AMD has declined the request of the AI startup Tiny Corp to put the Instinct MI300X AI accelerator on the MLPerf benchmarks.
AMD Shows Reluctance With Instinct MI300X AI Accelerators & Its Performance, Rejects Benchmarking Request
AMD's journey in the AI industry has been a roller coaster ride, and they are eager to achieve a breakthrough yet are restricted for multiple reasons. Despite having one of the most potent AI product lineups, AMD hasn't managed to attract market attention, well at least on the same level as NVIDIA but much better than what Intel has so far procured, where the obvious reasons include the inclination towards NVIDIA, but that isn't everything.
The whole Tiny Corp-AMD fiasco has shown that Team Red seems reluctant to cater to clientele interests. Despite multiple attempts by Tiny Corp to promote AMD's products, the chipmaker always takes a step back.
.@AMD @LisaSu For $1M and two boxes, we'll get the MI300X on the next MLPerf.
— the tiny corp (@__tinygrad__) June 18, 2024
A few days ago, Tiny Corp offered AMD to put their Instinct MI300X AI accelerators on the next round of MLPerf benchmarks for a million dollars and two units of the AI GPUs. For those unaware, the MLPerf is an industry-standard benchmark suite designed to measure the performance of AI products, especially accelerators. The important thing about them is that they are transparent, similar to PC counterparts like 3DMark. Hence, they are deemed important in the eyes of potential customers of an AI product.
As of right now, NVIDIA's AI accelerators remain at the top in MLPerf while Intel's Gaudi chips are the only benchmarked alternative, showcasing relatively strong perf/$.
Our offer to get MI300X on MLPerf was declined btw. https://t.co/I5w9cbpv63
— the tiny corp (@__tinygrad__) June 20, 2024
We weren't ignored. It's so strange, they don't even clearly say no to the offer, they sidestep it and use meaningless words like "partnership" and "collaboration"; want to have "phone calls." Would much rather a clear no.
Not, "what's your address we're sending two boxes"
— the tiny corp (@__tinygrad__) June 20, 2024
However, AMD might have shown reluctance to upload MLPerf benchmarks for its Instinct MI300X AI accelerators, invalidating the GPU's claims to be superior to competitor alternatives. To top it all off, Team Red decided to reject Tiny Corp's request as well, showing that something is wrong in the firm's camp, whether it is their reluctance to showcase the capabilities of their product or AMD has something under the covers that they won't like to be disclosed for now.
This isn't just it; we won't go into how AMD hasn't managed to compete with NVIDIA's CUDA despite seeing a huge market gap, but there's a lot to talk about the firm's approach. Recently, AMD's Executive VP & GM of Data Center Solutions Business Group, Forrest Norrod, claimed that the firm has adjusted its AI roadmap in light of NVIDIA's "push to the accelerator," which shows that for now, the company looks more focused towards the hardware side of things, rather than creating the right ecosystem, which NVIDIA has already done successfully.
It will be interesting to see how AMD moves in the future, given that the firm is slacking in multiple areas but in AMD's defense, we must state that asking straight up for a million dollars is a bit absurd. Regardless, with the recent Computex keynote, AMD did show optimism about how things are moving, so we can only wait and watch.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
