Samsung announced its next-generation mobile application processor, the Exynos 9 Series 8895, and said it's the first 10nm processor built so far. This means Samsung beat Intel and TSMC to the next-gen process node, but Qualcomm's Snapdragon 835 could soon follow on the same 10nm process.
10nm Process Node
The 10nm FinFET process brings an improved 3D transistor structure that allows for up to 27% higher performance or up to 40% power consumption when compared to Samsung’s previous 14nm FinFET process.
Samsung said the Exynos 8895 is also the first chip to embed a gigabit LTE modem that supports five carrier aggregation (5CA). This allows wireless operators to combine multiple fragments of a spectrum to deliver higher data throughput. The modem can achieve up to 1Gbps (Cat. 16) downlink with 5CA, and 150 Mbps uplink with 2CA.
Brand-New CPU And GPU
The Exynos 8895 uses Samsung’s 2nd generation custom CPU core in a big.Little configuration. The chip uses four custom Samsung cores and four Cortex-A53 cores. It seems that Samsung, much like everyone else right now, has decided to stick with Cortex-A53 for a while longer, even though the core has surpassed its two-year life cycle, and ARM has already released the faster and more efficient Cortex-A35 core.
On the other hand, the new Exynos comes with the latest Mali-G71 GPU, based on the brand-new “Bifrost” GPU “scalar” architecture. The new GPU supports all the APIs and features developers could want on a mobile GPU architecture: OpenGL ES 3.2, Android Extension Pack, Vulkan 1.0, OpenCL 2.0, as well as the next-generation Adaptive Scalable Texture Compression (ASTC) algorithm.
The GPU supports not just 4K UHD video decoding at 120fps, but also 4K UHD VR gaming experiences. The 4K resolution for VR is likely where the sweet spot between image quality and performance is right now. However, mobile chips are still constrained by how much heat they generate, so the graphics quality in a 4K VR game likely won’t be much better than what you can experience today on 1440p mobile devices. The performance should be more than enough for immersive 4K VR videos, though.
Vision And Security Processors
Last year, we were discussing how it wouldn’t be long until smartphones will start using their own specialized “vision” or machine learning-focused processors. It looks like the Exynos 8895 chip is one of the first to come with a Visual Processing Unit (VPU) that can recognize items and movement in videos, as well as help process panoramic images faster.
This sort of chip could drastically improve computational photography if Samsung takes full advantage of it, which can help the phone’s camera surpass its own physical limitations in terms of the image quality it can produce.
The Exynos 8895 also comes with a dedicated processing unit that handles sensitive information such as fingerprint or iris scanning data. This could make the data more secure than when it was being protected by an implementation of ARM’s “TrustZone” software environment.
Samsung’s 10nm chip has already begun mass production, according to the company, so it’s possible it will be used by the upcoming Galaxy S8 smartphone.
Intel's chips are more powerful because they have a much higher tdp due to not having to fit into the thermal constraints of a cell phone.
_________________________
Uhhhh .... No.
Intel transistor density (if that is what you are talking about) at 14nm hardly surpasses GloFo 28nm SHP. ARM transistor density is likely 2X+ at 14nm -- even more at 10nm.
The whole nm thing is mostly marketing albeit usually at least some parts do have a "feature size" of that size specified. All sorts of factors like number of fins, fin height, fin pitch, gate length, etc come in to play.
Then of course the manufacturing part is only one piece of the puzzle, the chip design also will impact logic densities since this is also a Samsung chip on the Samsung manufacturing process its worth noting. You know if AMD's old bulldozer chips were made on Intels 14nm they wouldn't all the sudden be Kaby Lake's.
Not true. Samsungs 10nm process is about 10% more dense than Intel's 14nm.
Do you work for Intel by chance?
This is why ARM cores clock higher than x86 at the same process node and TDP, but ARM can't get itself to 3ghz so went many-core with its first PC offerings (in servers). ARM cores are specifically designed to be small and efficient. The core itself is just too tiny to cool with heatsinks (unless there's 48 of them warming each other in a server chip) and thus clock higher.
TL DR: Transistor density comes at the cost of heat. ARM are passively cooled so even though they are super low power they chose a process that prefers power efficiency at the cost of lower raw density.