Goodbye dedicated AI hardware and hello to a GPU that fuses Xe graphics DNA with Habana chemistry
On paper, Intel's Habana Gaudi3 AI accelerators don't look like they're ready to take on Nvidia's H100 thanks to older process tech and slower HBM memory delivering fewer FLOPS. But Gelsinger's gang insists its latest parts can not only go toe-to-toe with the H100 in inference, but best it in training.
Intel's Gaudi3 accelerator boasts eight matrix math engines, 64 tensor cores, 96 MB of SRAM, and 128 GB of HBM2e memory – click to enlarge The often referenced figure is perhaps misleading. The 1,835 teraFLOPS claimed by Intel is dense floating point performance, while Nvidia is relying on sparsity to achieve its 4 petaFLOP claims. Taking this into account, Gaudi3 is only about 144 teraFLOPS slower than the H100 while offering more memory grunt.Then there's AMD's MI300X, which remains the FP8 FLOPS king, at least until Nvidia's Blackwell parts start making their way into customer's hands.
Here Gaudi3's older memory – HBM2e vs HBM3 on the H100 and MI300X, and HBM3e on the H200 – puts it in an odd spot.With eight stacks of HBM2e, Gaudi3 actually has more and faster memory than the Nvidia's H100 with its five stacks of HBM3. Despite this, the chip still falls well behind the H200 and MI300X's HBM3e memory, which deliver 4.8 GBps and 5.3 TBps of bandwidth respectively.
Each Gaudi3 accelerator features 24 200GbE interfaces, 21 for chip-to-chip comms and three for system-level networking – click to enlarge
Nigeria Latest News, Nigeria Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Nvidia turns up the AI heat with 1,200W Blackwell GPUs5x the performance of the H100, but you'll need liquid cooling to tame the beast
Read more »
Nvidia rival Cerebras says it's revived Moore's Law with third-gen waferscale chipsStartup is also working with Qualcomm on optimized models for its Cloud AI 100 Ultra inference chips
Read more »
Final Fantasy 7 Rebirth Gongaga Protorelic – all Gongaga Phenomenon Intel locationsAre you wondering how to get all the Gongaga Protorelic pieces in Final Fantasy 7 Rebirth? Defeat the Turks battle simulations.
Read more »
AI bubble or not, Nvidia is betting everything on a GPU-accelerated futureLLMs powering generative AI may be moving GPUs, but Huang and co already looking at next big opportunity
Read more »
UXL Foundation readying alternative to Nvidia's CUDA for this yearAn open standard challenger appears
Read more »
Nvidia software exec Kari Briski on NIM, CUDA, and dogfooding AIA RAGs to riches story
Read more »