quite vocal that it is out to disrupt the PC market with its low pricing model, and all three of the leading Ryzen models deliver on that front, especially in light of their beefy 8-core designs.

Ryzen 7 1800X

The Ryzen 7 1800X features 8 cores/16 threads with a 3.6GHz base and 4.0GHz boost frequency. The "X" designation in the product name denotes that the processor features AMD's XFR (eXtended Frequency Range) technology, which allows for higher clock speeds if you employ a more robust cooling device. XFR is just one of several key technologies in AMD's SenseMI suite, which includes Pure Power, Precision Boost, Smart Prefetch and Neural Net Prediction features.

AMD presented its own internal benchmarks comparing the 1800X to the Intel Broadwell-E 8-core/16-thread Intel Core i7-6900K. AMD claimed it offers 9% more performance in the Cinebench R15 multi-threaded test (noted as nT) and matches the i7-6900K's single-core score. More importantly, the company pointed out its $499 price, which is much lower than the i7-6900K's $1,050. It has a 95W TDP.

EDIT: AMD did not send the final test configurations until moments before launch, but it is worth calling out that the company tested the Broadwell-E comparison systems with a dual-channel memory configuration, though they support quad-channel memory. This could penalize Broadwell-E's Cinebench performance slightly.

Ryzen 7 1700X

The 95W Ryzen 7 1700X features a 3.4GHz base and 3.8GHz boost frequency. It also features 8 cores and 16 threads, which it leverages to beat the 6-core Intel Core i7-6800K by 39% and the 8-core Core i7-6900K by 4% in the multi-threaded Cinebench R15 test. Notably, AMD didn't share the 1700X's single-threaded results. The lack of a single-threaded benchmark result, or more expansive benchmarks of any variety, means that we will still need to wait to see reviews for the full story. Once again, the price is a big attraction: The Ryzen 7 1700X retails for $399 compared to the $425 Intel Core i7-6800K and $1,049 Core i7-6900K.

Ryzen 7 1700

The Ryzen 7 1700 steps back to a 65W TDP, hence the "world's lowest power 8-core desktop processor" designation, and it features 8 cores and 16 threads. It comes with a 3.0GHz base and 3.7GHz boost frequency along with 20MB of L2+L3 cache. AMD presented a multi-threaded Cinebench benchmark showing a 46% gain over the four-core Core i7-7700K. The Ryzen 7 1700 retails for $329, which is $21 less than the i7-7700K. Once again, AMD didn't present single-threaded benchmarks.

AMD Cooling Solutions And Retail Packaging

Ryzen processors will come with the logo etched on the heat spreader. AMD also unveiled its retail packaging and the new Wraith Spire cooling solution for select Ryzen processors, but details are slight. 

AMD indicated there would be 82+ motherboards at launch from the usual suspects, such as ASUS, Gigabyte, MSI, Biostar and ASRock. Full systems are available from 19 boutique builders. OEM gaming towers will follow in 1H17.

AMD's competitive pricing is aimed at the center mass of the desktop PC market--99% of the PC market buys CPUs below $500. AMD's pricing is definitely disruptive, but there are still some important facets we will discover in the coming weeks, such as performance in a more diverse range of benchmarks, single-threaded and gaming performance, and overclockability, among others. AMD's pricing strategy is encouraging, and it doesn't have to beat Intel in every category if it can provide a healthy price-to-performance ratio. A resurgent AMD will certainly help the struggling PC market and perhaps force Intel to alter its own pricing model.

AMD has a full lineup of new product launches this year, including Vega GPUs and Naples server CPUs in the second quarter. Ryzen mobile products with the Zen core arrive in the second half of the year. 

AMD indicated that it designed the pre-order strategy to satisfy pent-up demand, but it's always best to wait for product reviews and a more thorough examination before pulling the trigger. Products hit the shelves on March 2, and it's a safe bet that product reviews will arrive in the same time frame. Stay tuned.

OpenVPN, an open source VPN client on which a majority of VPN services rely, will be audited by cryptography and network security professor Matthew Green. The audit will be funded by Private Internet Access (PIA), one of the major VPN service providers in the United States.

Rise Of VPN In Surveillance States

Although VPN services saw much of their growth from customers looking to bypass geo-blocking of certain video streaming services, the new growth should be fueled by people’s desire to protect their privacy. With all the recent surveillance laws appearing in democratic countries, VPN services seem to have become more important than ever.

However, before using such a service, one also has to trust that it can guarantee the privacy it offers. Not too long ago we learned, thanks in part to Edward’s Snowden’s revelations but also to Dr. Green and his colleagues’ research, that up to two thirds of VPN service providers were vulnerable to NSA interception. That was because of the weak default Diffie-Hellman (DH) primes used by many internet servers, including VPN services, and the Logjam attack, which could downgrade connections to using the weak DH prime.

This sort of attack was believed to have been used by the NSA, and potentially other nation states, to easily spy on VPN users.

Dr. Matthew Green’s Audit

To ensure that such situations are avoided in the future and that there is no backdoor in the popular OpenVPN client used by most VPN service providers, Dr. Green will audit the open source software.

Dr. Green has also led the TrueCrypt audit project, has participated in the creation of Zerocoin/Z-cash privacy-friendly cryptocurrencies, has done research on Apple’s flawed iMessage encryption, and has commented on many other encryption and security issues more recently, including Android Nougat’s storage encryption weaknesses.

Dr. Green will audit version 2.4 of OpenVPN, which is the latest iteration (and is still in beta at the moment). As soon as it exits beta, the audit will commence.

“The OpenVPN 2.4 audit is important for the entire community because OpenVPN is available on almost every platform and is used in many applications from consumer products such as Private Internet Access VPN to business software such as Cisco AnyConnect,” said Private Internet Access in a recent announcement.“Instead of going for a crowdfunded approach, Private Internet Access has elected to fund the entirety of the OpenVPN 2.4 audit ourselves because of the integral nature of OpenVPN to both the privacy community as a whole and our own company,” PIA added.

After the audit is complete, PIA will share the results with the OpenVPN project and will work with OpenVPN’s team to fix all the issues before making the results available to the public as well.

Cisco’s Product Security Incident Response Team recently uncovered that the vulnerabilities revealed by the “Shadow Brokers” group as part of NSA’s set of hacking tools, were now being used against at least some of its customers:

“On August 15, 2016, Cisco was alerted to information posted online by the Shadow Brokers group, which claimed to possess disclosures from the Equation Group,” said Cisco in a recent security advisory.“The posted materials included exploits for firewall products from multiple vendors. Articles included information regarding the BENIGNCERTAIN exploit potentially being used to exploit legacy Cisco PIX firewalls.Based on the Shadow Brokers disclosure, Cisco started an investigation on other products that could be impacted by a vulnerability similar to BENINGCERTAIN.Cisco Product Security Incident Response Team (PSIRT) is aware of exploitation of the vulnerability for some Cisco customers who are running the affected platforms,” revealed the company.

Cisco did not reveal who exactly those customers were. However, the vulnerability in the IKEv1 protocol, used to negotiate cryptographic attributes (algorithm, mode, and shared keys) for communication sessions, exists in a wide variety of Cisco products. The devices most affected are those using Cisco’s IOS, IOS XE, and IOS XR software.

As Cisco said on its website, “Cisco IOS Software is the most widely leveraged network infrastructure software in the world.” That means it may now be an even more attractive target for malicious hackers due to its large deployment.

Its PIX firewall appliances version 6.x and below are also affected by the vulnerability, but version 7.x isn’t. Cisco’s PIX devices haven’t been supported since 2009. The company also confirmed that Cisco ASA 5500 and Cisco ASA 5500-X Series Adaptive Security Appliance devices are not affected by the IKEv1 vulnerability.

IKEv1 allows an unauthenticated attacker to steal the memory contents of devices, which could lead to disclosure of confidential information. Cisco said that there are no workarounds for this vulnerability until the company releases patches for its software. Until then, IT administrators are advised to closely monitor the affected systems.

Cisco said it will release updates that fix the vulnerability, but only those with valid licenses, procured directly from Cisco or from an authorized re-seller, will be able to receive them. It’s not clear whether PIX devices, which haven’t been supported since 2009, will receive any updates.

Before the world sees Ubisoft’s future plans next week at E3, the company decided to show one of its latest projects — Watch Dogs 2.

The sequel to the 2014 title will take place in San Francisco. Similar to Chicago (which was the setting of the first game), San Francisco also has the ctOS system to maintain its city infrastructure and surveillance, but the hacker group Dedsec aims to take the system down in an effort to remain free from watchful eye of Big Brother.

You play as Marcus Holloway, a young hacker from Oakland who was wrongly committed of a crime by the ctOS 2.0 system. As a fugitive, he eventually joins Dedsec in taking down the corrupt government as well as the technology that made him a criminal.

Similar to Watch Dogs’ Aiden Pearce, Holloway utilizes his parkour skills to traverse the city’s streets and rooftops. In addition to San Francisco, Holloway can also explore the areas of Marin County, Oakland and Silicon Valley. His weapon of choice seems to be a design of his own: a billiard ball attached to a long lanyard. Holloway also has access to a 3D-printed gun, a drone, and an RC car with a robotic arm.

Ubisoft also revealed a few details about Holloway’s hacking capabilities. The main focus was his ability to remotely control cars, which is used as a form of distraction or a way to kill enemies from a distance. You can also hack any individual on the streets in different ways, such as marking them as a threat so that police can arrive and arrest them, or you can ring their phone as a simple distraction.

The overall goal in Watch Dogs 2 is to gain more followers for the Dedsec cause. When you finish a mission, Dedsec gains more followers because of your actions. More people means that the group will use additional resources to take down ctOS 2.0, and Holloway is one step closer to getting his name cleared from the records.

You can also pre-order one of the game’s six edition variants. The Standard Edition ($59.99) includes the bonus Zodiac Killer Mission and a corresponding outfit. For an additional $10, you can get the Deluxe Edition, which includes the bonus mission and outfit from the Standard Edition as well as two Customization Packs. At $99.99, the Gold Edition includes all of the above plus the game’s Season Pass. There are also three versions of the Collector’s Edition that all come with an “app-enabled robot.” Each tier corresponds to the rewards of the Standard, Deluxe and Gold Editions and will cost you $109.99, $119.99 and $149.99, respectively.

You won’t have to wait long to play Watch Dogs 2. The game comes out on November 15, a mere five months from today. Those attending E3 next week will have the chance to try the game at the Ubisoft booth.

Follow Rexly Peñaflorida II @Heirdeux. Follow us @tomshardware, on Facebook and on Google+.

If you want to get your hands on Razer products, the first place you look is online at sites like Amazon, Newegg and the company’s own digital store. However, residents in Bangkok, Thailand; Taipei City, Taiwan; and Manila, Philippines can actually visit a RazerStore, a physical retail location selling all things Razer. On May 21, the company will open another branch, this time in San Francisco, its first U.S. location.

Specifically, the store is located at the Westfield San Francisco Centre, and it will take up nearly 1,300 square feet of space on two levels of the mall. Inside, customers can check out Razer’s lineup of products such as keyboards, mice, headsets, laptops and apparel. The store also features 20 gaming stations where customers can use Razer products to play games, listen to audio or see various peripherals in action.

With this RazerStore, as well as the Asian retail locations, the company joins the likes of Apple and Microsoft in providing a physical storefront to showcase its product lineup. Both Oculus and HTC are toying with the same with demos available at Best Buy, GameStop and Microsoft Store locations.

Unless you attended a tech or gaming show, it's usually rare for consumers to spend quality time with a Razer product before launch. With the RazerStore, you can now get first-hand experience with the latest peripherals and systems to give you a better idea of what to expect before buying anything. There’s no word yet on future expansion plans for other RazerStore locations in the U.S., but if the San Francisco store is successful, it could bring about more branches throughout the country.

Follow Rexly Peñaflorida II @Heirdeux. Follow us @tomshardware, on Facebook and on Google+.

Update, 3/15/16, 4pm PT: If you thought $399 was too good to be true, it is. During Sony's announcement, they led us to believe that the PSVR would be $399. In fact, we went back and listened to our audio recording, that is what was said from the stage. Not just the headset, but "PSVR." Later we got materials from Sony that paint a different picture, which is to say that price doesn't include the camera or the Move controllers.

Indeed, PlayStation UK eventually tweeted that "the PlayStation VR requires the PlayStation Camera, which is available separately." Further, the PlayStation VR website even states that the "PlayStation VR" will cost $399, but in small print it notes that the PS4 and Camera are sold separately. That, plus all the images Sony provided, would lead one to believe that the Move controllers are included in that price, but if you check the spec sheet, they are not.

Not every game will need the Move controllers, necessarily, but most people will surely want them. They cost $50 each, so you can tack another $100 on top of that $399, plus the cost of the camera, which is around $50.

Therefore, that $399 number is really a boondoggle. The real price, when you take into consideration all the peripherals you still need to buy, the PSVR as a whole will cost closer to $550. 

Original article:

Sony announced today at GDC in San Francisco that the PlayStation VR will be available globally in October for $399 (399 Euro, 349 GBP and 44,980 Yen, $549.99 CAD). The company also announced that the hardware has been finalized.

The specs of PSVR have not changed from what's been previously stated. As a reminder, it uses a 5.7-inch OLED display; with a resolution of 1920x1080, with R, G, and B subpixels; and a 120 Hz refresh rate. The company claimed a latency of less than 18ms. And of course, because it works with PS4, of which there are now 36 million according to Sony, developers have an exact target to hit, hardware-wise, without deviation.

The headset is visor style, and the company said it has been crafted to reduce pressure on a user’s face. It has a single band and a quick release button to make it easy to take on and off. The company also announced that it will include 3D audio that will work within a 360-degree space. That is, it will change in real time based on the orientation of your head.

A few other interesting tidbits: The PSVR will allow for cinematic mode, wherein you can project any PS4 game in 16:9 at three different zoom levels. There will be an updated media player that can expose any 360-degree content (photos, videos). And Playroom VR, a toybox of games from Sony Japan, will come as a free download. 

The company said there will be about 60 titles at launch, and today at GDC it is showing off about 20 games.

Our own Kevin Carbotte took the PSVR for a spin back at the Immersed conference, and he was surprised at how well the device performed.

Update, 03/15/16, 3:40pm PT: Included more comprehensive specifications chart

Fritz Nelson is the Editor-In-Chief of Tom's Hardware. Follow him on Twitter, Facebook and Google+. Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Sony has revealed the full list of what to expect at PlayStation Experience 2015 this weekend in San Francisco, and whether you are able to attend or have to watch from afar, Sony has plenty for you to take in at the event.

The Playstation Experience Keynote will kick off the weekend at 10am PT on Saturday, December 5. For those hoping to attend in person, the auditorium will open at 8am, but Sony suggested lining up sooner than that as space will be limited. For everyone not in attendance, the keynote presentation will be broadcast live on Twitch, and Sony suggested sticking around after it concludes. The company has over 30 live game demos and developer interviews planned throughout the weekend, in addition to a handful of panels to take in on Saturday afternoon.

If you will be attending Playstation Experience 2015, you’ll want to be sure to get the Experience Playstation App, which is available for iOS and Android devices. Sony said the app features a trackable program guide with details of all the booths and activities found at the event and an interactive map of the venue so you don’t get lost. The app can also be used to RSVP for Playstation VR demos, of which there are quite a few.

Sony revealed the list of “currently announced” playable Playstation VR titles on the show floor, insinuating that there could be some further reveals this weekend with playable demos at the event. So far, we know there will be at least 15 titles to try during Playstation Experience 2015.

There’s far more than just PSVR games being shown off during PlayStation Experience this year, though. Sony revealed that there will be over 130 playable titles for fans to explore.

On Sunday, starting at 10am sharp, the Capcom Cup 2015 Street Fighter world championship will kick off and run throughout the day. At the end of it all, the world will have a new Street Fighter champion, and that talented player will take home $120,000. Sony said this is “an event everyone should experience at least once in their lives.”

Sony will be broadcasting Playstation Experience 2015 on Twitch and live.playstation.com throughout the weekend. The keynote starts at 10am on Saturday, and the official event stream is scheduled for noon and will run until 9pm. Sunday’s broadcast resumes at 10am Sunday and wraps up at 4pm that evening.

Follow Kevin Carbotte @pumcypuhoy. Follow us on Facebook, Google+, RSS, Twitter and YouTube.

VRX is taking place this week in San Francisco, and Nvidia took the opportunity to make an exciting announcement about Gameworks VR and its forthcoming integration into Epic Games' Unreal Engine in the coming months.

When that happens, VR developers using Unreal Engine will have the ability to use Multi-Res Shading as well as VR SLI performance enhancements. Nvidia believes the addition of the integration will rapidly increase adoption of Gameworks VR features.

Some of the features being integrated into the engine are VR SLI, which enables each GPU to handle rendering for a specific eye. Nvidia has also considered systems with more than two GPUs installed by including a GPU Affinity feature that will let VR games scale over the extra processors.

Multi-Res Shading is also part of the Gameworks VR SDK. This is a technique that renders the image in different resolutions depending on the focus. VR lenses warp the image, leaving only the center in focus. Multi-Res Shading takes advantage of this and renders the least in focus areas with less clarity, which Nvidia said improves performance substantially where it counts.

Nvidia said that the company has seen as much as a 50 percent performance improvement in the UE4 Reflections Subway demo when using Multi-Res Shading, but the work on it continues. The company said that integration into the engine is still a couple months away, so final performance improvements could be better than internal tests are showing today.

The version of the Unreal Engine that first sees Gameworks VR integration has yet to be finalized, but the company expects version 4.11 to be the first public release.

Follow Kevin Carbotte @pumcypuhoy. Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Specifications, Features And Accessories

Today we're looking at the Linksys PLEK-500 Powerline adapter kit. Packaged individually, these Powerline adapter model numbers don't have the "K," so if you're looking for the two-piece starter kit, be sure to search for PLEK-500. On Linksys' website, the PLEK-500 is found in its "Wired & Wireless Range Extenders" products section, and is the company's most recent Powerline offering. Interestingly, the branding on our review unit belongs to Cisco, but since Belkin now owns Linksys, I'm sure newer versions will have that removed.

Specifications

MORE: Powerline Networking 101
MORE: How We Test Powerline Adapters
MORE: All Powerline Content


MORE: All Networking Content

According to Linksys' specifications, this model implements the HomePlug AV2 standard. Though its range isn't published, the number we saw in our HomePlug AV2 Powerline Networking Adapter Round-Up is 300 meters, so we can probably expect similar range in this case. Advertised data rates for this adapter claim up to 500 Mb/s. However, what isn't published is the power consumption or operating range in megahertz. What sparks my interest most is the fact that this Powerline adapter has a grounding pin in the male plug components. According to the HomePlug AV2 specification, that means MIMO can can be utilized.

The Linksys PLE-500 is also absent from the HomePlug Alliance's list of certified products with its HomePlug AV2 filter applied. However, I did find it in the filter list under HomePlug AV. Product packaging labels the PLE-500 kit as a "Powerline HomePlug AV2 Kit," and although the box sports a HomePlug AV Certification Mark, consumers don't really know if these adapters comply with the HomePlug AV2 standard.

This could be due to the HomePlug Alliance website not receiving dynamic updates for product certifications. If not, it shows a loophole in the HomePlug Certification Mark guidelines where product packaging can cite the logo and claim an alternate revision of the standard through a named reference in the product label. Is that what Linksys did here? I doubt it, but the point is that these discrepancies can cause confusion for the consumer. For more details on the HomePlug Certification Mark, check out our discussion of it in our How We Test Powerline Networking Adapters article.

Features

On the side panel, there's a button called the "HomePlug Simple Connect Button," which sets the Powerline network password. Communications in the Powerline network membership are then protected via 128-bit AES encryption, including key management. Pressing the button on one of the adapters starts a syncing period, during which you have to press the button on the kit's other adapter so it will sync appropriately within five minutes. If you want more granular control over network membership, you can download the Linksys Powerline Utility, which is discussed in a later section.

Linksys also includes a power-saving mode that you can tell is active when the Power LED flashes at a slow rate. Unfortunately, the consumption numbers aren't published, but it's nice of Linksys to think of our electrical bills.

Note that when you initially pair the adapters, the Power LED flashes at a slow rate at that time as well, but according to the user guide, they're performing the initial pairing.

Accessories

Inside the box we have the two Powerline adapters in a cardboard cutout, two RJ-45 cables, a quick-start guide and a software utility disc. There's no need to read a manual or instal the software utility; the PLEK-500 is plug-and-play. I think the RJ-45 cables are three-footers, so keep that in mind in case you need longer cables for your adapter placement.

Hardware, Software And Tear-Down

The adapter's form factor is similar in proportion to the D-Link DHP-600AV we reviewed in our round-up, except its chassis is more rounded. The PLEK-500 has what shows as a slight decline from the outer edge to the wall-adjacent edge, which provides clearance for a three-prong cable plugged in above from a slight angle. When comparing wall protrusion to the D-Link DHP-600AV, the Linksys PLEK-500 seems to stick out a little farther than an inch and a half.

There's a gigabit Ethernet port on the bottom, next to which is a hardware factory reset button. On the right side, seen in the picture above, you'll find the security button. Pressing it on one of the two Powerline adapters starts the re-sync process, and a randomized encryption key is generated.

There are three LEDs: Power, Powerline and Ethernet, and they indicate different statuses. While the Power LED is an on/off indicator similar to the Ethernet LED, the Powerline LED actually fluctuates between green and yellow. Yellow indicates a transmission rate lower than 15 Mb/s, while green means your transmission rate is above 15 Mb/s.

Software

In this case, Linksys provides a software utility for advanced configuration. Although it's not included in the starter kit, you can download it from the company's support site. Prior to using the utility, make sure that your Linksys PLE-500s are plugged in and accessible. Once you launch the tool, you'll see it has three tabs (as shown in the screenshot below).

Network Status is the first tab. We can see in the Devices List window the adapters discovered, along with the model number, MAC address, throughput quality and device password.  Under Detail Information you'll find the firmware version, which could be handy if Linksys pushes out an update at some point in the future.

On the Security tab, you can set the Network Membership Key for selected devices. The default key is printed on the back of the Powerline adapter.

The Administrator tab lets you restart the adapter, set it to factory defaults or upgrade its firmware.

I didn't need to use the advanced utility during testing, but it's nice to know you can upgrade the firmware this way.

Tear-Down

The first step in cracking open any case is to try to figure out if there are hidden screws. Some vendors hide them under rubber guards, but in this case, I found a screw behind the serial number bar code.

I cut away the relevant part of the label to reveal a hex-sided screw. With it loosened, the next step was to feel around the edges and find the plastic clips that needed to be depressed.

You can tell in the picture above and below that I had to do some guess-work to figure out exactly where those plastic clips were. To save you some time in case you want to break open the case yourself, here they are highlighted:

With the top off, we can get a sense of the parts inside:

Qualcomm Atheros AR8035-A: Marketed by Qualcomm as an ultra-low-power single RGMII gigabit Ethernet PHY and part of its Ethos product line (datasheet). Other interesting marketing details include:

• Supports voltage input/output levels from 1.5V to 3.3V
• "Error-free" operation up to 140 meters of CAT5 cable (that's about 160 meters short of this adapter's marketed 300 meter max range)

Qualcomm Atheros QCA7450/AR1540: According to a 2012 press release, this chipset is marketed by Qualcomm as its first that complies with the HomePlug AV2 specification. Reportedly included is a "SmartLink Plus" option to enable Powerline signal transmission on multiple wires. The SmartLink Plus option is elaborated on in this press release as utilizing all three electrical wires: line, neutral and ground. In the HomePlug AV2 specification, MIMO is a required capability, so this may be Qualcomm Atheros' implementation of that requirement,

UMEC UTB02123: This is a "Flyback Transformer" cited on the UMEC website as "reinforced" for the Atheros AR7450 chipset.

Test Results And Conclusion

Comparison Powerline Networking Adapters

Solo Round

We tested the Linksys PLEK-500 using the system configurations and procedures outlined in our How We Test article. Here's how the Linksys PLEK-500 performs:

According to traditional thoughts of signal degradation, the two gigabyte folder results follow the logic that the farther away the Powerline adapters are, the worse they perform. We then switched from testing a simple file copy via file shares to benchmarking throughput using Passmark's PerformanceTest 8 utility.

When using the PerformanceTest software, results were best the closer the Powerline adapters were to each other. While the graph above shows the absolute peak value observed through the testing period of three minutes, the graph below shows that the average throughput stayed relatively close to peak values.

In The Mix

Now let's compare how well Linksys stands up to the top three performers in our Powerline adapter round-up. First, the 2GB folder transfer:

Measured against the top three performers of our recent round-up in the Same Room test, the Linksys PLEK-500 falls just below ZyXEL's PLA5215 in aggregate numbers.

Linksys again appears under the top three when extending the length of the test from two outlets in the same room to two outlets on the same floor.

For the next test, I placed an adapter on the second floor, keeping the other on the first floor.

Linksys trails behind the top three yet again, and the trend continues in the next test of the first floor to the basement.

Saving the best for last, Linksys pulls ahead into the bronze position in the second floor to basement test.

Next, we switch our testing software over to PassMark's PerformanceTest8 and repeat the tests from the same positions to see how Linksys' PLEK-500 compares to the top three in each test from our prior round-up.

While not too far behind third place, Linksys' PLEK-500 finishes below our top three performers in the same room test.

In the same floor test, however, Linksys falls further behind the top three.

When testing from the second floor to the first floor, Linksys not only performs below the top three, but overall ends up the worst performance I observed in any test, even comparing Linksys' own numbers. Not sure what happened there.

In testing performance from the first floor to the basement, Linksys again falls short of breaking into the top three.

As if it had something to prove at this point, Linksys' PLEK-500 adapters edge their way into third place in the final test.

All of these tests show the peak values observed, but the averages yield the same finishing order. In short, Linksys takes a lower spot than the top three performers. Also, in the UDP tests, Linksys' PLEK-500 adapters completely flooded the switch I was using, invalidating the results of any UDP tests. Basically, there was too much packet loss.

Conclusion

Newegg lists the Linksys PLEK-500 kit for $100, and although it was one of the lower-performing Powerline adapters in our testing, its throughput remained above 55 Mb/s in every benchmark. Also, having the ability to upgrade the firmware through Linksys' advanced configuration utility might be a good thing if a future upgrade increases performance.

Overall, the PLEK-500 is a solid set of Powerline adapters worth considering if you can get them for less than competing devices, which are typically faster.

MORE: Powerline Networking 101
MORE: How We Test Powerline Adapters
MORE: All Powerline Content


MORE: All Networking Content

Matthew Matchen is an Associate Contributing Writer for Tom's Hardware. Follow him on Twitter @matchemm

Follow us on Twitter @tomshardware, on Facebook and on Google+.

Asus wants us all to know how much it loves gamers, and at its Republic of Gamers (ROG) Unleashed event in San Francisco, the company made it clear why. Taking the stage, Kirk Skaugen, Intel's SVP of client computer, laid out the numbers: Gamers refresh their PC system every two to three years, versus six years or more for general or business users. PC gamers are also spending more on gaming software than their console counterparts. While the general PC industry may be struggling, Asus, Intel and Microsoft want us to know that PC gaming is flourishing.

This helps explain the success of Asus' ROG program itself, a line of gamer-focused products that turned 10 this year. At ROG Unleashed, Asus revealed an array of new gaming hardware, including one of the craziest laptops we've ever seen.

Impressions

Desktops

Under the dark red lighting and pounding techno, Asus had a demo display showing off its new desktop and laptop lines. Keeping with Asus' fine tradition of baffling product names no one will ever remember, the compact desktop system is the G20CB, and the full tower comes in two flavors: the lower-end G11CD (for real?!), and the high-end G11CB.

Internally, the systems have almost identical options: Up to 32 GB DDR4 2133 MHz memory, up to a 512 GB M.2 SSD, and either the AMD R9 380, or Nvidia cards from the GTX 700 and 900 series. The only place they really differ is at the super high end, where the G11CB has the option for a hybrid SSHD and a GTX 980 Ti, while strangely, the compact G20CB has no option for a 980 Ti, but can be outfitted with a GTX Titan X.  

With that being the case, unless the lack of a 980 Ti is a dealbreaker for you, I can't imagine why anyone would opt for the larger, full tower G11CB. It's a large system in a fairly generic case, while the G20CB is not much bigger than an Xbox One, with an excitingly architectural style, and lighting effects that can be customized to pulse to ambient sound, which is a surprisingly cool effect.

Both of these systems will be re-released soon as "Oculus Ready editions," which simply means they'll meet the minimum system requirements to run an Oculus VR. Asus reps said they are planning to package their Oculus Ready computers with a discount voucher for the headset, likely $100 off.

Laptops

Desktops are all good and fine, but clearly the point of the entire ROG Unleashed event was to show off Asus' new laptop line. There were two models on display, both massive 17.3" affairs, weighing nearly 10 lbs, and clearly designed more for high performance than portability.

The G752 is the more normal of the two, but it's still a unique beast in its own right. The G752 has a vapor-chamber cooling design, a common cooling technology on desktop systems, but unique in a laptop. The system uses water to channel heat away from the hot components to a central radiator. Essentially, the heat of the CPU or GPU vaporizes the water on contact. The vapor is then channeled to the radiator, where it cools, condenses, and recirculates to the hot chips. The G752 uses two large radiators with blower fans to keep the system running, giving the back end of the laptop something of a hot rod effect.

Other than this innovative cooling system, the G752 is a fairly standard, high-end gaming laptop. It comes with options of Intel Skylake i7 chips (up to an unlocked i7-6820HK), a GeForce GTX 980m with either 4 or 8 GB GDD5, and various SSD / HD combos, up to a 512 GB SSD with a 1 TB HDD. It can also be equipped with a truly ludicrous 64 GB of DDR4 (more than the new desktops, it must be noted).

One nice feature is that all models of the G752 come equipped with a USB Type-C port. This particular port can be used either as a Thunderbolt 3 connection or 10 Gbps USB 3.1 port. It can also deliver up to 15 watts of power to a connected device. The 1080p screen looked as good as most Asus monitors do, with great color and contrast.

The G752 starts at $1,999, but if you want those high end specs, be prepared to shell out $3,499.

Now for the real showstopper, the Asus GX700. This laptop is a strange device, somewhere between totally ludicrous and just darn cool. Its specs are almost the same as the G752, with a few key differences. First off, it can be outfitted with two 512 GB SSDs using PCIe x4, it adds a second USB Type-C USB 3.1 10 Gbps port, and it has a full, desktop-grade GeForce GTX 980. It powers that massive graphics card with a one of a kind dock that not only delivers 330 W of power, but also hooks the laptop into a closed loop liquid cooling system.  

Outside of the dock, the GX700 shifts its graphics capabilities down to be closer to a GTX 980m, and it uses a 180 W power adapter to control heat. (Interestingly, the massive cooling fans on the G752 means it operates on a 220-Watt power adapter, suggesting that the laptop by itself will be more powerful than the GX700 outside of its dock.)

The computer snaps into the dock with a release on the top, and when it's hooked in, not only does it receive power, but four, no-drip nozzles latch onto the back of the machine, and connect it to the liquid cooling system and massive radiators built into the top of the dock. Disconnecting only takes seconds, and interestingly, the GX700 doesn't flush the coolant, (which is not water), back into the dock; its internal heat pipes stay filled with liquid. An Asus rep admitted that there can be some small coolant loss when the laptop is disconnected, but said they guarantee the system for at least two years, though after that you may need to get the dock serviced and recharged.

The display models we saw were all using 1080p screens, although a 4K option is supposed to be forthcoming. One rep said they kept with 1080p because they were worried that 4K would degrade the user experience, leading me to think that there are still some aspects of drivers or power management that need to be sorted out.

Overall I was impressed with these new laptops. They are massive, weighty powerhouses, but with beautiful screens, and surprisingly good keyboards. The keys were nicely clicky, with a good travel distance, and sufficient feedback. The machines feel fairly solid, challenging for devices this big, though they creaked and flexed a little more than I'd like.

That said, there are a few aspects of the designs that I wish Asus had improved. Maybe we've just been spoiled by sleek, well-designed laptops from Dell, Lenovo and Apple, but I wish there was a little more finesse to the build of these behemoths. 

For example, although the displays are beautiful, and 17.3 inches is a significant screen size on a portable computer, they're surrounded by huge bezels. Some quick Pythagorean math confirms that these machines are actually more than 21 inches corner to corner. It just feels like wasted space, and it gives them a little bit of a Laptop-From-The-90's vibe when opened.

On both models, the radiator assembly at the back also sticks out a fair amount past the screen hinge. These aren't fatal flaws, and clearly if you want to buy one of these machines you likely care more about performance than design, but for more than $3,000 of machine, it would be nice if it came in a more attractive package.

Check out the rest of our coverage of ROG unleashed.

Follow us @tomshardware, on Facebook and on Google+.

The unlocked midrange flagship space in the U.S. is getting a bit crowded with vendors like OnePlus, ZTE, Alcatel and now even Motorola all vying for your dollars. Now, upstart Nextbit is the latest to throw its hat into the ring. Don't worry if you've never heard of them -- they have a better pedigree than most, made up of former Android engineers combined with the design chops of former HTC head of design Scott Croyle.

Its first phone is called the "Robin," and with it, Nextbit is also taking a novel approach to storage, utilizing the cloud to offer almost limitless device storage. It launched the Snapdragon 808-powered phone on Kickstarter today for $300 for the first 1,000 early bird backers, which sold out quickly, and $350 thereafter (and then $400 after the Kickstarter ends).

Before Nextbit decided to make a phone, it licensed its cloud technology to be included with Android phones sold by Japan's NTT Docomo and were also working with Cyanogen to add its tech to its OS. For whatever reason, though, this approach wasn't getting what Nextbit thinks is a revolutionary approach to mobile device storage into enough people's hands. So it decided to go it alone, and take its technology back in-house and only offer it on its own device to help bring its vision to the masses.

What the Robin does differently is link together the phone's 32 GB of internal storage with an additional 100 GB of cloud storage. This, of course, means there is no need for a microSD slot (cue the angry comments). This is nothing new, you might say. Everyone has Dropbox/Google Drive/One Drive on their phones, but where the Robin differs is that it doesn't just back up your photos and documents into the cloud, but entire components of the phone's software.

It intelligently figures out which applications and data you haven't used for a while and sends them up into the cloud. What are left on your phone are ghost icons of the archived app or data, and one touch brings it back down from the cloud. Also, it backs the apps up in the state they were last at, so when you go back to an app after pulling it back into your phone, it starts off just where you last left it.

For those concerned that this unattended shuffling of your data from phone to cloud is going to chew through your data plan, it defaults to doing this on Wi-Fi only. Though for those with giant data buckets, this can be turned off. Helpful LEDs under the logo on the back of the phone let you know when the Robin is connecting to the cloud. Because it's constantly backing up your device, it can even archive unused information if storage space is needed when you're offline, knowing that if the data is needed again, it's only a trip to the cloud away.

What Nextbit hopes is that by leveraging the cloud, it can make a phone that gets better over time, but intelligently reconfigures itself to fit your current usage. It also means that, eventually, the phone in your pocket will be just one of many panes of glass that gives you access to all of your apps and data that now reside permanently in the cloud.

Along with its cloud-first technology, the other stand-out aspect of the Robin is its design. Seeing as its design came from the mind behind many of HTC's iconic handsets like the HTC One M7 and M8, the design of the Robin is quintessentially HTC-like. Or should we really say One & Co-like, as that's the design firm Scott Croyle was working at when HTC acquired it to bolster its design department.

The Robin is made from metal and plastic with a Gorilla Glass 4 display, and it has a unique squared-off design and choice of colors (mint and midnight) that helps it stand out from the sea of similar phones.

You can also see a lot of attention to detail, at least in the 3D renders we've been presented so far, from the perforations of the speaker grills on the front to the placement of the camera lens and sensors. Despite this apparent level of craftsmanship that should elevate it above other $400 phones, we are a little concerned by its ergonomics. We're not sure how comfortable its squared-off design will be to hold for extended periods.

Specifications

While the integration with the cloud is the Robin's headlining feature, the rest of the phone is nothing to sneeze at, either. As you can see from the specs above, although the Robin isn't a powerhouse device (to be expected from its price point), it is still a very handsomely equipped phone.

At least for now. There is the question of whether a Snapdragon 808 phone will be relevant four months down the road (when the first Kickstarter backers are due to get their units). However, if you look at the mobile SoC roadmap for the next six months, Nextbit doesn't really have any other choice, because the 808's replacement won't be out for some time. Although its specs do seem a little modest in other areas too, remember this is a $400 device, so there must be some compromises made.

We are pleased to see that equipping it with a decent set of front-facing speakers was important to Nextbit – this is a feature that oh so many smartphone vendors neglect. We also look forward to seeing how well the side-mounted power button/fingerprint scanner combo works, and it's nice to see that the Robin uses a USB Type-C connector for charging, and has Quick Charge support.

There are still quite a few questions remaining about the Robin's specs, and it may be simply because at this early stage perhaps they haven't locked down everything yet. For example, the information about the camera is quite vague, and we have no idea who is making its sensor. Of course, Nextbit could also be leaving some specs deliberately vague so it has the flexibility to perhaps upgrade some with Kickstarter stretch goals.

Other than the cloud integration, the Robin basically runs stock Android, lightly skinned with Nexbit's teal and gray RobinOS theme. The Robin is also more than just SIM unlocked -- it's also bootloader unlocked so you can install custom ROMs, and Nextbit promised that it will be "bloat free." It also said it will always be running the latest version of Android, so expect it to ship with Android Marshmallow unless there is some unforeseen delay in its release.

Although going the Kickstarter route is certainly less of a risk for Nextbit, there have been a few high-profile cellphone crowdfunding failures of late (the YotaPhone 2 on Indiegogo) which may scare off some potential backers. Despite the company's founder's backgrounds, with the Robin not in production yet, and the reliance on third parties (even if it is Foxconn), backing a Kickstarter is always a gamble.

The other problem is that by announcing this early in the development cycle to get funds from backers means that we won't be seeing the Robin is people's hands anytime soon. As we mentioned, the estimated delivery for early backers is January 2016, and that is if there aren't any of the characteristic Kickstarter delays. Regular backers won't see theirs until a month later at least, and those who miss the Kickstarter will have to wait even longer. Then, as we also mentioned above, who knows how well the Robin's specs will hold up in four to six month's time.

Despite all that, it doesn't seem that these concerns have deterred too many people, because presently, Nextbit has already blown past its initial $500K campaign goal. With 29 more days to go, Nextbit is surely going to raise a pretty big chunk of change.

Update, 9/02/15, 12:25am PT: Now that the initial goal has been reached, Nextbit has updated the Kickstarter with the first stretch goal. If it hits $1 million, "everyone gets a quick charger with their order.” Considering that quick charging is an advertised core feature of the Robin, it’s a little odd that a stretch goal is needed to provide such an essential accessory.

When the Oculus Rift finally ships, after nearly three years of manipulated anticipation and "hurry up and take my money" buyer frustration, it is going to blow your ever-loving mind. The additional industrial design touches Oculus' put on the final headset hardware, the use of XBox One controllers, and the where-did-this-come-from Oculus Touch (the company's new sneaky-obvious wristwear controllers), will all only add to the allure.

Oculus has never shied away from explaining that it has taken this long because the company wanted to get it right -- apparently down to the fabric on the headset.

I have watched the Oculus crescendo, practically from the beginning, experiencing the many demonstrations along the way, and I'm not going to hide it: I am in the tank for VR and for any company that will accelerate the adoption of VR, and for now one of those companies is undoubtedly Oculus.

My disclosures: None. My benefits are the same as yours, as a fan of technology advancement, as someone anxious to be surprised by the next big thing, as someone skeptical about almost anything pretending to be that. But I am not at all skeptical about this.

(Wait. Scratch that. I got a coffee mug and a t-shirt at the grand unveiling on Thursday.) 

I have pushed Tom's Hardware editors and myself to -- is there a more apt word for it? -- immerse ourselves in this technology because I believe it is, literally, a game changer in every sense. It will change the way games are conceived and played, the way we are entertained (and sure, read further into that as well, if it suits you), and the way interactive experiences and other human-oriented applications evolve.

I have tried every demo at every opportunity, as have many of our team. One of our editors bought DK2. We have Gear VR. We have tried Microsoft Hololens, HTC/Valve Vive. We will try them again. These things don't get old.

In short, I cannot wait. But that's just the problem. In less than nine months, the Oculus Rift will ship. That's still almost a year away, which is ages in tech time.

Beyond that, the hardware requirements to fuel the Oculus Rift are beyond the reach of too many. The typical consumer doesn't build PCs. Surely not even the majority of the Tom's Hardware crowd are packing the equivalent of an Intel Core i5 and an Nvidia GeForce GTX 970 and 8 GB of RAM. 

Would you upgrade for high quality VR? Some undoubtedly will, others will wait, and still others probably should wait until there's more content. Despite all of the demonstrations we've seen, it is still hard to believe that enough content will be ready. Anna Sweet, head of development strategy for Oculus, announced on Thursday that Oculus is giving away $10 million to spur game development. On the one hand, that's a rounding error for Facebook, and on the other, it's also a cry for help -- Oculus needs content, and it needs it yesterday.

Games like CCP Games' EVE:Valkyrie, Playful's Lucky's Tale, and many others are fun, but the demonstrations have been mostly of prototype games, snippets meant to excite interest. I'm told there will be more at E3, including a much-enhanced Lucky's Tale, now one year past my first experience with it. Will people even play VR games for an hour or more? Will the big studios jump in, or leave the early work to independent studios?

Oculus CEO Brendan Iribe proclaimed on Thursday: "For the first time, we will finally be inside of the game. It will change everything. We've been dreaming of it. It's finally here." (If by "here" you mean next year.) The more sobering fact is that when it ships, there will still be far too much work to do.

If Thursday's presentation and discussions revealed anything, it's that high quality VR takes time. It takes attention to detail. Without that attention to detail, the Oculus Rift and the experiences it delivers would not stand out.

The Details Matter

With each incremental dribble of news, there's a sense that Oculus is watching and learning and perfecting the first product it will ship. Did you catch, for example, that Oculus CEO Palmer Luckey said he had watched thousands of people using the early development kits? I've seen him, excited as a 12-year-old, hurrying a guest off to see a demo. I've seen him watch their reaction. (Don't tell him I was watching him watching them.)

(At the event yesterday, he practically ran me over to put the Touch controllers on a pedestal so all of us taking pictures could also see and capture them, too. I suggested he act as our head and hand model, and he gladly obliged, with gun gestures and various angles, talking about the nuances of the controllers like Willy Wonka describing the inner workings of his chocolate river.)

Over the past couple of years, as each person tried on the headset over time, jostling it on their misshapen heads, squeezing the headset like an accordion to make the lenses closer together, moving it gingerly around the frames of their designer glasses, mouths open like mesmerized toddlers at the circus, sometimes puking or feeling nauseated, the Oculus team had apparently been taking notes. The company has accounted for and fixed each of these problems.

The attention to detail on the final Rift headset almost seems to steal a page from the Apple playbook. Executives talked on Thursday about the unit's "strap architecture" and even called it a "facial interface." (Mm, OK.)

But once we're past the theater of it, it's also important to understand that unless a company thinks this way, that something like the strap of a headset is so important as to have an "architecture," then perhaps it's not taking the details seriously enough. If you play video games for an hour or more at a time, then comfort matters. If you've tried the various VR demos, you'll know that these details are important, whatever fancy nomenclature Oculus invents.

Oculus said that the Rift, now covered in some sort of fabric (what, no thread count details? No zigzag stitch architecture mumbo jumbo? No hand-rubbed seamless contours?), is lightweight (I tried to find out exactly how lightweight, but unfortunately I don't keep a portable scale on me), doesn't pull on your face, doesn't move around, it goes on like the old familiar ballcap, it fits around glasses with ease, it adjusts to each of our ocular idiosyncrasies, and the company believes it achieved its primary goal of making it so that you don't even know it's there, that it "just disappears."

Apple: It just works. Oculus: It just disappears. Apple: It feels good in the hand. Oculus: "It feels great when you put it on." 

Get it?

Not a stone is being left unturned. The Rift has an unmatched field of view and visual resolution. AMD and Nvidia have created new rendering schemes and exposed GPU functions for VR developers. And so on. But Oculus recognizes that there's still more work to do. Iribe, talking about some of the visual achievements accomplished so far, also indicated that the resolution wasn't as high as it will be one day. 

The new and final Sensor, which wirelessly syncs with the headset, is small and includes a constellation tracker for precise head tracking. Oculus didn't leave audio to chance either, promising that the latency needs to be nonexistent here, as well -- "Rift tricks your eyes, but it tricks your ears, too," Iribe said.

What of those images leaked from the Oculus website, the one that had a camera in the front of the headset? It was an old design concept. Oculus decided that the camera might overwhelm the processors with too much information right now.

The Xbox One controller came after what Oculus product VP Nate Mitchell described as years of work on the company's own controller. Those prototypes eventually became the Touch controller, Mitchell told me. Balance and ergonomics on a controller was difficult, Mitchell said, again pointing to the level of detail Oculus seems to put into everything it has created.

Iribe echoed those thoughts, pointing to how Sony "showed a consumer design earlier than anyone else, and yet they're really taking their time to get it right." 

Iribe also said that both the Touch controllers and the final headset design came from the company's acquisition of the Carbon Design team last year -- that's the team, he said, that created the Xbox 360 controllers and various keyboards and mice for Microsoft. Iribe said that Oculus is a team of mostly hardware and software engineers, so the company saw fit to bring in the experts to get the industrial design right.

I tried to do a little social engineering on Oculus execs to ascertain how it worked. For instance, it's wireless -- what kind? Not saying. What does it connect to, the PC or the tracking system? Not saying. What sensors are being used? Not saying. How is latency being managed if it's wireless? Good question, more details will emerge on this, possibly even at E3. (And can it check my heart rate? My sobriety? Or remind me to check on the kids?)

Half Moon, the prototype name for the Touch, will be out in the first half of 2016, which is to say after Rift ships. It provides high precision, low latency, and six degrees of finger tracking. It uses inertial measurement to do all of that tracking, and the intent is to allow precise manipulation of virtual objects as you would in real life. But it also includes traditional inputs such as buttons, a trigger, and an analog stick, and there is haptic feedback, as well. 

Mitchell did say that the controllers use the same kind of constellation tracking as the Sensor and pointed out the IR LEDs on the rings, and he indicated that more details about Touch would emerge in time. We'll have a chance to try it out at E3.

Amid even more wordplay about things like "hand presence," and "low mental load," there's yet another message in here. While Oculus tried to create a controller, they decided not to reinvent the same old thing, instead inventing something entirely new that made sense with the Rift -- something that they could control as part of the entire system and ensure that it kept to its low latency demands.

Remember that head tracking, audio synchronicity and hand tracking must be precise to feel real. Oculus' vision of VR is very much focused on the "R" part, and three years of engineering work is driven precisely by that. When Iribe talked about mobile, console and PC VR, he emphasized that PC VR, in the form of the Rift, is where you fully believe that you're "there," with a full sense of presence, because of the high fidelity the PC can muster.

In a land o' leaks, this was truly the first we had heard of Touch. In addition to the hardware, Luckey talked about the Toybox, a sandbox where Oculus has created robots to control, garden gnomes to punch, and tetherballs to smack. Again, these should be ready for us to try out at E3. 

Content 

Above all, content is the final hurdle. Jason Rubin, the head of Oculus Studios, invited a few guests on stage, including the CEOs of CCP Games, Gunfire Games, and Insomniac. Once again, the content looked compelling, if brief. Rubin talked about some of what we'd see at E3, and he finally showed a list of developers working with Oculus. 

It's an impressive list, but save for Square Enix, there's not a major game developer among them, and Rubin wouldn't say what the game developer was creating for the Rift. If this truly is the opportunity to re-create gaming, where are the big boys? Rubin insisted that "big studios will come and will play inside VR and they're extremely excited about it." He hinted that there would be much more prior to the official Rift launch next year.

I pushed further, asking what hurdles those developers saw, and he talked about how the independent developers started early and had a passion about VR from the get go. The bigger studios require more thought and more reasons to jump into VR, he said, and they are "not driven by the same immediate desire to fulfill the dream...of making worlds in VR."

When asked about the nascent nature of VR and the lack of an installed base, he countered that most of the developers realize that the learning curve for VR is steep, as it has been for other transitions in game development, and that those who "don't jump in early aren't fast followers, they're the ones who never catch up." In other words, the bigger players get it, and they know they must experiment, "even if an installed base isn't there on a spreadsheet."

Rubin did say that most studios typically have three-year plans for games, but he said he's encouraging them not to think that way, that so much is changing that three years is too long. (Yes, three years is too long. I think I've heard that somewhere before.)

Rubin said that most of the demonstrations and games we've seen so far are 15 minutes or less, and that this is a good way to figure out whether you have enough to make a full game out of the experience. He called 2014 the year of the demos, said 2015 will be about making what sells, and asserted that 2016 will be about making longer VR games.

Some day. Some day.

Meanwhile, I have the mug and the t-shirt.

Fritz Nelson is the Editor-In-Chief of Tom's Hardware. Follow him on Twitter, Facebook and Google+. Follow Tom's Hardware on Twitter, Facebook and Google+.