Intel CEO Pat Gelsinger countered NVIDIA’s Grace CPU announcement
Intel CEO Pat Gelsinger said in an interview with Fortune:
We announced our Ice Lake server line last week with an overwhelmingly positive response. And in Ice Lake we have consistently expanded its capabilities with regard to artificial intelligence. NVIDIA is responding to us. We are not the ones who answer them. Clearly this idea of AI-powered CPUs is where Intel is a very strong leader.
NVIDIA's Grace processor for AI and HPC machines promises to be more than 10x faster at AI and in select HPC workloads versus today's x86 processors, but it will only be launched in early 2023. Expectations like these always seem pretty incredible, but it's worth bearing in mind that the competition - AMD and Intel - won't stand still. on our hands.
Intel's latest Xeon processors integrate AI acceleration technologies under the generic name DL Boost, which currently includes extended instruction sets such as AVX512_VNNI (Cascade Lake and Ice Lake) and AVX512_BF16 ( Cooper Lake only). Both have proven to be quite competitive overall (at least according to Intel's testing) and when it comes to optimized algorithms they can make Intel CPUs outperform NVIDIA's GPUs by up to 15x.
Also, because Intel's Xeon CPUs are so popular they are widely used for inference workloads, and ISVs optimize their engines for these processors. In short, Intel's modern scalable Xeons can already count on improvements to AI, and their successors will naturally expand their capabilities.
In general, ARM, AMD, Intel, NVIDIA and other companies are working hard to improve the performance of their processors (CPU, GPU, IPU, VPU, etc.) in AI and / or HPC workloads due to a constant and growing demand for such operations. As a result, there will be more progress related to these areas in the coming years than in the previous thirty.
Looking for a new PSU to power your next GPU? Corsair RM750X, 750W modular power supply, is available on Amazon.
