On-Prem

HPC

AMD opens wallet to lure scientific computing boffins away from Nvidia's CUDA onto its Instinct accelerators

Only academics, researchers, nonprofits need apply to AIER initiative


AMD has rolled out a programme that it hopes will tempt scientific computing types away from Nvidia and the CUDA ecosystem and onto its own Instinct accelerator range and ROCm run-anywhere software platform: the AMD Instinct Education and Research (AIER) initiative.

"At AMD, we're committed to engineering the best products and tools to make that happen – and the new AMD Instinct Education and Research (AIER) initiative makes those innovations easy to access," claimed AMD senior product marketing manager Guy Ludden of the programme's launch.

"If you're looking for ways to solve big challenges, make discoveries, analyse more data, or make better decisions faster, taking part in the AIER Initiative can give you access to the resources you need to get the job done."

Unveiled at the ISC HPC 2021 conference this week, AIER is aimed at convincing academics to jump across to AMD's high-performance computing product lines. Membership is free, but limited: only "academics, researchers, and scientists working at higher education universities or institutions, along with non-profit organisations" are invited.

The initiative appears positioned to pull users away from rival Nvidia's CUDA platform for general-purpose GPU offload and towards AMD's Instinct family of GPU-based accelerators. Back in November, AMD launched the Instinct MI100 accelerator, described by corporate VP Brad McCredie as "designed to provide scientists and researchers a superior foundation for their work in HPC."

While McCredie didn't specifically name Nvidia as the inferior foundation, some have inferred it.

Once approved for the AEIR initiative, members receive access to a range of support services from code validation and certification through to community-driven support forums, workload containers, and tutorial walkthroughs demonstrating just how to get started with what will be for most an unfamiliar environment.

There's one resource that may tempt interested parties into signing up more than the others, however: cold hard cash. In an apparent indication that AMD is putting its money where its mouth is to take on Nvidia in the high-performance computing arena, the AIER initiative documentation included the promise that "AMD often provides grants and/or hardware discounts for our academic and research colleagues" – though stopped short of promising payouts for members.

"High performance computing is critical to addressing the world's biggest and most important challenges," crowed AMD's senior VP Forrest Norrod in a canned statement.

"With our AMD Epyc processor family and Instinct accelerators, AMD continues to be the partner of choice for HPC. We are committed to enabling the performance and capabilities needed to advance scientific discoveries, break the exascale barrier, and continue driving innovation."

The announcement came as AMD celebrated an increasing presence in the TOP500 list of the world's most powerful supercomputers, with its processors driving five times more systems than at this time last year – though it still powers less than 10 per cent of the listed systems, compared to rival Intel's 86.2 per cent.

Interested parties can find out more about the AIER initiative, which has Dell, Gigabyte, HPE, and Supermicro as its industry partners, here. ®

Send us news
6 Comments

AI cloud startup TensorWave bets AMD can beat Nvidia

Starts racking MI300X systems - because you can actually buy them and they beat the H100 on many specs

TSMC boss says one-trillion transistor GPU is possible by early 2030s

Timing predictions aside, multi-chip designs with 3D stacking will be the path forward

Intel Gaudi's third and final hurrah is an AI accelerator built to best Nvidia's H100

Goodbye dedicated AI hardware and hello to a GPU that fuses Xe graphics DNA with Habana chemistry

Tiny Corp launches Nvidia-powered AI computer because 'it just works'

Startup slams AMD for buggy firmware

Los Alamos Lab powers up Nvidia-laden Venado supercomputer

Promises up to ten AI exaFLOPs of grunt – if you turn down the resolution a bit

China scientists talk of powering hypersonic weapon with cheap Nvidia chip

Jetson module can efficiently process computational fluid dynamics models

AMD to open source Micro Engine Scheduler firmware for Radeon GPUs

And it was all thanks to peer pressure

Latest AMD Ryzen Pro chips are similar silicon, more smarts

That other processor company really wants you to use AI at work

China orders its telcos to rip and replace US chips with homegrown silicon by 2027

There's no Huawei we saw that coming

Next Vision, or Vision Next? What we really thought about Google and Intel's AI events

We sat through these conferences so you didn't have to

Lambda borrows half a billion bucks to grow its GPU cloud

Will buy tens of thousands of Nvidia's prized accelerators, which will be collateral for the loan

How this open source LLM chatbot runner hit the gas on x86, Arm CPUs

Way to whip that LLaMA's ass