Loongson Technology has announced the Loongson 3C3000, a new 16-core server processor aimed at low-cost general-purpose server systems. Unveiled on June 26, 2026, via a public corporate release, the chip is based on Loongson’s in-house architecture and is designed for small- and medium-sized business workloads, including file, database, web, and business process servers. Loongson says the 3C3000’s general-purpose computing performance matches that of the company’s earlier 3C5000 server processor.

The Loongson 3C3000 — which is based on a 64-bit architecture and supports the LoongArch instruction set — uses the company’s LA364E processor core design and comes in an FCBGA1371 package measuring 37.5mm by 37.5mm. The chip is pin-compatible with the Loongson 3B6000 processor, which should make it easier for system builders to reuse existing platform designs.

The processor features 16 physical cores and 16 threads, with clock speeds ranging from 1.5 GHz to 1.8 GHz. Each core supports 128-bit vector instructions and three-issue out-of-order execution. Loongson says each core integrates two fixed-point units, one vector unit, and two memory access units.

Cache and memory support are modest by modern server standards, but in line with the chip’s low-cost positioning. Each core includes 64KB of private L1 instruction cache and 64KB of private L1 data cache, while all 16 cores share 16MB of L2 cache. The integrated memory controller supports two 72-bit DDR4-2400 memory channels with ECC support, giving the processor server-class error correction for business and infrastructure workloads.

For expansion, the 3C3000 provides two PCIe x16 interfaces, totaling 32 PCIe lanes. These can be split into up to four x4 or x8 interfaces, depending on platform requirements. The chip also includes another PCIe x16 interface that can be configured as LCL (Loongson Coherent Link) for dual-processor interconnection. Other interfaces include SPI, UART, three I2C interfaces, AVS, and 16 GPIOs.

Loongson lists a typical power consumption of 40W when running at 1.5 GHz. The processor supports dynamic shutdown of the main module clock and dynamic frequency adjustment of the main clock domain, helping reduce power consumption under lighter workloads. It also integrates a Loongson-developed security and trust module that supports Chinese national cryptographic algorithms for encryption and decryption.

Unlike the 36C000, which Loongson says matches Intel’s Xeon, the company is positioning the 3C3000 as a low-cost, high-performance server CPU for customers that need general-purpose compute rather than high-end acceleration or AI performance, slotting it below the higher-core-count 3C6000 server lineup the company launched a year earlier. Support for China’s local software and hardware ecosystem also appears to be a selling point. The company has not publicly disclosed pricing, a common move for server processors, which are often sold through system builders and negotiated enterprise deals rather than as boxed retail chips.

Tramplin Electronics, a Russia-based microelectronics company, has announced that it has obtained what it claims to be the first samples of its Irtysh processors based on the LoongArch instruction set architecture (ISA) developed by Loongson, a Chinese CPU developer, reports Tramplin.Media. The processor features 16 or 32 cores, and their specifications and packaging match Loongson's LS3C6000 CPUs, so we most likely are looking at re-badged products from the Chinese company.

Tramplin Electronics claims that it has obtained its 16-core Irtysh C616 and 32-core Irtysh C632 processors for sovereign data centers and HPC applications. Notably, Russian firms cannot rely on industry-standard x86 CPUs from AMD and Intel because they are sanctioned by the US and cannot be legally obtained from nearby countries. You can see the Russian chips in the flesh by expanding the tweet below.

The company's product catalogue also includes the 64-core Irtysh C664. The CPUs are based on the LA664 microarchitecture that features a 6-way out-of-order execution and simultaneous multi-threading (SMT) technology, as well as support 128-bit vector processing extension instructions (LSX) and 256-bit advanced vector processing extension instructions (LASX) for the new CPUs. Loongson and its Russian ally Tramplin claim that LA664-based CPUs are competitive against AMD's Zen 3 and Intel's Ice Lake-based offerings.

Among the advantages of the Irtysh processors, Vasily Vorobushkov, director of development at Tramplin Electronics, cited a proprietary boot environment, high energy efficiency, stable production, and uninterrupted supply. Development is said to be handled by the company's own team of engineers and designers, along with partner companies that build complete hardware-software solutions. The company is also said to operate its own design center focused on developing domestic IP blocks and maintains a broad ecosystem, though Vorobushkov did not elaborate on which blocks and elements of the ecosystem have been developed by Tramplin so far.

The specifications of Tramplin's 16-core Irtysh C616 (2.20 GHz, 32MB L3, quad-channel DDR4-3200 memory, 844.8 GFLOPS, 100W – 120W TDP) and 32-core Irtysh C632 processors (2.10 GHz, 64MB L3, octa-channel DDR4-3200 memory, 1612.8 GFLOPS, 180W – 200W TDP) are identical to those of Loongson's 16-core LS3C6000/S and 32-core LS3C6000/D CPUs down to a single number, which isn't something that happens usually unless we are dealing with the very same silicon.

Indeed, Tramplin Electronics was first registered on April 4, 2025, so the company is less than a year old. It is impossible to develop a processor from scratch (even based on a known/licensed ISA), find a production partner, build its physical design, tape it out, and get samples in this short of a time frame. In fact, a year is barely enough to bring up a new CPU based on an existing platform (this may easily take a couple of years for a company of AMD's or Intel's size), not to mention developing one from scratch. That said, even though the processors were made in the third week of 2026, it looks like these are regular Loongson LS3C6000 CPUs that carry Cyrillic inscriptions.

Now that Russia-based entities cannot legally obtain high-performance CPUs from companies like AMD or Intel, the only way for the country to retain access to more or less contemporary processors is to buy them illegally in nearby countries, or get Chinese processors from the People's Republic. Apparently, we are dealing with the second option here, albeit with an attempt to disguise Chinese processors as those developed in Russia. Interestingly, the source of the river Irtysh — after which the CPUs are named — is in China.

It's rare to find Loongson CPUs outside of China, but a Linux reviewer has managed to get one of the CPU manufacturer's 12-core chips for testing. Phoronix reviewed Loongson's 12-core 3B6000 processor in a plethora of Linux-based benchmarks. Despite its high core count, Phoronix's benchmark numbers revealed that the chip can't even approach the performance of modern Western six-core chips like the Ryzen 5 9600X.

The 3B6000 was reportedly sent over to the Linux-based outlet by the Loongson Hobbyists Community. The 3B6000 came inside a 3B6000x1-7A2000x1-EVB micro-ATX with two DIMM slots, one M.2 slot, two PCIe x16 slots, and just a handful of USB ports.

Phoronix benchmarked the 3B6000 across dozens of benchmark results, including applications with AVX-512 support. Overall, the 3B6000 was in basically last place, performing three times worse than AMD's six-core Ryzen 5 9600X. The only CPU the chip was able to overtake was the quad-core ARM CPU in the Raspberry Pi 500.

That said, the 3B6000 did not have such gloomy performance in all of the benchmarking applications Phoronix ran. For instance, in C-Ray 2.0, the 3B6000 was able to match the performance of the Ryzen 5 9600X, and in OpenSSL 3.6, it approached the performance level of the Core Ultra 5 245K. In QuickSilver 20230818, the 3B6000 managed to outperform the 245K slightly and achieve a score on par with the Core Ultra 9 285K. But these were the exceptions; in almost all other benchmark runs, the 3B6000 was significantly slower than all of AMD and Intel's x86 chips.

These performance results aren't surprising at all. The 3B6000 runs at just 2.5GHz, which is half the clock speed of most of Intel and AMD's latest chips. The Chinese chip's LA664 CPU architecture reportedly has the IPC performance of Zen 3, but the chip's abysmally slow clock speed hinders any strong IPC gains the architecture might have.

Loongson is working on newer CPU architectures that it hopes will be much faster than what the 3B6000 and its LA664 CPU architecture offer right now. Last year, we heard that the Chinese CPU maker is working on a new LA864 architecture that will reportedly to potentially have 13th/14th Gen Raptor Lake Intel CPU performance. Chips based on this architecture will also reportedly have much better clock speeds of 3 to 3.5GHz. It's still nothing close to the 5+GHz that Intel and AMD can reach, but it is a big jump for Loongson regardless.

Multiple publicly traded Chinese companies have notified their investors that the recently imposed tariffs from the trade war do not concern them, reports the South China Morning Post, largely because US sanctions have already prevented them from selling into the US. 

China's high-tech industry has a number of world-class companies that develop products that are competitive far beyond Chinese borders. But given the sanctions imposed against these entities in recent years, it is hard to find products from Huawei, Loongson, or Longsys outside of China.

Cambricon Technologies (an AI processor developer), Loongson (CPU designer), Leaguer Microelectronics (an IoT IC designer), Longsys Electronics (a maker of storage systems), and Maxscend Microelectronics (an RF chip developer) all said that they were not going to be impacted by massive tariffs imposed by the U.S. government in their filings for investors, according to SCMP. 

Based on the SCMP report, these companies will indeed not suffer from the punitive tariffs the U.S. has imposed on products from China: 

• Cambricon stated that revenue from overseas markets contributed less than 1% of its total income in both 2023 and 2024. The company had already been impacted by its 2022 placement on the U.S. Entity List. The new trade measures would not significantly affect its business.
• Huawei apparently did not make any noticeable announcements, though its business in China, Russia, and Iran will clearly not be hurt by the U.S. government's sanctions.
• Loongson Technology reportedly said that the new import duties have had no negative impact on its operations.
• Leaguer Microelectronics disclosed that its products are made in China and bought in China. This internal sourcing shields it from external disruptions.
• Longsys Electronics, a major supplier of storage solutions, reportedly said that its Brazilian subsidiary plays a key role in buffering the effects of ongoing trade disputes. As it sources 3D NAND memory from all major vendors, it can probably supply to customers in the U.S.
• Maxscend Microelectronics noted that its RF chips are focused on domestic supply, so the U.S. market and tariffs are not an issue.

Hardware makers from China ship tons of hardware, including domestically developed chips and domestically developed systems on those chips. That industry has been aimed mostly internally so far though, so no financial impact on the aforementioned companies at this point. However, what about indirect impacts? That remains to be seen. 

Chinese chipmaker Loongson recently announced in an investor call its 3B6600 CPU, which will likely succeed the 3A6000 processor that debuted in April 2024. According to Fast Technology (machine translated), the new desktop chip features eight cores and will have an integrated GPU. But, more importantly, the company claims that the Loongson 3B6600 is competitive with Intel Core i5 and i7 chips of the Alder Lake and Raptor Lake processor families. While these processors are already four years old, having been launched in November 2021, they still mark an advancement over the 3A6000 which is compared to 14nm chips, which were in production from 2014 to early 2021.

The 386600 CPU is expected to maintain the 2.5GHz frequency of the previous generation 3A6000 chip. However, its single-core turbo frequency is expected to increase by 20%, allowing it to hit as much as 3GHz. Loongson even claims that its single-core performance would be one of the world leaders, but we’d like to see proper benchmarking to back this up. Aside from its single- and multi-core performance, the new chip will also support DDR5 memory, PCIe 4.0, and HDMI 2.1 output.

The company has been working on these chips since April 2024, and we expect them to arrive on the market later this year. Despite that, they’re still years behind the cutting-edge chips produced by TSMC, AMD, and Intel. Because of this, even local companies are seemingly reluctant to adopt these homegrown processors, so Loongson is likely facing an uphill battle even as China blocks Western-made chips from government offices and servers.

Despite that, Loongson's silicon has come quite a long way, especially as the company seems to be starting from scratch and doesn’t have the decades of expertise that American and Taiwanese firms have. American sanctions are likely pushing Chinese companies to develop alternatives to chips that have had export controls applied to them, and we’ve seen innovations like “tock-tock-tick” strategy to help close the gap.

Loongson - the Chinese chipmaker gunning for Intel and AMD - has joined the cosmic competition (literally) as its chips are now available in space. ITHome reports that China's latest Tianzhou-8 spacecraft was packaged with a space-based system powered by Loongson processors - reportedly fabbed using an unnamed "advanced" process node. This cargo spacecraft delivered six tons of supplies and materials to the Tiangong Space Station and successfully docked within three hours of flight.

The new space-computing platform was dispatched via the Tianzhou-8 payload and successfully installed on docking with the Tiangong Space Station. This is a major step forward for China as it looks to achieve self-sufficiency in the semiconductor realm. A domestic and locally produced chip mitigates the risks associated with foreign supply and reduces China's dependence on Western hardware.

While the report doesn't go into specifics about the processor - it is likely based on a modified version of the MIPS ISA (Instruction Set Architecture). There are also various mentions that this Loongson chip leverages a cloud computing setup - though that could be a mistranslation. As per the details, the chips are equipped to undergo self-testing and offer on-orbit processing. They have dedicated storage, networking capabilities, and a thermal management system.

The report suggests that the chip is designed to execute tasks such as ground radiation source identification, which is likely to detect fissile material on Earth, optical remote sensing, monitoring electromagnetic radiation emitted or reflected from the Earth, and image capturing. The key notion here is that the Loongson chips do all this while in orbit under harsh space conditions - including but not limited to cosmic radiation exposure and limited power resources.

The payload also delivered several utilities and packages to the astronauts - like birthday gifts and other seasonal offerings for the Spring Festival, Mid-Autumn Festival, and the Dragon Boat Festival to keep the indigenous cultural spirit alive - even in space. The supplies also included lunar regolith simulant, which is essentially synthetic moon dust and is currently being employed to study the feasibility of space stations on the moon.

The Chinese chip industry still faces several hurdles - primarily the inability to source EUV equipment. This has crippled progress for SMIC and even Huawei, whose flagship Ascend AI chips are now limited to 7nm technology until 2026 - per rumors. While this is a step in the right direction, China must find breakthroughs in EUV technology if it wishes to stay competitive against global players.

Chinese CPU manufacturer Loongson is purportedly closing the gap quickly to Western CPU manufacturers regarding raw performance. Fast Technology reports that Longsoon's fourth generation is adopting a "tock-tock-tick" strategy to accelerate innovation.

Over the past several years, Loongson has taken inspiration from Intel, using its widely known tick-tock strategy to innovate on its past three generations of CPUs. The tock refers to architectural innovation, designing a new chip architecture around an existing processing node. The tick represents porting that existing architecture to a new design node. Intel generally does not utilize this design strategy anymore, with its latest Arrow Lake chips functioning on a brand-new architecture and a completely new process node.

However, Loongson has adopted the tick-tock strategy for the past three generations, bringing consistent performance improvements to its chips. That all changes with its fourth-generation chips, however. The Chinese manufacturer has purportedly decided to switch gears, utilizing a new strategy deemed tock-tock2-tick. This method adds a second architectural design optimization "pass" before the design is ported to a smaller node. Effectively, this gives Loongson's designers more time to improve their CPU designs before moving them to a new process node.

Fast Technology reveals that the first chips to operate on its first tock were the 3A6000, 3B6000M, and 3C6000, nicknamed the "Three Musketeers." Chips running on the second tock are the Loongson 3A6600, 3B6600, and 3C6600, with the difference being the "600" numbering scheme added to the chips trio. This trio of chips is Loongson's current iteration of CPUs.

The 3B6600, for example, is claimed to perform at par with Intel 12th—and 13th-generation CPUs thanks to its new eight LA864 cores with a clock frequency of just 3 GHz. It will be tapped out in 2025.

Loongson's fourth generation of CPUs is still well into the development phase, and we have no word on what CPU models will be powered by Longsoon's next-gen design. Regardless, if Loongson's third-gen chips can already compete with Intel Raptor Lake CPUs, it stands to reason that these fourth-gen chips could compete directly with Zen 5 and Arrow Lake. Loongson's fourth-gen chips will purportedly operate on a new process node comparable to 7nm nodes from competing Western manufacturers.

Chinese chipmaker Loongson has a new GPU on the way that blows its current-gen 9A1000 out of the water. According to a senior company executive speaking today, the upcoming Loongson 9A2000 will deliver performance that is “8-10 times that of 9A1000,” reports Chinese language tech site ITHome. 

The machine translation suggests that the new 9A2000 is comparable to Nvidia’s GeForce RTX 2080. Nvidia’s Turing architecture RTX 2080 is still a very serviceable graphics card in 2024, as evidenced by our frequently updated GPU benchmarks rankings.

The headlining information comes from Loongson Chairman and General Manager Hu Weiwu, who spoke during a Q&A session at the firm’s 2024 semi-annual results briefing. Hu answered a question about the status of the current generation 9A1000 and positioning plans for the 9A2000.

The answer from the top Loongson exec revealed that the 9A1000 and 9A2000 GPUs will co-exist as they address pretty different markets. Specifically, the 9A1000 GPU will be taped out by year-end and address entry-level users with what is expected to be Radeon RX 550-level performance. The 9A1000 card will be suitable for budget systems, with minimal 3D rendering demands, and where 32 TOPS of AI inference is sufficient.

However, we were much more interested in hearing some of the first details regarding the Loongson 9A2000 GPU. Hu stated that the new 9A2000 GPU’s performance will be “8-10 times that of 9A1000.” Moreover, for those of us needing a more familiar frame of reference, he told briefing listeners that “it is comparable to Nvidia RTX 2080.”

PC enthusiasts outside of China might not flock to a Loongson GPU that can match the 2018 vintage RTX 2080 and currently lacks DX12 support (DX12 compatibility is just a ‘maybe’). However, if the price is right and the performance lives up to the tease today, some may consider buying a 9A2000 for testing, tinkering, or even out of sheer curiosity.

The Loongson 9A2000 GPU and Black Myth: Wukong gaming

An RTX 2080 is still higher than most PC games’ minimum to recommended spec levels in 2024, but will you be able to enjoy titles like Black Myth: Wukong?

Thankfully, Hu answered this question during the Q&A. The exec explained, “In terms of GPGPU performance, the display performance of 9A2000 should be able to support it.” We guess that’s one way of saying no and hinting that the platform/drivers won’t allow gamers to have any anthropomorphic monkey fun unless there are some key software and driver developments. However, we note that DX11 is indeed enough to run this title.

We published a deep dive on Black Myth: Wukong PC gaming last month. Referencing our GPU hierarchy data, an RTX 2080 user could likely enjoy about 90 to 100 FPS in the game at 1080p medium settings (no RT). Officially, the recommended hardware levels call for an RTX 2060 or better GPU, but we aren’t sure what quality settings and frame rate that target.

Six major Russian supercomputer centers located across the country have teamed up to form the 'Distributed Scientific Supercomputer Infrastructure' consortium, reports the Russian Academy of Sciences. This new alliance aims to pool resources for the performance needed to process modern scientific workloads. In total, the six supercomputing centers have around 900 servers and these fleets cannot be upgraded due to the U.S. sanctions. To that end, pooling resources is largely a forced measure. Meanwhile, DSSI has a plan to collectively upgrade their hardware.

Six supercomputing centers join forces

The consortium's consolidated infrastructure includes 900 servers with a total peak performance of 1.5 FP64 PetaFLOPS and scientific data storage systems exceeding 15PB. This infrastructure supports distributed systems for the collection, storage, and analysis of scientific data across various regions. To put the number into context, 1.5 FP64 PLOPS performance trails the slowest supercomputer in the Top 500 list, which has Rmax performance of 2.13 PetaFLOPS. Members of the DSSI are providing computing resources and support to 240 organizations across the country and by working together they will be able to offer more compute performance to their customers as the latter's requirements increase. 

The consortium does not include Russia's most powerful state-owned supercomputer, the Lomonosov 2 located in the Lomonosov Moscow State University, which has Rmax performance of 2.48 FP64 PFLOPS. It should be noted that the MSU introduced its MSU-270 supercomputer with a peak computational power of 400 'AI' PFLOPS last Fall and this one also does not participate in the DSSI project. Commercial supercomputers, such as Yandex's Chervonenkis, Russia's most powerful machine with Rmax Linpack performance of 21.53 FP64 PFLOPS, will also not be a part of the consortium. 

The DSSI consortium brings together multiple players from different regions, including the Far Eastern Federal Research Center in Khabarovsk, the Institute of Automation and Control Processes in Vladivostok, the Matrosov Institute in Irkutsk, the Institute of Computational Mathematics and Mathematical Geophysics in Novosibirsk, the Krasovsky Institute in Yekaterinburg, and the Space Research Institute in Moscow.  

"We need to move away from competition between collective-use centers and instead focus on uniting their resources and expertise," said Alexey Sorokin, director of the computing center of the Far Eastern Branch of the Russian Academy of Sciences and a member of the consortium's Coordinating Council. "There is a growing demand for a distributed infrastructure that connects technological sites across different regions of the country via high-speed communication channels. These sites should be equipped with modern computing systems of various architectures and specialized data storage systems. It is impossible for even the largest organization to tackle this complex task alone. It makes sense to leverage the groundwork laid by existing supercomputer centers, which have already proven their efficiency. This approach will undoubtedly lead to more efficient use of financial resources and, just as importantly, foster the development of scientific teams in the regions, particularly in HPC space."  

And have upgrade plans

Due to sanctions imposed by the U.S. after Russia invaded Ukraine, Russian state-owned entities have a hard time acquiring supercomputer hardware, such as the latest processors from AMD, Intel, and Nvidia. While smuggling exists (and largely prospers), it is likely that Russian institutions cannot get enough new machines to modernize their fleets.  

Pooling in resources will enable Russian scientists to get access to higher performance. Also, the DSSI consortium aims to develop a comprehensive project to upgrade scientific supercomputer centers for collective use with modern compute hardware. After discussions with all interested participants, the project will be submitted to the Ministry of Science and Higher Education of the Russian Federation and the Russian Academy of Sciences. 

Given the sanctions imposed by the U.S., EU, Taiwan, South Korea, and other countries, it will not be easy to upgrade all of the aforementioned datacenters with modern Western hardware. But perhaps the plan is to use parts that can be obtained without any curbs: including usage of cut-down components like Nvidia's HGX H20 or hardware from Chinese developers. However, it is evident that the Russian scientific community wants access to higher performance.

Following the successful tape-out of its 16-core 3C6000 processor in February, Loongson has been bringing up the CPU and has tested actual samples of the chip. According to data shared by the company at the 2024 Global Digital Economy Conference in Lhasa, the processor is just as fast as Intel's 16-core Xeon Silver 4314 processor, reports ITHome.

The Loongson 3C6000 is a monolithic chip packing 16 cores featuring the LA664 proprietary MIPS-derived microarchitecture supporting simultaneous multithreading technology (SMT) that operates at an undisclosed frequency. It also has four DDR4-3200 memory interfaces and 64 PCIe Gen4 links. The CPU's performance is supposedly comparable to that of Intel's 16-core Xeon Silver 4314 processor, released in Q2 2021 and powered by the Ice Lake microarchitecture. However, Loongson hasn't disclosed the specific benchmarks.

Loongson once said it expected its LoongArch 6000 microarchitecture to match the instructions per clock (IPC) performance of AMD's Zen 3 cores. Hence, the current performance indicators are more or less in line with that expectation, a notable achievement for the Chinese CPU designer.

There is a catch, though. According to the report, Loongson's 3C6000 will be available in the fourth quarter of 2024, and production will ramp up in 2025, so the processor will still be about four years and three generations behind Intel's Xeon 6 for servers.

There are some other things to note as well. One of the notable innovations in the Loongson 3C6000 is the introduction of the Loongson Coherent Link (LoongLink) technology. This interconnection enables chiplet-to-chiplet communication, akin to AMD's Infinity Fabric and Nvidia's NVLink. This technology makes Loongson's 3C6000 essentially a building block for constructing CPUs with higher core counts.

In particular, Loongson is working on its dual-chiplet 3D6000 with 32 cores and 64 threads and quad-chiplet 3E6000 with 64 cores and 128 threads. We don't know when these processors will come to market, though we expect they'll only arrive after the initial single chiplet version, mostly likely some time in 2025.

Two new motherboards for the Chinese-made Loongson 3A6000 CPU are now available on AliExpress. These motherboards are available in various configurations ranging from a barebones motherboard with CPU through a complete prebuilt computer that also includes RAM, a hard drive, and a discrete graphics card.

AliExpress has two motherboard models featuring the Longsoon 3A6000 that are now beginning to ship worldwide. The first, the Loongson 3A6000 7A2000 main board, sells for $372.91 without memory, a storage device, or a discrete graphics card. It includes the Loongson 3A6000 processor with a maximum clock speed of 2.5 GHz.

The 7A2000 main board offers modest expansion possibilities, with a total of three PCIe slots, two DDR4 memory sockets, and support for both NVMe SSD and SATA drives. It also includes HDMI as well as VGA graphics output, plus HDA audio interfaces.  

The second motherboard, the XC-LS3A6M, costs a little more at $389.93 and offers slightly more by the way of expansion possibilities. It provides the same PCIe expansion slots as the 7A2000 but also includes an extra Ethernet port and a header for a front Type-C USB 3.2 port.

Other available bundles for both boards allow consumers to add memory and storage, or purchase the motherboards preinstalled as a complete PC, with or without a discrete GPU installed.

Due to the lack of any further specifications for the Loongson 3A6000, we assume these motherboards only support PCIe 3.0 and nothing newer. This does go along with Loongson’s comparison of its CPU with the Intel 10th Gen and AMD Zen3 CPUs, both of which are rather long in the tooth. An earlier release of a mini PC featuring the Loongson 3A6000 also advertised performance on par with those older CPUs.

The fact is, Loongson has developed its own CPU just a few generations behind Intel and AMD. Considering the Chinese firm has to work around U.S. sanctions and claims to insist on doing everything based on its own IP, the performance could actually be considered impressive.

With the wider availability of these Loongson processors / motherboards, due to the reach of AliExpress, we hope will will see some reviews by western media, with clear comparisons against AMD and Intel solutions.

This week, Chinese CPU developer Loongson published 105 programs from 53 developers that natively support its 5000- and 6000-series processors based on the proprietary LoongArch architecture. As the list revealed, Lenovo has quietly deployed Loongson's processors in its datacenters and is running cloud services on them, reports The Register. The scale of the deployment is unclear, but the revelation highlights Lenovo's commitment to using Chinese CPUs.

For now, Lenovo offers three software packages that support Loongson's LoongArch-based platforms: Wentian WxSphere Server Virtualization System Software V8.0 (16-core 3C5000L/3C5000), Wentian WxCloud Cloud Computing Management Platform V3.0 (16-core 3C5000L/3C5000), and Wentian WxStack Hyper-converged System Software V8.0 (quad-core 3A6000). For Lenovo, this is enough to deploy Loongson's 5000-series CPUs commercially for its cloud services and prepare to deploy the next-generation Loongson's 6000-series processors. 

Loongson has quietly gained traction in China with mini PCs aimed at the channel, NAS, and the education sector. These moves align with China's increasing urgency to replace Western technology with homegrown solutions, driven by policy objectives and necessity due to U.S.-led sanctions. 

Deploying 16-core 3C5000 processors for cloud services is something new, but it shows that Lenovo is confident in these CPUs and their successors, which will feature up to 128 cores. Lenovo's support for Loongson's architecture is crucial in making Chinese hardware a viable alternative to existing enterprise technologies. This support is expected to challenge companies like AMD and Intel, especially given China's vast market, which includes major telecommunications companies with extensive customer bases. 

It is unclear whether it makes much financial sense to use 16-core CPUs for cloud services nowadays, as there are more powerful equivalents from traditional x86 CPU vendors specifically architected for such workloads. However, Lenovo needs to learn how Loongson's CPUs behave with its instances today and try out the next-generation DragonChain microarchitecture-based processors that will be rolling out over the next couple of years.  

Notably, Lenovo's software stack is not the only cloud platform in China to support Loongson's processors; there are ten more platforms from various vendors, so there are more Loongson-based cloud deployments in the country. 

China's domestic chip production is managing impressive feats regardless of U.S. restrictions. CPU developer Loongson announced to shareholders that its upcoming 3C6000/3D6000/3E6000 series of server-grade processors had been successfully sampled and returned and that these first chips are now being tested. Loongson plans to release the 3C6000 line in Q4 2024, exactly matching its roadmap. 

Loongson's new server chip line is unique for its chiplet-based architecture. While the 3C6000 is a 16-core/32-thread processor on its own, the 3D6000 contains two 3C6000 chips connected with Loongson's "Loongson Coherent Link" technology, creating a 32-core/64-thread processor. The 3E6000 takes this up another notch, connecting four 3C6000 chiplets for a monster 64-core/128-thread beast. 

Loongson compares its Coherent Link to the performance of Nvidia's NVLink, though perhaps a better comparison would be to AMD's Infinity Fabric, which enables its Zen chiplet-based architectures. The 3C6000 contains 16 LA664 cores with Loongson's LoongArch microarchitecture. Unfortunately, no information on the speed or performance claims of the 3C6000 has been confirmed, but Loongson claims the LA664 cores offer double the performance of their predecessors inside the last-gen 3C5000, which operated at 2.2 GHz. The 3C6000 family will also support up to four channels of DDR4-3200 memory. 

Chiplet architectures have long been considered the future of microprocessors in a post-Moore's Law world, and nowhere is this more true than in China. As U.S. sanctions restrict Chinese chipmakers from access to advanced chip processes, chiplet architecture is viewed as a way to allow Chinese developers access to more performance in a roundabout way. However, Loongson is relatively unique in its pursuit of cutting-edge chips within China, as many chipmakers have transitioned to investing fully in legacy chip production, a niche China is dominating. 

Loongson's efforts have been impressive despite regulations and Chinese industry trends moving back to legacy production. Thanks to its own in-house MIPS-based LoongArch ISA and the use of domestic Chinese fabs, Loongson can avoid U.S. trade blockades on the greater Chinese chip market. Loongson has been populating China's schools with high-power domestic chips, and its most recent chip releases meet Intel's 10th-gen single-core performance. 

As China continues to ban non-domestic computers for government use, Loongson will continue to grow its market share. Intel CEO Pat Gelsinger may claim that Chinese chips are ten years behind American chips, but that may not matter for Loongson — and if the 3E6000 manages an impressive showing, it may not even be true.

Loongson is one of the few CPU developers from China not expected to be directly hit by U.S. sanctions. It uses the MIPS-originated LoongArch instruction set architecture (ISA) and employs a China-based foundry to make its chips. This is perhaps why Loongson has quite an ambitious roadmap that spans across markets and includes datacenter CPUs with up to 128 cores, according to slides published by EE Times China.

Loongson's CPU product roadmap until 2025 outlines an ambitious plan to improve performance and expand market reach across desktops, laptops, and servers. As expected, the company aims to enhance single-core/single-thread performance, develop up 128-core server CPUs, and introduce energy-efficient mobile processors to compete with AMD and Intel in client PCs.

The roadmap describes desktop CPUs, starting from the 3A1000 series and advancing to the 3A5000 and 3B5000 series between 2020 and 2022. Significant improvements are marked by the introduction of the 3A6000-series for client PCs in 2023, which operates at clock speeds between 2.5 GHz and 4.0 GHz using the LA464 architecture, an alleged Loongson ISA. These CPUs aim to offer performance akin to that of Intel's 10th generation Core processors, according to the story.

When it comes to server-class CPUs, things are getting more interesting. The 3C/D/E6000 series processors are set to be launched by 2025, featuring up to 128 cores. This is the pinnacle of Loongson's server CPU development, and we also know that this 128-core product is based on multiple chiplets, another major development for China's semiconductor industry.

In the laptop CPU segment, Loongson has its 2K series, starting with the 2K2000 (for now) and progressing to the 2K3000 and 2K5000 by 2025. These CPUs are designed for energy efficiency, but Loongson does not really go beyond that.

In the meantime, Loongson's roadmap highlights Loongson's strategic focus on catching up with multinational CPU developers and their performance standards (e.g., AMD, Apple, Arm, Intel), aiming for significant improvements in single-core performance over the decade.

Loongson's roadmap clearly illustrates the company's commitment to its own progress in particular and perhaps China's technological self-sufficiency in general, particularly important in the context of geopolitical tensions and trade restrictions. The use of the proprietary LoongArch architecture represents a significant step towards reducing reliance on foreign technologies and enhancing domestic capabilities. However, production at SMIC may be a whole different story.

Zhang Ge, the CEO of Loongson, revealed the company's next-generation octa-core 3B6600 and 3B7000 CPUs for the Chinese market. Both these CPUs are made for the mainstream market, emphasizing single-core performance rather than multi-core. 

Loongson admitted that while the 3B6600 and 3B7000 are behind the competition in multi-core performance, the company has significantly closed the gap in single-core performance. Loongson claims a 20X uplift compared to the company's previous processors. We've already seen the 3A6000 in action, exhibiting a similar IPC level close to AMD's Zen 4 and Intel's Raptor Lake.

The 3B6600 features eight LA854 CPU cores and an LG200 on-chip graphics core clocked at 3.0 GHz. The 3B7000 CPU also has eight LA864 cores clocked with 3.5 GHz, but it doesn't mention its iGPU.

Based on the information provided by the company, the 3B7000 supports PCIe 4.0, SATA III, USB 3.0, GMAC, and HDMI output. It also has integrated INT8 Tensor accelerators for LLM workloads. It also supports OpenCL 3.0 computing and OpenGL 4.0 graphics acceleration. Its on-chip graphics can get up to 256 GFLOPs of computing performance, and with a discrete graphics solution, it can achieve up to 1TFlops. There's no other information left, but the mini PC with the 3A6000 CPU only supported 4K at 30Hz. Hence, there's a possibility that this iGPU will lack some functions and support. Interestingly, it seems to support multiple RAM interfaces such as SRAM, SDRAM, DDR2, and DDR3, followed by DDR4.

There is conflicting information presented in its slide: one shows these chips to be 'next-generation notebook chips,' while the roadmap shows both to be desktop processors. Time will tell. While these chips are made for the domestic market, one can't help but speculate about the possibility of seeing them retail in a country affected by the sanctions imposed by the US and the EU.

Loongson is ambitious to compete against x86 and ARM offerings with its 'Dragon' architecture. With China already blocking the use of Intel and AMD CPUs for its official government use and schools ordering domestically-made chips, Loongson will have a much larger share of the domestic market in multiple sectors. However, its success remains to be seen in its mainstream market.

The IPC comparison with the Loongson CPUs reminds me of Pat Gelsinger's comment about Chinese-made CPUs being decades behind. But with companies like Loongson making such progress, one wonders if the domestic chipmakers can close the gap sooner than ten years.

Making CPUs is easier said than done, but China has consistently imported chipmaking tools by hook or crook. The sanctions would be seen as a blessing in disguise for the Chinese chip production business in terms of market share and innovation. That said, Russia still has to achieve the same level of success with its chipmaking ventures.

A Chinese-made Loongson 3A6000 SoC is making its way into the retail market in the Morefine M700S mini PC. The CPU isn't a top-of-the-line performer compared to currently available AMD and Intel chips, though; The A36000 CPU's performance is equivalent to an Intel Core i3-10100F and about Zen 3 levels of IPC. 

Loongson's chip has four cores and eight threads with clock speeds between 2.0 and 2.5 GHz. At max, it supports up to 64GB DDR4-3200 SO-DIMM memory. According to the manufacturer, the SoC is made for industrial use, government, education, medicine, and other fields. 

Based on earlier reports about this CPU's performance, it makes sense to pitch such a device towards these fields, which wouldn't need the highest-end performance but will appreciate a PC with a small footprint and adequate I/O options. The M700S retails for 2,799 Chinese Yuan, the equivalent of $387.

The M700S looks like a NUC-style mini PC, measuring 149 x 145 x 50 mm, and weighs 812 grams. It supports 2.5G and 5G dual-band Wi-Fi 5 and Bluetooth 5.0 with one PCIe 3.0 slot for an M.2 drive and a SATA connection for a 7 mm 2.5-inch drive. The built-in graphics supports a 4K resolution at 30 Hz and dual screens via two HDMI ports. The M700S has four USB 2.0 ports, two USB 3.0, and one USB Type-C with 30W PD 3.0. For wired internet connection, it has one Gigabit Ethernet port and a 3.5 mm headphone jack for audio.

According to the listing, it is pre-installed with Linux-based Loongnix OS but is compatible with many operating systems like Windows 10, Windows 11, Tongxin UOS, Galaxy Kirin, and Ubuntu.

Challenges and Opportunities in Sanction-Imposed Nations

Making a CPU from scratch is no small feat. This CPU will likely be featured in desktops and notebooks intended for some fields. Loongson may make its CPUs available in Russia and Belarus as the countries have trade sanctions imposed by the EU and the US. Russia is currently struggling with its chip-making venture; about half of its recently manufactured chips are defective. The country is also transitioning towards its locally made software since Microsoft is shutting down its Office 365 cloud computing services. Loongson may enjoy a certain market share in Russia for now because of these issues.

Despite that, it will take time for companies like Loongson to make CPUs that rival Intel and AMD's performance. Intel's Pat Gelsinger has said the chipmaking companies in China are ten years behind. But as China has been on a shopping spree for chip-making tools and with opportunities to do business in countries imposed with sanctions, we'll see if it's able to catch up.

As China is accelerating its transition to its own computer platforms, local developers of processors are set to benefit. Recently the Hebi city government procured as many as 10,000 computers based on processors from Loongson for 50 schools in the city. The new PCs are not only based on domestic hardware, but they also use homegrown software, according to Weixin.QQ.com. Loongson helped to adjust learning experiences and eventually could ship considerably more CPUs for China's educational sector.

The 10,000 desktop systems are based on Loongson's 3A5000-series quad-core processors based on the homegrown LoongArch GS464V instruction set architecture and will be used in the Hebi city and Qi Bin District. These computers run on the domestic UOS operating system and come with a full set of licensed software, including the WPS Office, the Mythware Classroom Management software, as well as 104 applications covering a wide range of educational and management needs, according to the report.

Selling 10,000 CPUs is hardly a big deal for Loongson as this is essentially a relatively small commercial batch of processors. But this is a pilot program that involves 50 schools. If the program achieves its goals and is considered a success then the company will be able to sell considerably more of its CPUs, which will be a big deal. 

In addition to building desktop computers, Loongson has reportedly helped develop educational materials by assisting in the creation of six textbooks based on the latest standards. Loongson has also supplied teaching robots and set up education scenarios in some pilot schools.

Looking ahead, Loongson plans to expand the 'Hebi Experience' to more regions across the country, which will provide the company with a plethora of new business opportunities to sell its computer platforms to the educational sector. 

According to a recent Financial Times report, China has begun a policy shift to phase out foreign processors from its government computers and servers. This move is aimed at gradually replacing AMD and Intel processors from PCs used by Chinese government agencies with domestic CPUs, which will decrease sales of American CPUs and increase sales of Chinese CPUs in the government sector. 

Loongson has taped out its 16-core LS3C6000 processor, designed primarily for servers, reports MyDrivers. The new CPU is based on the latest iteration of the LoongArch instruction set architecture, and Loongson believes that its latest LA664 can offer performance comparable to AMD's Zen 3 cores.

The Loongson LS3C6000 processor packs 16 LA664 cores supporting simultaneous multithreading technology (SMT) that operate at an unknown frequency. The new cores are interconnected using the company's proprietary DragonChain technology, which is supposed to resolve bottlenecks associated with expanding the number of processor cores. This technology is supposed to enable the company to build 32-core and 64-core processors in the future. 

Loongson expects that its upcoming CPUs based on the LoongArch 6000 architecture will match the instructions per clock (IPC) performance of AMD's Zen 3 cores. Achieving IPC performance comparable to AMD's Zen 3 would be a significant milestone for Loongson, as its current CPUs trail behind those from AMD and Intel. Matching the IPC performance of AMD's Zen 3 could potentially position these processors as viable competitors for AMD's 3rd Generation EPYC processors and Intel's Xeon CPUs.

However, achieving IPC parity alone does not ensure competitiveness for the 3A6000/3C6000/3D6000 processors against AMD's Ryzen 5000-series and 3rd Gen EPYC parts, even with similar core counts. Factors like clock speed and other platform features, like the memory subsystem, will be crucial in determining the overall performance. 

Meanwhile, Loongson is doing a lot to make its LA664 cores, in particular, and LoongArch-based CPUs, in general, more competitive. In addition to enabling SMT on its latest cores, Loongson's last year also introduced support for 128-bit vector processing extension instructions (LSX) and 256-bit advanced vector processing extension instructions (LASX) for their new CPUs. While LSX and LASX are part of the LoongArch microarchitecture found in the current 3A5000-series processors, it is blurred if they were previously activated and, if so, what impact they had on performance. 

Taping out of the 16-core LS 3C6000 processor is a big deal for Loongson as the company needs to offer higher-performing CPUs to compete against offerings from AMD and Intel on the Chinese market. Meanwhile, what remains to be seen is whether Loongson's 6000-series processors with 16, 32, and 64 cores can offer decent performance in the high-performance computing (HPC) space.

China's market for client and datacenter computers is vast and growing, but Chinese fabless chip designers that build processors for these applications are struggling and bleeding money, reports the South China Morning Post. Cambricon Technologies, a developer of artificial intelligence (AI) accelerators, and Loongson Technology, a CPU designer, are expected to lose tens of millions of dollars in 2023.

Loongson, one of the leading CPU developers in China, faces a drastic downturn with anticipated full-year revenue of ¥508 million ($71.524 million), a 31% year-over-year decrease, and an expected loss of ¥310 million ($43.36 million). This contrasts sharply with a profit of ¥51.8 million yuan in 2022. This decline is attributed to a sluggish market for industrial chips and significant expenditures on research and development. 

Cambricon Technologies, an AI chip specialist, has been in red for years and is also bracing for a tough year. It is projecting a full-year loss of up to ¥924 million ($129.3 million). Although this figure represents a 40% improvement over its losses in 2022, thanks to aggressive cost-cutting strategies, its losses still exceed its revenue. Cambricon's sales for 2023 are anticipated to be slightly down from ¥722 million ($101.05 million) in 2022. Cambricon did not elaborate on the reasons for its weak performance, though there are two obvious ones. Huawei, once its largest client, no longer buys its AI processors, and Nvidia was particularly strong in China last year with its AI GPUs. 

One of the biggest problems for Cambricon and Loongson is that they can address only a fraction of the market due to their narrow product lineups. Meanwhile, government procurements fail to provide the necessary scale to sustain these businesses amid international sanctions and market pressures. 

"It is not just them [Loongson and Cambricon], everyone struggled a bit last year," said Stewart Randall, head of electronics and embedded software at consultancy Intralink, in a conversation with the SCMP. "A lot of companies like Loongson and Cambricon, they do not really sell into the mass market, and that is where semiconductor companies make real money."  

Loongson and Cambricon are also hampered by their inclusion on the US Commerce Department's Entity List, which severely limits their ability to procure advanced American-origin technologies and address certain customers. The U.S.'s tightening of sanctions against China's semiconductor and supercomputer sectors has significantly affected these industries. These measures have disrupted the supply chain and stifled the progress of the People's Republic semiconductor development.

Hot on the heels of releasing surprisingly competitive new CPUs, Loongson also announced that it is developing a new GPGPU -- but its performance specs fall far below competing GPUs from Nvidia. 

With demand for artificial intelligence (AI) and high-performance computing (HPC) accelerators on the rise as US sanctions severely impact China's access to the fastest GPUs, multiple domestic companies are trying to address the market with specialized processors. Loongson, a prominent CPU developer from China, recently introduced its LG200 accelerator that can process AI, HPC, and graphics workloads, reports El Chapuzas Informático.

At a high level, the Loongson LG200 is a highly parallel processor akin to AI and HPC GPUs by AMD and Nvidia. Loongson's LG200 supports the OpenCL 3.0 application programming interface (API) for compute, which is good enough for high-performance workloads including AI and HPC. It's not enough to run an operating system, but support for OpenGL 4.0 for graphics workloads is good enough for some games.

The block diagram of Loongson's LG200 depicts a processor organized in four clusters, each featuring 16 small ALUs, four bigger ALUs, and one huge ALU or a special-purpose unit. Unfortunately, we cannot draw any conclusions from analyzing the diagram, as it is light on actual technical detail. 

Loongson has yet to disclose the specifications of its LG200 processor. We know it supports INT8 data format for AI workloads and probably FP32 and FP64 for graphics workloads, respectively. Also, Loongson claims that the LG200's compute performance is from 256 GFLOPS to 1 TFLOPS per node, though it didn't disclose the precision it used for the metric.

Even if the company used FP64 for its performance claims, the processor is dramatically slower than modern GPUs. For instance, Nvidia's H100 delivers 67 FP64 TFLOPS. That means this GPU could be targeted more at lower-power inference applications.

Loongson calls its LG200 accelerator a "GPGPU," which certainly implies that the part not only supports AI, HPC, and graphics workloads but can also perform general-purpose computing. But we can only wonder what the company means by general-purpose computing. We are unsure if the processor can run an operating system, though OpenCL can be used for certain general-purpose workloads.

Chinese chip designer Loongson has finally launched its loong teased "next-generation" 3A6000-series processors based on the LoongArch microarchitecture. IPC tests showed the 3A6000 matching Intel's Raptor Lake i5-14600K in IPC (instructions per clock), with both chips clocked at 2.5GHz.

As well as the headlining x86 compatible processor came the announcement of numerous partner desktop, laptop, and all-in-one machines — plus a consumer-grade motherboard from Asus. It was also entertaining to see a recorded overclocking session, which took an LN2-cooled 3A6000 chip to the current maximum 3 GHz.

Starting with the basics, the Loongson 3A6000 desktop processor is fabricated using a 14 / 12nm process and features a 4-core/8-thread configuration. It runs at clock speeds from 2.0 to 2.5 GHz, consuming up to 50W. According to the chip designer, this new CPU has 256KB of L2 cache and 16MB of L3 cache, and still only supports DDR4-3200 — like its 4C/4T predecessor, the Loongson 3A5000.

Loongson compared its new 3A6000 with Intel 10th Gen and AMD Zen 3 CPUs prior to today's official launch. It continues to make comparisons to the higher-clocked 4C/8T Intel Core i3-10100 even now. Note that the Core i3-10100F features on our list of the best budget CPUs, and it features the Comet Lake architecture that was a rehash of the Coffee Lake architecture, which itself was a refresh of the Kaby Lake and Sky Lake architectures.

Chinese TechTuber Uncle Tony took the Loongson 3A6000 desktop processor for a spin a few hours ago. In his BiliBili video, he tested a sample on the new Asus XC-LS3A6M motherboard. Apparently, the new Chinese CPU can be "easily" overclocked to 2.63 GHz on air. However, Uncle Tony pushed things further for fun, using liquid nitrogen to coax the CPU to 3.0 GHz. According to the linked ITHome report, 3.0 GHz isn’t going to be an impenetrable barrier for the 3A6000, and it says it thinks things will improve in the future and indicates that 3.0 GHz is a limit currently set in the Asus motherboard BIOS.

Regular readers probably won’t be too impressed at the performance of the out-of-the-box and overclocked Loongson 3A6000. Pre-launch we saw it was roughly on par with the Intel Core i3-10100 in SPEC CPU 2006 and UnixBench (at iso-clocks). This isn’t a trivial achievement, if Loongson is indeed achieving this with its own core IP and Dragon architecture, as it claims. Moreover, even Intel’s Core i5-14600K wasn’t significantly superior in SPEC CPU 2006, when restrained to 2.5 GHz. But IPC is only half the equation, and Intel's modern chips can clock about twice as high.

Uncle Tony confirmed a number of benchmark comparisons between the 3A6000 (and the previous gen 3A5000) vs the Intel Core i3-10100. He also highlighted some of the Asus motherboard features, and checked out a few games and apps, before starting his OC session. As mentioned above, the OC fun was muted due to a hard limit in the BIOS, but it's always quite theatrical to see plumes of LN2 vapor boiling off a CPU.

The source reports suggest that future desktop CPUs from Loongson are going to make their greatest strides by using process refinements. ITHome says China processor makers like Loonson will, in the future, “use mature processes to achieve the performance of Intel and AMD's advanced process CPUs.”

It would be fun to get a Loongson 3A6000 CPU in the lab, with a suitable motherboard. Now that it has finally launched, and with the plethora of partners announcing desktops, laptops, and AiOs, it's likely we will be able to buy some of this new Chinese hardware in the coming weeks or months — for science.

The Alt operating system developed by Moscow, Russia's Basalt SPO has been recompiled to support Chinese Loongson processors based on the LoongArch architecture. Blacklisted Loongson has actively supported the porting process. The OS is available as a distro with a basic set of programs (Firefox, GIMP, LibreOffice) that can be installed on desktops, workstations, and servers. The company says the product is offered 'as is,' and its commercial distributives will be available later.

Alt is now the first Russian operating system capable of running on Loongson's processors based on the 64-bit LoongArch architecture, such as LS5000 and LS6000 series, which some in Russia consider alternatives to x86 CPUs from AMD and Intel. Recently, China lifted the export ban on Loongson's latest CPUs to Russia, which opened doors to using these processors in desktops, laptops, servers, storage equipment, and other applications.  

The distro is based on the unstable, experimental branch of the Sisyphus project repository, which developers use to test new ideas and developments, including support for various processor architectures, reports CNews. The distro is available for download, but a stable release with LoongArch support is set for Q1 2024. 

"In early 2024, Basalt SPO will release a new stable branch of the Sisyphus project repository — the Eleventh platform," said Alexei Novodvorsky, an advisor to the CEO of Basalt SPO. "We plan to release Alt OS p11 distributions, including distributions for the LoongArch platform in Q1 2024."

The Alt OS is only the first step towards broader support of LoongArch-based CPUs in Russia, but it is undoubtedly essential.

"Basalt SPO is presenting a unique technology — the first Russian operating system for the Loongarch64 architecture," said Sergey Trandin, CEO of Basalt SPO. "Loongarch64-based processors have recently entered the Russian market, which means organizations and enterprises face a large-scale task of adapting software to work with the Chinese architecture. The Alt operating systems can become a platform for testing such products."

Basalt SPO developers adapted the Alt OS for full-fledged work with Loongarch64 in just nine months. This quick result was possible thanks to the automation of the assembly process using its unique process flow called catch-up package assembly. 

"The catch-up assembly technology significantly accelerates the process of adapting the OS for new architectures," said Trandin. "Originally, packages for Loongarch64 were assembled 'manually' — the assembly of the first thousand took about half a year. In July, an assembly line was set up for this platform, and the process proceeded much faster. Today, more than 16,000 software packages for the Chinese architecture have been compiled in the Sisyphus project repository, and their base is constantly replenished. The source codes of the Loongarch64 storage packages are also published and are developed as part of the Sisyphus project."

Software specialists from Loongson actively supported Basalt SPO by providing detailed documentation and modern Loongson-based servers for testing the OS.

Russia's adaptation of the Linux-based Alt to support the LoongArch architecture from China's Loongson company is a crucial development. On the one hand, by mating the Alt OS with Loongarch64 processors, Russia is moving towards greater technological independence from Western technologies and diversifying its use of processor technology. On the other hand, it introduces a dependency on Chinese technology.

MyDrivers has published a review of Loongson's 3A6000 quad-core CPU, confirming that the chip's IPC improvements are real. Benchmarks reveal that the 3A6000 enjoys an impressive 60% performance uplift in single-core performance and an even more impressive 2x performance multiplier in multi-core performance over its 3A5000 predecessor. With these improvements, the 3A6000 features performance comparable to a Core i3-10100F, with the IPC performance of a Zen 3 chip.

Of course, both Intel's Comet Lake 10th Gen architecture and AMD's Zen 3 architecture are now coming up on three years old. They're nowhere near the top of our list of the best CPUs for gaming or other purposes. But it still represents a step in the right direction.

Loongson's 3A6000 is the company's latest CPU featuring its home-brewed Dragon CPU architecture. The chip sports a quad-core design featuring multi-threading support, with a new 6-way multiple-issue design that makes the 3A6000 substantially more efficient than its predecessor. The chip's clock speed is rated at 2.5GHz. However, the CPU does not feature native x86 support due to political reasons. As a result, it features fully proprietary instructions in place of x86-supported instruction sets and utilizes virtualization and translation techniques to run x86 applications (when needed).

The 3A6000 was tested in SPEC CPU 2006 and UnixBench against Intel's Core i3-10100F and AMD's Ryzen 3 3100 entry-level CPUs. In Spec CPU 2006, the Loongson chip was 5% slower than the i3-10100F, and anywhere between 10-15% slower than the Ryzen 3 3100 in the single-core and multi-core tests.

UnixBench showed similar results for the 3A6000. The Chinese CPU was 2% slower than the i3-10100F and 8% slower compared to the Ryzen 3 3100 in the single-core test. In the multi-core benchmark, the 3A6000 was 8% slower than the i3-10100F and 11% slower than the Ryzen 3 3100.

These new benchmark reports confirm previous performance analyses by Loongsoon themselves stating that the A36000 chip would feature comparable performance to Intel's older 10th Gen Comet Lake CPUs. The 3A6000 doesn't beat Intel's Core i3-10100F counterpart, but it does get relatively close. At the same time, these tests represent a limited overview of performance and are generally synthetic in nature.

Longsoon's original IPC analysis has so far been proven to be legitimate. Even though the 3A6000 is only targeting Comet Lake performance, it's doing it at just 2.5GHz, which is substantially lower than the Core i3-10100F's 4.3 GHz peak turbo clock. If Longsoon can figure out how to boost clock speeds to 4GHz or higher, its CPU architecture might be able to compete with the likes of AMD's and Intel's more recent CPU architectures.

This will be important as Longsoon continues to try and compete with its Western competitors in the race for higher CPU performance. And with increasing export restrictions, the Chinese government will have plenty of reasons to continue supporting one of its most successful home grown alternatives.

China is speeding up its efforts to lessen reliance on foreign technologies, focusing on nurturing and implementing domestic alternatives, according to Reuters. This strategic shift has been influenced by the tightening restrictions from U.S. government on high-tech exports to China. The initiative has seen a surge in government, military, and state-linked entities pushing for an accelerated adoption of domestic PCs, telecommunication equipment, and software.

Hundreds of Billions in the Game

In 2022, China allocated ¥1.4 trillion (equivalent to $191 billion) towards the substitution of foreign hardware and software with domestic alternatives, representing a 16.2% increase compared to the previous year, as per insights from the IT research entity, First New Voice.

The number of payments and bids aimed at nationalizing equipment used by state-owned enterprises (SOEs) has seen a two-fold increase, rising from 119 to 235 from September 2022 to September 2023, Reuters found out from finance ministry database. The value of projects awarded during this period has also tripled, reaching a total of ¥156.9 million ($21.448 million). Although only capturing a fraction of the national tender bids, the database stands as the most extensive public repository of state tenders, furthermore, it corroborates with information from independent third-party sources (e.g., research documents), the news agency notes.

Replacing Hardware and Software

A significant portion of China's shift initiative is directed towards computer hardware, aiming to supplant foreign-branded machines and components with locally developed and manufactured ones. The telecommunications industry is also under scrutiny, with plans underway to transition towards domestic technologies, minimizing the reliance on Western companies. Additionally, there is an effort to revamp sensitive infrastructural systems, such as usage of domestic intelligence-gathering equipment. 

China is strategizing to overhaul various software sectors with homegrown solutions too. Office software systems used by state enterprises are earmarked for replacement with local versions, aligning with directives to enhance domestic utilization. Financial technologies, particularly those governing digital transactions and banking databases, are also slated for a revamp, fostering a transition towards domestic software to bolster security and autonomy. 

Huawei has emerged as a main player in China's technological transformation, witnessing a substantial growth in its enterprise business sector, which encompasses software and cloud computing services. In 2022, Huawei's enterprise business reported sales amounting to ¥133 billion ($18.181 billion), marking a 30% increase from the previous year. The company's comprehensive range of products and its agility in product deployment are the key factors contributing to its thrive in China despite being sanctioned by the U.S. government.

China Faces Challenges

Despite the vigorous push towards domestic technology, China faces significant challenges, primarily attributed to its limited capabilities in advanced chip manufacturing. This limitation hinders the country's ability to fully realize a comprehensive substitution of foreign technologies with domestic counterparts. Additionally, sectors such as financial institutions have exhibited reluctance in transitioning towards domestic database systems, citing stability and reliability concerns.

In late 2022, the Chinese government banned exports of Loongson's latest LS5000-series CPUs to Russia, citing national security concerns. But in March 2023, the U.S. blacklisted Loongson and limited its access to American technologies, which is perhaps why the Chinese government decided to lift the ban on exports of these CPUs. As a result, at least two Russian companies are planning to produce Loongson-based systems, reports Kommersant.

Norsi-Trans and Promobit are leading the initiative, aiming to produce servers, storage systems, and computers powered by Loongson's 5000-series processors using the LoongArch architecture. So far, Norsi-Trans has purchased around 100 Loongson processors for the initial production phase. They have also received the green light for incorporation into the official registry of Russian electronics.

These Russian tech companies eye the Loongson processors as viable alternatives to their American counterparts from AMD and Intel. However, there is a slight catch: software support. Norsi-Trans will collaborate with Basalt SPO, an Alt operating system developer, to adapt the OS to work seamlessly with the Loongson processors. Meanwhile, it is unclear when Basalt SPO will tailor its OS for Loongson's CPUs.

The decisions by the Russian companies to opt for Loongson processors were made with strategic considerations. The Chinese government previously restricted these processors from export due to their critical technological significance, including applications in China's military-industrial complex. The loosening of these restrictions and the resurgence of Loongson processors in international collaborations coincide with rising geopolitical tensions. Essentially, Norsi-Trans and Prombit want to build a more self-sufficient tech ecosystem, less vulnerable to foreign sanctions and supply chain disruptions. 

"Because the X86 and Arm architectures are controlled by the American and British governments, there is always a threat of disconnection or incorrect processor operation," said Grigory Sizonenk, chief executive of IVK, in a conversation with Kommersant. "Therefore, Loongson chips look more attractive. 

However, despite these collaborative advancements, industry experts express caution. While mitigating reliance on American technologies, the move might introduce a new dependency on Chinese tech. Voices within the industry, such as the president of Russoft, Valentin Makarov, underscore the importance of negotiating mutual market and technology access with Chinese partners, ensuring a balanced technological exchange and collaboration.

 "We will just replace dependence on American processors with the same dependence on Chinese ones," Makarov said. "Therefore, it is necessary to negotiate with Chinese partners for mirror access of Russian IT products to the Chinese market or access to Chinese technologies."

Chinese chipmaker Loongson has long promised that its next-generation 3A6000-series processors based on the LoongArch microarchitecture will match AMD's Zen 3 CPUs in terms of instructions per clock (IPC) performance. This week the company disclosed (via @9550pro) some of the actual test results of its quad-core 3A6000 processor and said that they were comparable to Intel's quad-core '10th-Gen Core' processor from 2020. But there's a catch.

"Based on the relevant test results, the overall performance of the Loongson 3A6000 processor is comparable to that of Intel's 10th generation Core quad-core processor launched in 2020," a statement by Loongson reads.

Loongson's quad-core 3A6000 processor has a 2.50 GHz clock rate. The chip hit 43.1/54.6 points in the SPECint_base2006/SPECfp_base2006 benchmarks, as well as 155/140 in SPECint_rate_base2006 (8 copies)/SPECfp_rate_base2006 (8 copies), at least according to the tests conducted by Saixi Laboratory of the China Institute of Electronic Technology Standardization.

However, SPEC discontinued its SPEC 2006 CPU benchmark in 2018, making it impossible to compare Loonson's performance numbers to independently obtained results approved by SPEC. Meanwhile, the new 3A6000 is noticeably faster than its predecessor, based on previously published test results of the quad-core 3A5000 at 2.50 GHz.

The same institute also tested the CPU in UnixBench v5.1.3 and got a 2284.5 single-thread score and 7438.4 eight-thread score. If Intel Core i7-10750H's UnixBench performance numbers published in a Reddit post are to be believed (1713 single-thread, 8248 12-thread), then the 3A6000 indeed outpaces Intel's six-core Comet Lake at 2.60 GHz when it comes to single-thread workloads at around the same clock, but falls behind a six-core CPU with SMT enabled. It's noteworthy that Intel's Comet Lake uses Skylake cores from 2015.

According to Loongson, its 3A6000-series processors employ a brand-new 6-way multiple-issue Dragon microarchitecture that is significantly more efficient than its predecessor. Meanwhile, the company has so far published the results of its 3A6000 CPU at 2.50 GHz and has not disclosed the final clocks of the actual processors that are due to ship several months from now.

To that end, it is too early to draw any conclusions about the performance of Loongson's upcoming processors. On the one hand, it looks like the 3A6000 is faster than the 3A5000 at the same clock in the discontinued SPEC CPU 2006 benchmark, but that's based on results published by the company itself. The chip also seems to be faster than its predecessor and Intel's Skylake in UnixBench at 2.50 – 2.60 GHz. 

Since there are no independently obtained benchmark results of the 3A6000, we cannot really say whether or not Loongson has succeeded in developing a microarchitecture that matches AMD's Zen 3 in terms of IPC performance or not.

Chinese chipmaker Loongson's next-generation 3A6000-series CPUs will feature simultaneous multithreading support, the latest Linux 6.5 patches that the company submitted (via Phoronix). The addition of SMT will double the number of threads these chips will process simultaneously.

Loongson's SMT implementation in its next-generation 3A6000-series processors is similar to AMD's and Intel's and enables one physical CPU core to process two threads at the same time. Therefore, the company's quad-core 3A6000 processor for client PCs will be able to process eight threads simultaneously. Meanwhile, Loongson's datacenter grade 32-core 3D6000 CPU will be able to process 64 threads at once. 

"Loongson-3A6000 has SMT (Simultaneous Multi-Threading) support, each physical core has two logical cores (threads)," Loongson's description of the technology reads. "This patch adds SMT probe and scheduler support via ACPI PPTT."

In addition to enabling SMT, Loongson's patch also enables support for 128-bit vector processing extension instructions (LSX) and 256-bit advanced vector processing extension instructions (LASX) for the new CPUs. While both LSX and LASX are parts of the LoongArch microarchitecture that powers existing 3A5000-series processors, it is unclear whether they were ever enabled and if they were, what performance benefits did they bring. 

Loongson anticipates that its forthcoming LoongArch 6000-based CPUs will equal AMD's Zen 3 in terms of instructions per clock (IPC), potentially positioning Loongson as a competitor against top-tier processor manufacturers. Matching IPC performance AMD's Zen 3 is a big achievement for Loongson as its existing CPUs lag behind those from AMD and Intel. 

But IPC alone does not guarantee success and make 3A6000/3C6000/3D6000 processors competitive against respective AMD Ryzen 5000-series and AMD 3rd Gen EPYC parts even with similar core counts. Factors such as clock speed and other platform features like the memory subsystem will significantly influence the final performance. We'll have to wait to see the chips when they emerge in the wild to suss out those details. 

Loongson has posted the first Linux patches to enable support for its next generation 3A6000-series processors, reports Phoronix. The company expects its upcoming LoongArch-based CPUs AMD's Zen 3 in instruction per clock (IPC), which will enable Loongson to challenge leading processor manufacturers.

Loongson shared details about the progress of its 3A6000-series CPU development last November when it revealed that the design phase of the project had been concluded and that samples of the processors would be available in the first half of 2023.

Now, the company's engineers posted patches enabling the 3A6000's new memory management unit (MMU) or page table walker (PTW) that can handle address translation exceptions (like TLBI, TLBL, TLBS, TLBM) directly in the hardware, boosting performance. The CPU will only require software handling in situations like page faults.

Another feature enabled by another patch for Loongson's 3A6000's processors is moving away from full completion barrier (dbar 0) hint to a set of more fine-tuned hints for different memory barriers, which can improve performance.

CPU enablement in Linux is an important milestone for any processor development cycle, since it signals that development is proceeding. Enablement alone does not necessarily mean that the new chip is about to be taped out, or is progressing rapidly, but at least it means that its designers are confident enough about its success.

Matching IPC performance of AMD's Zen 3 microarchitecture or Intel's Tiger Lake microarchitecture is a big deal for Loongson, whose current CPUs are considerably slower than processors from the leading suppliers. Meanwhile, it should be noted that IPC alone does not necessarily mean that Loongson's 2nd Generation CPUs that rely on its LoongArch microarchitecture will be as fast as AMD's Ryzen 5000-series or Intel's 11th generation Core processors. Clock speed and other aspects of the platform will also play a role.

Loongson, a Chinese fabless chipmaker, has launched the new 3D5000 processor for data centers and cloud computing. MyDrivers reported that Loongson claims its 32-core domestic chips deliver 4X higher performance than rival Arm processors.

The 3D5000 still leverages LoongArch, Loongson's homemade instruction set architecture (ISA) from 2020. The chipmaker was previously a firm believer in MIPS. However, Loongson eventually built LoongArch from the ground up with the sole objective of not relying on foreign technology to develop its processors. LoongArch is a RISC (reduced instruction set computer) ISA, similar to MIPS or RISC-V.

The 3D5000 arrives with 32 LA464 cores running at 2 GHz. The 32-core processor has 64MB of L3 cache, supports eight-channel DDR4-3200 ECC memory, and up to five HyperTransport (HT) 3.0 interfaces. It also supports dynamic frequency and voltage adjustments. Officially, the 3D5000 has a 300W TDP; however, Loongson stated that the conventional power consumption is around 150W. That's roughly 5W per core.

The 3D5000 flaunts a chiplet design since Loongson has glued together two 16-core 3C5000 processors. Loongson developed the 3C5000 server part to compete with AMD's Zen and Zen+ architectures. The latest 3D5000, which measures 75.4 x 58.5 x 7.1mm,  slides into a custom LGA4129 socket. 

The processor supports 2P and 4P configurations; therefore, Loongson has launched the 7A2000 bridge chip to manage the communication between the processors and other components. As per the chip designer, the 7A2000 is up to 400% faster than the previous generation. Furthermore, with the help of the 7A2000, there's a possibility to scale up to 128 cores per motherboard.

According to Loongson's provided numbers, the 3D5000 scores over 425 points in SPEC CPU 2006, a depreciated benchmark replaced with the newer SPEC CPU 2017 version. The 3D5000 also delivers over 1 TFLOPs of FP64 performance, up to 4X higher than regular Arm cores. Meanwhile, the processor's stream performance with eight channels of DDR4-3200 memory crosses the 50GB mark.

While performance isn't the 3D5000's strong suit, security is. The 32-core processor allegedly has a custom-made mechanism to defend against vulnerabilities such as Meltdown or Spectre. The chip also has its Trusted Platform Module (TPM), so it doesn't rely on an external solution. In addition, according to MyDrivers' report, the 3D5000 also supports a secret national algorithm with an embedded security module that seemingly delivers excellent encryption and decryption efficiency higher than 5 Gbps.

In addition to the 3D5000 and 7A2000, Loongson also announced the 2K050, the company's baseboard management controller (BMC). The 2K050 features LA264 cores at 500 MHz, integrated 2D GDP, 32-bit DDR3 support, and outputs at a 1080p (1920x1080) resolution at 60 Hz.

Loongson's 3D5000 is no match for AMD's EPYC Genoa or Intel's Sapphire Rapids Xeon processors. It was never about beating the foreign competition but pushing for self-sufficiency. Unfortunately, with the ongoing U.S. sanctions, Chinese companies have no means to secure chipmaking tools originating from the U.S. In addition, the U.S. Department of Commerce recently blacklisted Loongson, which likely derailed some of the company's plans.

Loongson, one of the few Chinese CPU developers that can make competitive processors for client and server applications, is facing massive tailwinds due to its recent inclusion on the U.S. Department of Commerce’s Entity list. It’s spot on that list, which restricts its access to technologies from the U.S., won’t curtail its ability to make chips overnight, but a DigiTimes report suggests it could cause major issues in the coming years.

Unlike some other Chinese companies that use Arm and x86 instruction set architectures controlled by Western companies, Loongson's CPUs rely on the company's proprietary LoongArch ISA, which is backwards compatible with the MIPS architecture. As a result, it is impossible for the U.S. government to cut Loongson's access to the latest CPU technologies. But Loongson uses American electronic design automation (EDA) software to develop its processors, whereas its manufacturing partner SMIC uses wafer fab equipment that originates in the U.S. 

To sell or support EDA programs to Loongson, companies like Ansys, Cadence and Synopsys — which control over 90% of the Chinese EDA market — have to obtain an export license from the U.S. Department of Commerce. Furthermore, to make chips for Loongson using its 14 nm and 12 nm-class process technologies, its manufacturing partner SMIC has to obtain other export licenses from the U.S. DoC as well.

Earlier this year it turned out that the Chinese government restricted exports of Loongson's latest CPUs based on the LoongArch microarchitecture citing strategic importance of the technology and national security concerns. 

Given historical use and positioning of Loongson's processors today, it is unlikely that the U.S. Department of Commerce will grant export licenses for U.S. originating technologies to be sold to Loongson. As a result, eventually Loongson will not be able to develop new CPUs using EDA tools it has today. Furthermore, SMIC and other chipmakers will not be able to legally produce processors for Loongson unless they manage to obtain an appropriate license. 

While Loongson's proprietary LoongArch microarchitecture can ensure that its processors can evolve, the curbs imposed by the U.S. government restrict development and manufacturing of actual CPUs.  

For now, the company has its quad-core 3A5000 processor for client PCs and 16-core 3C5000 as well as 32-core 3D5000 CPUs based on the for servers, so it is going to keep ramping up production of these chips while it can. Design of the company's 6000-series processors based on the next-generation LA664 cores (which promise performance comparable to that of AMD's Zen 3) is probably ready, but Loongson's access to advanced process technologies is uncertain. It remains to be seen whether the company will be able to commercialize even its next-generation CPU family.

This week, the U.S. government blacklisted CPU designer Loongson and server maker Inspur, alleging that these companies support the modernization of the Chinese People's Liberation Army. The addition to the Entity List limits Inspur's and Loongson's access to technologies originating from the U.S., which essentially curbs their existing business and ability to develop new products.

The U.S. Department of Commerce added Inspur and Loongson to the Entity List after it found that the entities acquired 'U.S.-origin items in support of China's military modernization efforts.' The U.S. DoC does not determine which items were procured but could include U.S.-sourced goods, software, services, and technologies. 

From now on, U.S.-based producers of hardware, software, and services and foreign companies that use American technologies (e.g., contract makers of chips that use tools made in the U.S.) must obtain an appropriate export license from the U.S. Department of Commerce. License applications are set to be reviewed with a presumption of denial.

No More Chip Production for Loongson?

Loongson is a major CPU developer from China, which has for years used the MIPS instruction set architecture (ISA) originally designed in the U.S. Loongson's latest quad-core 3A5000, 16-core 3C5000, and 32-core 3D5000 processors based on the LA464 cores use the company's own LoongArch ISA, which reportedly features nearly 2,000 proprietary instructions. As a result, these CPUs can take advantage of multiple architectural innovations — such as binary conversion extension instructions (LBT), vector processing extension instructions (LSX), and advanced vector processing extension instructions (LASX) — yet they maintain backward compatibility with MIPS. In addition, they can also execute code written for previous-generation Loongson processors. 

Virtually all chips designed in China are developed using electronic design automation (EDA) tools of U.S. origin. Furthermore, they are made — either by China-based SMIC or Taiwan-based TSMC or UMC — using tools produced in the USA.  

Now that Loongson has joined the Entity List, all of its EDA and chip fabrication partners (SMIC) will have to apply for an export license to the U.S. DoC to continue working with the CPU designer. SMIC, which has been under heavy attack from the U.S. government for years, will unlikely to get one, so formally it will not be able to service Loongson, which uses SMIC's 12nm-class manufacturing technology for its latest processors. 

No More Hardware for Inspur?

Backlisting Inspur is a big deal. Inspur is a huge IT conglomerate focused on AI, big data, cloud computing, servers, and storage. It also happens to be the world's third-largest maker of servers, with a 10% market share. Inspur does not develop its own silicon, but relies on CPUs, GPUs, and other chips designed by industry leaders, such as Intel and Nvidia.

Inspur already cannot procure advanced CPUs and GPUs used to build powerful supercomputers after the Biden administration imposed curbs against the Chinese supercomputer sector in October 2021 without an appropriate export license. But now, the company's suppliers will have to obtain export licenses for virtually all components and software they sell to Inspur, which includes memory modules, storage devices (HDDs, SSDs), network controllers, NVMe/SAS/RAID controllers, software, and perhaps even power supplies. 

While we would expect the U.S. DoC to grant export licenses to many Inspur suppliers, it might not be enough to support the company's needs. This will hurt many Inspur suppliers, which include familiar names like Intel, Nvidia, Broadcom, Microsemi, Samsung, Solidigm, and SK Hynix, to name a few.

35 More Entities Blacklisted

"The ERC determined to add 4Paradigm Technology, Inspur Group Co., Loongson Technology, National Research Center for Parallel Computer Engineering and Technology, Qingdao National Laboratory of Marine Science and Technology, Wuxi Institute of Advanced Technology to the Entity List for acquiring and attempting to acquire U.S.-origin items in support of the China’s military modernization efforts. This activity is contrary to U.S. national security and foreign policy interests," a statement of the Bureau of Industry and Security of the U.S. Department of Commerce reads. "All of these entities will require a license for items subject to the EAR, which will be reviewed under a presumption of denial. They are also given a footnote 4 designation, which means that 'items subject to the EAR' for the purpose of these license requirements include foreign-produced items that are subject to the Export Administration Regulations." 

In addition to Inspur and Loongson, the BIS of the U.S. DoC added 35 more entities from Belarus (1), Burma (3), China (26), Pakistan (4), Russia (1), and Taiwan () into the Entity List.

When we did our The Rise of China GPU Makers story inspired by Jon Peddie Research and then detailed by Jon Peddie himself, we were somewhat surprised that Loongson — a Chinese CPU champion — was not a part of the game. Well, now it looks like things have changed.

One of the announcements that flew under our radar this month was the introduction of the Loongson LS2K2000 system-on-chip (SoC) that features the company's proprietary LoongArch microarchitecture as well as its first own-designed graphics processing unit (thanks to Heise.de for the tip).

The LS2K2000 CPU is hardly anything special, as it is aimed at various embedded applications. As such, it only has two LA364 cores featuring the proprietary LoongArch microarchitecture with all of its peculiarities as well as some basic I/O capabilities like a 64-bit DDR4-2400 memory interface, PCIe 3.0, SATA 3.0, USB, Ethernet, audio, and everything else that one comes to expect from a simplistic SoC.

What grabbed our attention was the phrase that the LS2K2000 features the LG120 GPU 'core independently developed by Loongson, further optimizing the graphics algorithm and performance.'

Given the applications that the LS2K2000 is aimed at, it is unlikely that we are dealing with something very advanced in terms of performance. In fact, we believe it would be pretty much underwhelming when it comes to power consumption. Yet, a basic GPU from a CPU-centric company looks curious, and these types of foundations can be built upon.

When it comes to the LG120 GPU itself, it seems to be a very basic unit, the company would probably mention its capabilities if there was something to brag about. What seems important is that Loongson claims that this is a unit designed internally.

We do know that many Chinese companies license Arm and Imagination Technologies GPU IP to integrate into their designs in a bid to save time and money (read our China GPU story for details). Meanwhile, Loongson has publicly said that it was adopting the RISC-V architecture for its next projects. While it was disclosed in the context of CPUs, the architecture is flexible enough for GPUs, too. So, who knows?

Of course, it is possible that Loongson only has a GPU for very basic needs. Yet, the company that has ambitions to challenge AMD's Zen 3 on the CPU side of the business might well have GPU ambitions, too. If it were designed for scalability, a proprietary GPU architecture would be a good start, even if it begins on the low end.

Building large monolithic many-core CPUs is extremely hard even for renowned chip designers. For Chinese CPU developers, many of whom do not have access to leading-edge production nodes, the only way to build a processor with high core count is to adopt a chiplet design. As it turns out, this is exactly what Chinese CPU maker Loongson does with its 32-core 3D5000 processor, Sina reports.  

Earlier this year Loongson began to ship its 3C5000 processor that relies on 16 LA464 cores featuring the company's LoongArch microarchitecture, up to 64MB of cache, and four 64-bit DDR4-3200 memory interfaces with ECC support. The Loongson 3D5000 takes two 3C5000 CPUs and places it on a single piece of substrate to build a 32-core processor with eight memory channels. The 32-core processor supports up to 4-way simultaneous multiprocessor configurations and therefore it is possible to build a server with up to 128 cores. 

Loongson recently completed verification of its 3D5000 processor, the report says. The CPU reportedly consumes 130W at 2.0 GHz as well as 170W at 2.20 GHz. Loongson's 3D5000 CPU comes in an LGA-4129 packaging. 

While building a 32-core CPU based on a proprietary microarchitecture is an achievement, it should be noted that product is a way to test ability to build a chiplet-based design. While China-based SMIC — which produces processors for Loongson — slowly adopts more advanced nodes, it is significantly behind market leader TSMC. Therefore, companies like Loongson cannot offer products that are comparable to those from AMD and Intel.  

Chiplets represent a real opportunity for Loongson to build rather serious processors and servers platforms with significant number of cores and proprietary microarchitecture enhanced for servers and supercomputers. Meanwhile, we have no idea whether other CPU developers from China will follow the suit.  

When it comes to performance, Loongson says that its 32-core 3D5000 CPU scores 400 points in the SPEC CPU2006 base test, whereas the result of a 2-way 32-core 3D5000-based machine exceeded 800 points in the SPEC CPU2006 base benchmark. The CPU designer believes that a 4-way machine will hit 1600 points. 

Loongson is gearing up to ship samples of its 32-core processors in the first half of 2023, whereas commercial versions will be shipped later.

The Chinese government has reportedly banned exports of Loongson CPUs based on the LoongArch microarchitecture to Russia and other countries, citing the strategic importance of these processors that are used by the country's military. For some Russian companies, Loongson chips could have become an alternative to x86 processors from AMD and Intel if partners of these two companies cease to ship these CPUs to Russia via other countries.

No Loongson for Non-Chinese Buyers?

Sources close to the Russian Ministry of Digital Development, Communications and Mass Media as well sources with knowledge of the local high-tech industry told Kommersant business daily that the Chinese government has banned exports of Loongson processors featuring the LoongArch microarchitecture to other countries. This is not going to have an immediate impact on the local market, but it could leave Russia without alternatives to processors from AMD and Intel.

"Although Russian companies do not depend on supply of Chinese processors significantly, in case of a hypothetical blocking of 'parallel imports,' they hoped to switch to Loongson solutions," a source explained to the newspaper. "The best [Loongson CPUs] are used by the Chinese military-industry complex, this is the main reason why they are not available to foreign markets."

Kommersant admits that the Chinese government and Loongson have not yet formally put restrictions on exports of LoongArch-based CPUs. At present, it is still possible to get Loongson LS3A5000-based systems and motherboards from AliExpress, but these parts are rather expensive. Given that the performance of Loongson CPUs is significantly lower compared to processors from AMD and Intel, it's hard to expect these products will get any traction anywhere except China.

LoongArch for China

For quite a while, several Russian companies have been exploring alternatives to x86 processors and compute GPUs from American developers. After Russia invaded Ukraine in late February, 2021, AMD, Intel, and Nvidia ceased operations in the country and stopped selling CPUs and GPUs to Russian companies. While grey imports of CPUs and GPUs now thrive in Russia, it is possible that at some point they will be stopped or significantly reduced and the country will be left without essential technology.

That's when Chinese 'alternatives' will become the only options for at least some of the Russian PC and server makers. However, it looks like the Chinese government does not want its Loongson chips to become a substitute AMD and Intel offerings — in Russia or anywhere else.

"For now, the decision to ban exports of Loongson CPUs to Russia will not cause much damage, as big projects to deploy Loongson in Russia most likely have not yet begun," said Maxim Koposov, chief executive of Prombit, a Russian PC maker which has tried to use non-x86 processors in its systems.

Loongson has two CPU models based on its LoongArch microarchitecture: the quad-core LS 3A5000-series processors for client PCs as well as 16-core LS 3C5000-series chips for servers. These CPUs use LA464 cores that can execute code written for Loongson's previous-generation processors featuring the LoongISA microarchitecture, which is a custom subset of the MIPS64 architecture. Meanwhile, the latest cores also feature 2,000 LoongArch-specific instructions, binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ). 

It's possible that the Chinese government does not want other countries to learn how to take advantage of LoongArch. Another reason why the Chinese government might not want to export Loongson CPUs is because they are made by SMIC on its 12nm-class fabrication process. Assuming that SMIC's 12nm/14nm capacity is limited, China may simply want to keep all the processors it can get for its own projects. While some Loongson chips may ship from AliExpress, there will be no official sales of these CPUs to customers in Russia or other countries.

Chinese chipmaker Loongson Technology has held a performance briefing on its upcoming fourth-generation 3A6000 CPU, which has completed its design phase and is now entering production. The first samples of the chip, which is expected to approach Ryzen 5000 levels of performance according to Loongson, should be available in the first half of 2023.

Ryzen 5000, let’s not forget, is Zen 3 technology, roughly on a par with Intel’s 11th generation chips from 2020. Loongson’s LoongArch processor architecture, however, isn’t compatible with X86 code, instead mashing together bits of MIPS and RISC with custom instructions that may help with the emulation of other systems. Introduced in 2021, it primarily uses Linux for an operating system.

The performance claim comes from Loongson’s internal simulation testing, using the SPEC CPU 2006 benchmark that was retired by its maker in 2019. The 3A6000 chip offers a 68% improvement in single-core floating-point performance over its predecessor, the 3A5000, which if correct is quite impressive and matches claims made back in June. The chips will be built on a 12nm process (Zen 3 was a mix of 7nm CCD chiplets with a 14nm IOD chiplet), but it’s not clear who will be making the chips as Loongson Technology is a fabless company.

Software support for the chips is also in the works, with Loongson Technology chairman Hu Weiwu, speaking at the 2022 Information Technology Independent Innovation Summit Forum in Nanjing, committing to the creation of Linux-based desktops and applications. China is attempting to rid its computing ecosystem of foreign technology, just as the United States has introduced export controls to prevent the country from getting hold of advanced smarts.

While it’s hoped that the 3A6000 and follow-up 3A7000 will become mainstream products, Weiwu also launched the Loongson 100-Core Project at the forum. Current Loongson server chips have 16 cores, with 64-core products not expected until 2015, so 100 cores may still be some time in the future.

The 486 CPU is somewhat of a relic these days, but its legacy in the Linux kernel has lived on. The i486 has been the de facto minimum for decades.  Even Linux, that long-term supporter of outdated architectures, is considering giving up on the chip and removing support for the 486 processors, just like it did for the 386 back in 2012.

The news comes via a post on the Linux Kernel Mailing List from Linus Torvalds himself. Recently keen on adding things like the Rust programming language and support for Intel Arc GPUs and Loongson CPUs to the Linux kernel, Torvalds is now considering removing the venerable 486, writing: “We got rid of i386 support back in 2012. Maybe it's time to get rid of i486 support in 2022?”

The idea, which seems so obvious in these days of Raptor Lake and Ryzen 7000, received a certain amount of pushback, with the claim from some users that new hardware based on the superannuated silicon was still being shipped. When the same plan was raised a year ago, one user said they were still using a 486, and wanted to continue doing so.

The 486, which dates back to 1989, is currently the minimum possible spec for running Linux, and works best with lightweight distros such as Tiny Core Linux.

It all comes down to cmpxchg8b, an instruction that compares then exchanges eight bytes (or 64 bits) of information in the computer’s memory. Mailing list member Peter Zijlstra suggested Linux should only support processors capable of carrying this out, leaving the 32-bit 486 behind and meaning new Linux kernels would run on P5-class hardware or newer. The cmpxchg8b instruction is the culprit behind the original Pentium’s ‘F00F’ bug, which saw an affected CPU without operating system mitigations in place cease to function until it was rebooted when asked to execute the instruction.

Kernel 6.1, out this year, will be a Long Term Support kernel, but with Torvalds’ backing to the idea it could be the last one that will run on the 486, with 6.2 removing the ability to run Linux on ancient beige 486 towers.

Owners of Raptor Lake and Arc Alchemist rigs desperate to escape the Microsoft hegemony and explore the world of open-source software awoke to good news today, with the announcement that version 6.0 of the Linux kernel will directly support their cutting-edge platforms. 

It’s also good news for owners of Qualcomm Snapdragon 8cx Gen3-powered laptops, plus other Arm-based SoCs and Chromebooks, including the Allwinner H616, found in TV streaming boxes and the NXP i.MX93 embedded processor board. Raspberry Pi 4 users also get a v3d graphics driver, updated drivers for AMD’s RDNA 3 GPUs, RISC-V platforms get additional extensions, and the H.265/HEVC codec has moved to 'stable' status. Rockchip RK3588 devices, such as the Pine64 QuartzPro64 (pictured above), also receive an MMC driver for their onboard storage.

Bigger news than that, however, is the improved support for the Chinese LoongArch CPU architecture, which received tentative backing in the previous 5.19 kernel, but was unable to boot thanks to missing drivers. That’s all fixed in kernel 6.0, opening up the OS family to a whole new set of computers. It also improves the amount of hardware that can be used in Linux on LoongArch, with initial support for PCI on the platform. This means Linux could soon run on the recently announced Loongson 16-core server chips.

The jump from 5.19, the kernel used in the new Ubuntu 22.10 beta, to 6.0 comes from a quirk of Linux overlord Linus Torvalds, who doesn’t like to see version numbers get too big. “The major version number change is more about me running out of fingers and toes than it is about any big fundamental changes,” he wrote in a Linux Kernel Mailing List message, adding: “Of course there's a lot of various changes in 6.0 - we've got over 15k non-merge commits in there in total.”

Torvalds did, however, hint at big things to come in the future: “Tomorrow I'll open the merge window for 6.1. Unlike 6.0, which has a number of fairly core new things lined up.”

The new kernel should be available to install in major Linux distros with a simple update as soon as the package maintainers push it through. Of course, users of smaller or more exotic flavors of the OS may have to take matters into their own hands, but that’s the joy of Linux.

Loongson has become yet another Chinese GPU designer as it rolled out a new chipset with built-in graphics capabilities for its quad-core 3A5000 and 16-core 3C5000 processors. The new iGPU has capabilities comparable to graphics chips from 2004 ~ 2005, but it offers sufficient performance in office workloads. 

Loongson's 7A2000 bridge chip (aka chipset) integrates an OpenGL 2.1/OpenGL ES 2.0-compliant GPU that works at 400 MHz ~ 500 MHz, supporting up to two displays with a 1920x1080@60Hz or 2560x1440x30Hz resolution over two HDMI connections (or an HDMI and a D-Sub/VGA connection).  

Based on data from Loongson, the GPU scores 300 FPS in GLMark2, a benchmark from 2004 ~ 2005, and 1,800 FPS in Glxgears, a performance testing program from 1999 ~ 2000. Unfortunately, the manufacturer does not disclose which operating system it used or how it compiled the benchmark. In addition, it is noteworthy that the GPU does not support even basic DirectX features and, therefore, can't even Windows 7. 

The iGPU is said to be an in-house design, and we are not surprised by this since both Arm and Imagination have inexpensive GPU IP blocks that offer a much more sophisticated feature set. However, developing a low-end in-house GPU may make a lot of sense for Loongson. 

The 7A2000 chipset is designed for Loongson's 3-series processors that use proprietary LoongArch architecture and therefore require an operating system and software that support this architecture. Instead of rewriting third-party drivers for Loongson's platform, designing a low-end GPU and developing drivers in-house may be easier. At the end of the day, Loongson's customers are various government or local authorities-run organizations that see China-developed IP as a huge benefit even if it is years behind equivalents designed in Europe or the U.S.  

Other features of the 7A2000 core logic include PCIe 3.0, SATA 6 Gbps, USB 3.0, GbE PHY, UART, and even GPIO if someone wants to connect something fancy to a Loongson-based platform.  

Another notable improvement of the 7A2000 mentioned by Loongson is the fact that the CPU and chipset are now connected using a HyperTransport 3.1 32-bit interface at 3.2 GT/s, which improved PCIe bandwidth by 2.4 times as well as SATA read/write performance by 82%/97% when compared to the previous generation Loongson 7A1000 chipset. 

Loongson says that multiple motherboard makers and ODMs have already developed numerous boards featuring a quad-core 3A5000 processor with a 7A2000 chipset and a 16-core 3C5000/3C5000L CPU with a 7A2000 chipset.

Having launched its first 16-core CPU for servers earlier this week, Loongson is looking forward to releasing a 32-core processor in the coming months and matching the performance of AMD's Zen 3-based products in the coming years.

Chinese CPU developer Loongson formally introduced and started to ship its first 16-core 3C5000 processor based on its proprietary LoongArch instruction set architecture earlier this week. At a conference where the CPU was officially unveiled, the company also talked about its future endeavors, and they look pretty ambitious. In the coming months, the company will reveal its 3D5000 CPU with 32 cores, which will combine two 3C5000 processors on a single piece of substrate. In contrast, its next-generation 6000-series processors may match the instruction per cycle (IPC) performance of AMD's Zen 3 microarchitecture, reports EET-China.

Loongson's current-generation 5000-series (quad-core 3A5000 and 16-core 3C5000/3C5000L) processors are on its LoongArch-based 64-bit superscalar LA464 cores. They are compatible with the MIPS ISA (to maintain compatibility with applications written for Loongson's previous-generation processors) but can also execute code explicitly written for LoongArch. With four general-purpose ALUs, and two 256-bit vector operations units, the LA464 cores look promising. Still, once the software is recompiled to take advantage of 2,000 proprietary LoongArch instructions, they promise to shine.

Based on the numbers demonstrated by Loongson at its conference, Loongson claims the quad-core 3A5000 processor operating at 2.50 GHz can challenge popular eight-core Armv8 implementations as well as Intel's 10th Generation Core 'Comet Lake' processors in terms of single-thread performance in UnixBench. However, the chip falls behind in multi-thread execution speed for obvious reasons. Meanwhile, Loongson did not publish the performance of its 16-core 3C5000 CPU in a similar scenario but only mentioned that the processor operates at 2.0–2.2 GHz and delivers up to 560 GFLOPS of raw performance at 130W.

For its next-generation Loongson 6000-series processors, the company plans to use its upcoming LA664 core design to increase efficiency and boost IPC throughput, the company said at its event, citing the results of its simulations. As a result, Loongson's 16-core 3C6000 and 32-core 3D6000 processors due in 2023~2024 will supposedly be able to challenge AMD's Ryzen and EPYC CPUs based on the Zen 3 microarchitecture in terms of performance per cycle. (Yes, that's an architecture from 2020, but the company has to start somewhere.)

For Loongson, matching AMD Zen 3's IPC in 2023 is a big deal since the company will be somewhat able to close the gap with modern chips from leading developers. Still, matching instructions per cycle alone may not be enough to reach the actual performance of AMD's CPUs. AMD's Zen 3 chips can clock in the 4.5–4.7 GHz range, which is significantly higher than the 2.5 GHz clocks Loongson has mentioned.

Perhaps Loongson's 16-core 3C6000 and 32-core 3D6000 processors will be able to challenge the real-world performance of AMD's contemporary low-power EPYC processors in programs that can take advantage of LoongArch's innovations. Still, we would not expect Loongson's 2023 offerings to be competitive against AMD's 2021 CPUs.

Interestingly, due to its 16-core 3C7000, 32-core 3D7000, and 64-core 3E7000 processors due in 2024~2025, Loongson will continue to rely on its LA664 core design, though it will probably produce them on a new node to increase their performance per watt efficiency.

Up to 64 cores and improved performance per watt might pose a more significant challenge to CPUs from AMD and Intel, but expecting Loongson's products to replace processors from leading designers in performance-demanding applications is too optimistic.

China's Loongson plans to formally introduce and start shipping its 3C5000 processors. based on its own LoongArch instruction set architecture and featuring up to 16 cores. The CPUs will be aimed at servers, and promise to bring a number of advantages, due to their ISA enhancements.

Loongson’s LS3C5000 processors feature up to 16 LA464 cores, up to 64MB of cache, and four 64-bit DDR4-3200 memory interfaces with ECC support, according to eZone.hk. Each of Loongson’s LA464 cores have four general-purpose ALUs, and two 256-bit vector operations units. Also, being LoongArch-based, they feature 2,000 proprietary instructions, binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ). The CPU is reportedly made using SMIC’s 12nm fabrication process, an enhanced version of its 14nm manufacturing technology.

Previously it was reported that Loongson’s 3C5000 CPUs with up to 16 cores use four quad-core LS3A5000 CPUs, connected on the same piece of substrate using AMD’s HyperTransport 3.0 technology, but this is incorrect, the 3C5000 chip uses a monolithic core, but has HyperTransport 3.0 links for symmetric multiprocessor configurations. The 3A5000 chips for client PCs were introduced last year and have been shipping since then.

There is no information about clock speeds for the 3C5000 CPU, or thermal design power, but keeping in mind that 3A5000 chips (operating at 2.30 GHz – 2.50 GHz) were designed with notebooks in mind, the 3C5000 should not be too termally demanding.

Loongson originally planned to formally introduce and start shipments of its 16-core 3C5000 processors for servers by the end of 2021, but it looks like it had to delay its product launch a bit, possibly to polish off the platform and/or wait before server software gains support for at least some of the proprietary instructions supported by the CPU.

The GNU project has announced that its GNU Compiler Collection (GCC) version 12.1 now supports Loongson's proprietary LoongArch microarchitecture. The addition of LoongArch support to a popular set of compilers will enable software developers to better address systems based on the latest processors from Loongson. 

"Support for the LoongArch architecture instruction set has been added," release notes for the new GCC read. "The Loongson CPU codename LA464 and LoongArch 64-bit generic CPU codename loongarch64 are supported through the -march= and -mtune= options (GCC identifiers in parentheses)." 

It is unclear whether the new compiler collection supports all ~2,000 new instructions introduced by the LoongArch architecture, but hopefully it supports those that bring the most significant performance improvements. 

For many years Loongson's processors used various kinds of its developer's LoongISA architecture, which in turn was a custom subset of the MIPS64 architecture, which was designed for various applications, including general PCs as well as high-performance computing. This made a lot of sense for Loongson as it allowed the company to maintain compatibility with programs designed for MIPS64 and introduce its own extensions to improve the performance of specific programs. But no matter how good MIPS64 used to be, it is an older architecture and Loongson needed something new. 

Last year Loongson finally introduced its 3A5000 and 3C5000 processors based on its own LoongArch instruction set architecture, which maintains compatibility with MIPS, but has 2,000 proprietary instructions, binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ). The first attempt to enable the new CPUs in Linux was not an outright success as the chips largely used MIPS code, but now that GCC 12.1 finally gains support for LoongArch, this situation may change. 

Competitive processors are crucial for China, which cannot access HPC hardware developed in the U.S. For now, Loongson's CPUs are aimed primarily at client machines and servers, but once LoongArch gets proper support from software (and this includes popular compilers on the first place), it will certainly make sense to expand the list of applications addressed by these processors.

The Fenghua No.1 GPU recently passed an important milestone for mass adoption in China. The GPU, created by Xindong Technology and Innosilicon, was initially announced in November 2021. We received some interesting but not entirely revealing performance indicators a month later. This month the GPU was certified for running stably and offering "outstanding performance on the Tongxin UOS operating system," reports ITHome.

Tongxin UOS is quite important in China as it was developed as part of a government initiative to usurp Windows. It is based upon Debian Linux and is adapted to support the latest homegrown silicon. So the Fenghua No.1 GPU certification is a significant step, alongside the implementation of support for Chinese-designed semiconductor products such as Zhaoxin CPUs and GPUs, processors from the likes of Sunway and Loongson.

The GPU sounds like it is definitely up to par for providing a 'Chinese' solution for graphics acceleration and other GPU tasks. We put the provenance in quotes as our previous investigations indicated that Fenghua No. 1 is based upon Imagination Technologies' PowerVR architecture, though we can't be 100% certain about this aspect of the GPU at this time.

The new GPU completed "comprehensive in-depth adaptation certification" in Tongxin UOS. Moreover, it could be China's first domestic " 4K-level high-definition desktop and high-performance server-level graphics card," to have been certified. It is claimed to exhibit "excellent performance, power consumption, reliability and compatibility, and fully supports mainstream graphics frameworks such as OpenGL/ OpenGL ES/ OpenCL/ Vulkan to achieve performance breakthroughs," says the Chinese source publication.

We hope to see systems featuring the new GPU available for sale soon, and some third-party benchmarks give us insight into the nature of this desktop GPU.

Before signing off, it would be amiss not to remind readers of the specs of the first Fenghua No. 1 graphics cards. In brief, we hear that the PowerVR architecture GPU is built on a 12nm process in China, and it supports up to 19Gbps GDDR6/6X. SKUs are being configured with 4, 8 and 16GB of VRAM, and outputs include HDMI 2.1, DP 1.4, and VGA with multiple monitor support. One of our December reports on the Fenghua No.1 GPU indicated that there would be a version of the card with dual GPUs, effectively doubling up all the specs such as FP32 and INT8 performance.

More GPU rivals are always welcome, and it would be a remarkable chain of events if the Fenghua No. 1 becomes popular in China. However, we aren't sure we need another company aiming for the near-RTX 3060 space.

Rather than wait for China to rescue the affordable GPU market, we already have hopes of a significant correction coming, as GPU pricing seems to be taking a nose-dive. While you wait for your own GPU pricing sweet spot to be reached, check out our Best Graphics Cards for Gaming in 2022 guide.

Update 4/16/2022 7:45am PT: Corrected article to clarify Intel's statements and specify that the comments did not occur at a political event. 

Amended article: 

Some interesting comments made by an Intel senior executive in China have come to light. According to a report from Taiwan's DigiTimes, an Intel exec said that Chinese CPU makers could become "strong rivals" in the next three to five years. Rui Wang, SVP of Intel Corporation and chair of Intel China, made the comments to Guancha.cn, a Chinese media outlet.

"So far there has not been any local companies that are able to deal a substantial threat to Intel," stated the Intel SVP. "But in 3-5 years, it will become clear that local companies will emerge as strong rivals," envisioned Ms. Wang (as translated by DigiTimes). Having said this, Wang went a little distance to temper expectations that US tech giant Intel would easily be displaced, dislodged, or usurped as the CPU leader in China. "Intel won't be polite, and will exert its power to compete fairly."

Chinese Companies With Intel in Their Sights

Unfortunately, Intel China's Chair wasn't specific about the avenue from which Intel's business in China would be assaulted. No specific Chinese CPU or IT company was highlighted, and no mention was made of any specific architecture with a chance at becoming an Intel competitor.

China has quite a wide range of CPU design outfits that could potentially surprise us in the next three to five years. We have previously and recently reported on the successes of China-made Zhaoxin CPUs. One of these LuJiaZui microarchitecture quad-core chips features in what looks like a China-only version of the Yoga Slim 7i laptop, dubbed the Lenovo Kaitian N7. The immature architecture combined with a custom China OS means that the N7 isn't a great performer, however, and it looks like it has issues with waste heat and battery life.

We have also posted news in recent months discussing Chinese chip advances by the likes of Loongson (MIPS based), Sunway (supercomputers), and Hygon (AMD Zen 1 based, under license). China also has at least a couple of firms building processors on the Arm architecture. Huawei's HiSilicon and Phytium Technology are the two best know Arm architecture proponents in China. Phytium has a particularly wide portfolio with products for the server, workstation, and embedded markets, as well as five new high performance CPUs penciled in for the coming months.

China has homegrown GPUs and DDR4 (currently planning DDR5 production) too, and hopes to make its semiconductor foundries, especially the state backed SMIC, more competitive in the not-too-distant future.

Some other important data with which to frame the above news includes a recent statement by Xiao Yaqing, China's Minister of Information and Technology. Just ahead of the National Party Congress, Yaqing asserted that the domestic chip industry had grown by a third compared to a year ago.

Even though Loongson claims that its latest 3A5000 and 3C5000 processors are based on its in-house developed proprietary architecture and no longer rely on the MIPS architecture, the code the company uses to enable the new CPUs in Linux is actually the same code it used for its MIPS-based chips. Furthermore, the CPU developer fails to demonstrate advantages of its architecture even on paper to software developers. 

Historically, Loongson's CPUs relied on various types of the company's LoongISA architecture, a custom subset of the MIPS64 architecture. Such a tactic enabled the company to preserve compatibility with programs designed for MIPS64 (which includes software for high-performance computing applications) while bringing in its own extensions to improve performance in contemporary applications. 

But to make significant leaps in performance, Loongson needed a brand-new architecture, which is why it developed its LoongArch. The architecture continues to be MIPS-compatible, but it has about 2,000 of proprietary instructions, binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ). 

But instead of writing up new code to enable LoongArch-based CPUs in Linux, the company continues to use the old code that was written for MIPS64-powered processors, which causes some frustration in the community, reports Phoronix. 

"You keep saying 'not MIPS,' and yet all I see is a blind copy of the MIPS code," a software developer wrote to Loongson. "This is still the same antiquated, broken MIPS code, only with a different name." 

Since Loongson's LoongArch-based 3A5000 and 3C5000 CPUs can execute code designed for MIPS64 platforms and there may not be too many differences between the company's LoongArch and MIPS64 platforms, there is nothing bad about using proven code to enable them in Linux. However, using old code means that any new capabilities of the platform remain hidden. 

"What is that? Yet another MIPS legacy? Why does it have to be per interrupt if it obviously apply to each and every root interrupt?" another developer wondered. "You still obviously have some static partitioning of the interrupt space, which is not acceptable for a new architecture."

As of now, it remains to be seen when Loongson enables all-new capabilities of its LoongArch architecture in Linux.

Loongson, a Chinese CPU developer, has formally introduced its next generation 3A5000 CPUs that are based on the company's proprietary microarchitecture. The company says that the new processors for client PCs deliver up to 50% higher performance amid lower power consumption than predecessors, which may make it more competitive against CPUs developed by leading designers.

Loongson's 3A5000/LS3A5000 processor is a quad-core CPU operating at 2.30 GHz – 2.50 GHz. The cores are superscalar, each core has four general-purpose ALUs and two 256-bit vector operations units, reports CnTechPost. The cores are based on the LoongArch GS464V instruction set architecture, which reportedly features almost 2,000 proprietary instructions. In addition to the base ISA, LoongArch supports binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions. 

As far as performance is concerned, Loongson claims that its 3A5000/LS3A5000 CPU is over 50% faster than its predecessor LS3A4000 (probably when running software designed for LoongArch) and also consumes 30% less power. Keeping in mind that the LS3A4000 was on par with AMD's excavator, the new LS3A5000 will offer performance close to that of AMD's 1st generation Ryzen. The developer says that the new CPU hits fixed-point and floating-point single-core base scores of 26+ and quad-core scores of 80+ in SPEC CPU2006 when compiled using GCC. 

To cut down power consumption, the 3A5000 implements dynamic frequency, dynamic voltage regulation, and can dynamically shut down unused hardware.

The 3A5000 CPU features two DDR4-3200 memory controllers with ECC and four HyperTransport 3.0 controllers with coherency support to enable SMP (symmetric multiprocessing) capability. Like all Chinese processors, Loongson's 3A5000 chips support the country's encryption standards, such as SM2, SM3, and SM4 in hardware. 

Previously Loongson called its next-generation processors 'MIPS64-compatible,' which suggests that the new chips can still execute programs designed for the company's LoongISA architecture, which is a subset of the MIPS64 ISA. Meanwhile, GCC, LLVM, and GoLang compilers already support the LoongArch GS464V instruction set architecture. Also, the ISA is supported by several virtual machines.

Loongson's 3A5000 CPUs will be sold primarily to Chinese PC makers, so their pricing is unknown for now and will hardly be ever disclosed publicly. 

Loongson Technology is one of a few well-known CPU designers from China that is not on the U.S. blocklist, therefore potential investors can invest in the company without major risks. This week the company filed for an IPO on Shanghai's STAR Market in a bid to raise money and fund the development of its future CPUs and GPUs. 

Loongson plans to raise 3.5 billion yuan ($544 million) to replenish its working capital and increase its R&D spending on its next-generation CPUs and GPUs, reports Nikkei. The news agency claims that Loongson earned 1.1 billion yuan in operating revenue, two times more than in the previous year. But at the same time, its profitability in 2020 dropped to 72 million yuan from 193 million in 2019. 

Earlier this year, Longson Technology introduced its own LoongArch architecture to power its processors. The company's quad-core 3A5000 CPU for client PCs and 16-core 3C5000 CPU for multi-processor servers are the first chips to use the new architecture. The 3A5000 is already available, whereas the company will release the 3C5000 later this year. 

Due to US restrictions, China cannot rely on hardware from companies like AMD, Intel, or Nvidia, making developing its own platform for exascale supercomputers one of its main challenges. To do so, China needs to develop high-performance GPUs in addition to high-performance CPUs. As it turns out, Loongson will be one of the companies that will develop GPUs that could potentially be used for next-gen supercomputers.   

While Loongson is a privately owned company, the Chinese government has supported it through subsidies and tax breaks. Also, government offices have been buying PCs based on Loongson-developed CPUs since 2018.

Chinese processor developer Loongson Technology this week announced that it had developed its own CPU instruction set architecture (ISA), Loongson Architecture (LoongArch) that has nothing to do with architectures designed outside of China. Surprisingly, the first processors based on the new architecture have already been taped out and will be released commercially this year. 

For many years Loongson's processors relied on different versions of the company's LoongISA architecture, which were subsets of the MIPS64 architecture. This approach allowed the company to maintain compatibility with software developed for MIPS64 (which includes software for supercomputers) and also introduce its own extensions to boost performance in modern applications.  

By contrast, LoongArch is said to feature almost 2,000 of proprietary instructions. Meanwhile, the company said that it had removed outdated instructions not suitable for modern hardware designs to ensure low power consumption and simpler designs. In addition to the base instruction set, LoongArch features binary conversion extension instructions (LBT), vector processing extension instructions (LSX), advanced vector processing extension instructions (LASX), and virtualization extension instructions (LVZ), reports PC Watch.

Apparently, the first CPUs to use the LoongArch are Loongson's quad-core 3A5000 for client PCs and 16-core 3C5000 for multiprocessor servers. Previously Loongson called these processors 'MIPS64-compatible', so they can execute programs originally developed for previous-generation Loongson processors that relied on a superset of the MIPS64 architecture. The Loongson 3A5000 is set to ship in the first half of 2021, whereas the 3C5000 is projected to be available towards the end of the year, according to a previous report. 

Loongson Technology has released its LoongArch infrastructure instruction system manual to some interested parties for evaluation, but is yet to publish it for a broader audience of software developers. Also, the LoongArch is currently evaluated by the Chinese IP agency.

China has been striving for semiconductor self-sufficiency for years now and even created its Made in China 2025 program that set the time-frame by which this self-sufficiency is set to be achieved. Analysts are skeptical that China will be self-sufficient by 2025 as far as semiconductors are concerned, but the country is moving towards its goal. 

A proprietary CPU ISA is certainly a step towards self-sufficiency as Loongson no longer has to rely on Arm, MIPS, x86, RISC-V or other Western processor architectures. However, only time will tell whether the company can actually develop a competitive ecosystem for its LoongArch. 

It is noteworthy that Loongson Technology has once announced plans to transit to open-source RISC-V, so if LoongArch fails, the company will have a plan B.

Executives of Loongson Technology, a subsidiary of the Chinese Academy of Engineering, said at a recent conference that the next-gen Loongson 5000-series processors were on track to be released this year. The new MIPS64-compatible CPUs are aimed at client PCs as well as multiprocessor servers. Interestingly, the new chips may be the last high-end MIPS64 offerings from the Chinese Academy of Engineering.

The chips in question are the 2.50 GHz quad-core Loongson 3A5000 for client PCs and 16-core Loongon 3C5000 for servers with up to 16 processors. Both chips are set to be made using a 12nm process technology (most likely one of TSMC's nodes), reports CnTechPost, citing a small conference that was held earlier this year. Both CPUs are said to be based on a new internal architecture that is compatible with the MIPS64 instruction set, feature enlarged caches, and a new memory controller.

Based on some previous reports, the 3A5000 was taped out in April 2020, which is why it is due in the coming months; whereas the 3C5000 was taped out in August, 2020, so it will be released towards the end of 2021 if everything goes as planned.

One interesting thing about Loongson Technology is that the company is reportedly 'looking forward to join the open-source instruction consortium.' The consortium mentioned by Loongson's executives is almost certainly RISC-V International, which essentially means that going forward, the company will focus on RISC-V.

Loongson has historically developed MIPS-compatible CPU cores, so switching to RISC-V should not be too challenging for the company as the architectures have many similarities. Meanwhile, the adoption of RISC-V means that Loongson's upcoming processors (or cores) will be supported by a broad ecosystem of software and hardware, something that will inevitably make them more competitive.

Developing new RISC-V-compatible microarchitectures and cores will take several years, so for now, Loongson will have to promote its 3A5000 among PC makers and its 3C5000 among server and HPC customers.