Samsung readies Gaia AI accelerator for PCs — HP and Lenovo are reportedly validating the NPU
How might this NPU be used?
Samsung is reportedly sampling its dedicated AI processor for next-generation AI PCs with leading PC makers, such as HP and Lenovo. The chip, codenamed Gaia, was developed by the company's System LSI business unit, and it is designed to offload AI-related workloads from the CPU and GPU, reports Chosun.
Samsung's Gaia is designed to accelerate generative AI workloads on PCs and is made using the company's 4nm-class fabrication process. The chip, which is essentially a neural processing unit (NPU), is currently being evaluated by HP in the U.S. and Lenovo in China to verify its performance and evaluate whether it makes sense to integrate Gaia into their systems due in late 2027 or early 2028.
The report does not detail how Gaia differs from NPUs that are integrated into AMD's Ryzen, Intel's Core, or Qualcomm's Snapdragon X processors as well as whether it can offer significant performance advantages. Meanwhile, the report implies that the NPU (or perhaps its derivatives based on the same architecture) could be used for Samsung's next-generation implementations of its processing-in-memory (PIM) technology.
Samsung's original PIM was designed to embed compute logic directly within the HBM memory array and reduce data movement between HBM memory modules and host processors. PIM was aimed to accelerate select workloads, but did not take off because AI and HPC GPUs became very efficient and were supported by mature ecosystems, unlike PIM.
Perhaps if Samsung's upcoming Gaia NPU gains support from hardware makers and ecosystem partners, then this will give a boost to Samsung's next-generation PIM implementation as well. However, standalone NPUs and PIM are so fundamentally different that we can barely imagine that they can share a common architecture. Yet, PIM logic can be a subset of an NPU in terms of supported instructions and data formats and they can certainly share a common software framework.
One of the interesting things to note about Gaia is that it was reportedly developed by Samsung's LSI division, the same business unit at the company that is responsible for Exynos processors, automotive solutions, connectivity chips, ISPs, DSPs, display drivers, and image sensors. Given the multi-faceted nature of Samsung's LSI unit, as well as its strategic importance for the company, Samsung must be pinning some hopes on Gaia.
We have contacted Samsung and asked for a comment about the report, but we yet have to hear back from the company.
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
-
usertests Who is it for? Are office PCs going to be running 8B LLMs with an add-in card?Reply
Nova Lake-S desktops should be integrating a >70 TOPS NPU, and AMD might end up joining the fun with 50 TOPS (XDNA2) or XDNA3.
I guess Gaia would have to take up an M.2 slot if they want to get it into a lot of "AI PCs". -
theverge Reply
Yet to find a single use for my current NPU... but sure, lets buy another oneAdmin said:Samsung reportedly preps Gaia AI accelerator for client devices that is already being tested by HP and Lenovo.
Samsung readies Gaia AI accelerator for PCs — HP and Lenovo are reportedly validating the NPU : Read more -
salgado18 Reply
No, 30B LLMs and up. Once onboard memory catches up, there will be little need for external servers, as most tasks will be local. And there will be a lot of cases for AI work, everywhere.usertests said:Who is it for? Are office PCs going to be running 8B LLMs with an add-in card?
Nova Lake-S desktops should be integrating a >70 TOPS NPU, and AMD might end up joining the fun with 50 TOPS (XDNA2) or XDNA3.
I guess Gaia would have to take up an M.2 slot if they want to get it into a lot of "AI PCs". -
bit_user Reply
No way it'll be a M.2 slot. Several reasons for that, but the first two are deal-breakers:usertests said:Who is it for? Are office PCs going to be running 8B LLMs with an add-in card?
Nova Lake-S desktops should be integrating a >70 TOPS NPU, and AMD might end up joining the fun with 50 TOPS (XDNA2) or XDNA3.
I guess Gaia would have to take up an M.2 slot if they want to get it into a lot of "AI PCs".
power limits
limited space for DRAM
cooling-related challenges
limited PCIe bandwidth
Most PCs that even have a PCIe 5.0-capable M.2 slot only have one and it's being used by the system drive. So, I'd bet on a PCIe 5.0 add-in card. -
bit_user Reply
Uh, that's topologically impossible. The memory controller sits inside the CPU package. For Samsung's NPU to talk to your system RAM, it must go through the CPU.dva852 said:Gaia is a derivative of Samsung's PIM (processing in memory) work, so the likelihood is that it would need to be in close proximity to system RAM. Shorter trace length means lower latency.
Or, Samsung already has lined up production for a 1st-party AIB (similar to their Samsung-branded SSDs) and therefore doesn't need to talk with 3rd party AIBs. They just need to talk to systems vendors to ship PCs with their new AIB pre-installed.dva852 said:Talks with system vendors and not AIB vendors affirms the point.
But, common AI frameworks reduce this burden, because all you need is for your desired AI framework to have a backend that supports your hardware.dva852 said:A large problem is that 'NPU' is largely a marketing term with no standard API. Each vendor (Intel/AMD/Apple/Qualcomm/et al) requires separate supporting code for support. ONNX runtime, used in Windows, is the de facto standard for NPU interoperability.