Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why isn't AMD's MI300X competitive?

newsletter.semianalysis.com

Post reply on HN
Print view View original

21–30 of 38 posts

Re: Why isn't AMD's MI300X competitive?

#21

Post by roenxi »

andy_ppp wrote: post #7

Please just get everything in PyTorch to work, and work well (and across all graphics cards too). This is the starting point and it doesn't matter how you do it. But the fact you cannot even do some very basic stuff on AMD is going to mean you are left unused by researchers, so getting further up the stack is going to be almost impossible.

Does PyTorch not work on AMD cards? I remain very nervous about returning to the AMD ecosystem. On paper AMD has been a compelling choice for GPGPU work for years, up until it turns out the hardware can't actually do what it claims. But the PyTorch problem seemed to be largely solved years. The issues weren't on the application layer, it was crippling firmware bugs that they didn't seem interested in getting a handle on. PyTorch ran fine until the computer kernel paniced or whatever, but that isn't a PyTorch problem.

Re: Why isn't AMD's MI300X competitive?

#22

Post by agunapal »

Nvidia had the first movers advantage. Nvidia spent so many years perfecting CUDA to work well with PyTorch. Before ROCM, there was only CUDA. There were so many developers building their use cases on top of PyTorch+CUDA, and bringing all that feedback to PyTorch, this made CUDA battle ready and stable. AMD can get there, especially now with demand for compute, but as someone already said here, the biggest focus needs to be on PyTorch

Re: Why isn't AMD's MI300X competitive?

#23

Post by dragandj »

oneofthose wrote: post #19

> AMD’s software experience is riddled with bugs [...] AMD’s weaker-than-expected software Quality Assurance (QA) culture and its challenging out of the box experience. This has anecdotally been true since forever. Back in the day, OpenCL implementations were passing conformance test but performance was poor. They could not turn hardware capabilities into performance for compute users. Drivers were buggy. Documentati…

FWIW Back in 2015 OpenCL 2.0 performance was quite good on then-current AMD GPUs (IMO), but the problem was that 1. You had to implement everything yourself, from scratch, since AMD's GPU BLAS was barely compilable, and 2. They abandoned OpenCL that year, and switched to HIP (or whatever their copy of CUDA was called) which didn't even compile (in practice) for quite some time, and 3. Even with HIP, you were on your own when it comes for any BLAS and other standard library implementations because AMD provided nothing of the sorts for a long time.

All in all, it's not that the drivers performance was poor per se, but AMD did nothing about providing a software ecosystem, which amount to its hardware wasn't realistically usable unless your pockets were so big that you can do AMD's job and fund the re-development of the whole ecosystem from scratch.

In other words, it made MUCH better ROI to just use Nvidia, pay a little bit more for the hardware, and save millions on software :)

Re: Why isn't AMD's MI300X competitive?

#24

Post by HarHarVeryFunny »

dingdingdang wrote: post #11
pstuart wrote: post #2

If AMD's betting the company on their AI compute, they had best follow the advice in the article because the only way to compete with NVIDIA is to meet/exceed not just the performance but also the DevX.

These days it's for sure the dev environment that is lacking, hardware is okay (potentially great?!), software abysmal. To run a local llm in a stable manner implies using Vulkan.. any attempt at ROCm is totally hamstrung by haphazard support of hardware alongside with an online presence poisoned by people primarily discussing work-arounds rather than work when it comes to AMD as a platform. Argh.

Is there any benefit of Vulcan vs ROCm on a card where ROCm is fully supported?

Re: Why isn't AMD's MI300X competitive?

#26

Post by wongarsu »

arka2147483647 wrote: post #14

The important part of Hardware, is Software After all, if the Software does not work, its just a Paperweight

And yet hardware companies with good software are the exception, not the norm. Is it just the cultural mismatch between hardware and software development life cycles and planning philosophies, or is there more to it?

Re: Why isn't AMD's MI300X competitive?

#27

Post by sorenjan »

dragandj wrote: post #23
oneofthose wrote: post #19

> AMD’s software experience is riddled with bugs [...] AMD’s weaker-than-expected software Quality Assurance (QA) culture and its challenging out of the box experience. This has anecdotally been true since forever. Back in the day, OpenCL implementations were passing conformance test but performance was poor. They could not turn hardware capabilities into performance for compute users. Drivers were buggy. Documentati…

FWIW Back in 2015 OpenCL 2.0 performance was quite good on then-current AMD GPUs (IMO), but the problem was that 1. You had to implement everything yourself, from scratch, since AMD's GPU BLAS was barely compilable, and 2. They abandoned OpenCL that year, and switched to HIP (or whatever their copy of CUDA was called) which didn't even compile (in practice) for quite some time, and 3. Even with HIP, you were on your…

Cuda also compiles to PTX, which makes it much easier to distribute and therefore also easier for users to actually use. Doesn't matter that much when you're writing code for specific hardware like MI300X, but it's part of the developer story.

Re: Why isn't AMD's MI300X competitive?

#28

Post by geremiiah »

I wonder if hiring is a big factor here. I presume, all the really good systems+parallel programmers would rather gain more experience on NVIDIA hardware than AMD, so given the choice, they'd go with NVIDIA. Does AMD do enough to win them over?

Re: Why isn't AMD's MI300X competitive?

#29

Post by shrubble »

oneofthose wrote: post #19

> AMD’s software experience is riddled with bugs [...] AMD’s weaker-than-expected software Quality Assurance (QA) culture and its challenging out of the box experience. This has anecdotally been true since forever. Back in the day, OpenCL implementations were passing conformance test but performance was poor. They could not turn hardware capabilities into performance for compute users. Drivers were buggy. Documentati…

Even before ATI was acquired by AMD they had driver support problems.

When I was working for a Unix commercial graphics software company, the CTO told me how bad the information he received under ATI’s NDA was: different revisions of the same chipset had contradictory register settings, so the driver had to identify the revision before writing a value to the write-only configuration registers. The same chipset might need a 0 or a 1; writing the wrong values could crash the driver.

Re: Why isn't AMD's MI300X competitive?

#30

Post by ElectricalUnion »

HarHarVeryFunny wrote: post #24
dingdingdang wrote: post #11

Earlier quoted context omitted.

These days it's for sure the dev environment that is lacking, hardware is okay (potentially great?!), software abysmal. To run a local llm in a stable manner implies using Vulkan.. any attempt at ROCm is totally hamstrung by haphazard support of hardware alongside with an online presence poisoned by people primarily discussing work-arounds rather than work when it comes to AMD as a platform. Argh.

Is there any benefit of Vulcan vs ROCm on a card where ROCm is fully supported?

On my gfx1030 "consumer grade hardware", ROCm means using SDMA, and that is broken for my system. Forcing `HSA_ENABLE_SDMA=0` makes it "work", but also makes loading tensors to VRAM take 15x longer.
Post reply on HN