News for hackers. Stuff that matters.
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by shmerl on Sunday November 12, 2023 @10:00PM (#38246390)
Why is AMD not compared?
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by ioedward on Sunday November 12, 2023 @10:07PM (#38246428)
AMD didn't submit results, despite geohot/Lisa Su's plan to get AMD on MLPerf: https://twitter.com/realGeorgeHotz/status/166980346408248934...
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by dhruvdh on Monday November 13, 2023 @12:11AM (#38247170)
If Dr. Lisa Su, the CEO, wants AMD to submit to MLPerf they would have already submitted to MLPref. This tweet is about a discussion George had with her on AMD's open-source practices and code quality for the contributions they make.
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by zwaps on Monday November 13, 2023 @03:03AM (#38248024)
AMD does not do ML. They don't have the capability and engineers, nor desire in their leadership, to develop the software stack.
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by shmerl on Monday November 13, 2023 @09:59AM (#38250889)
Makes no sense and not true most likely.
Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by brucethemoose2 on Monday November 13, 2023 @10:42AM (#38251459)
AMD does have a history of spinning their wheels on the software side. Not graphics specifically (anymore), but definitely with their compute initiatives like HSA or even OpenCL.

I dunno what to think about ROCm. It does seem like they should have thrown the MI300 into the mix, but maybe its not ready.

Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by zwaps on Tuesday November 14, 2023 @07:32AM (#38262518)
The proof is in the pudding. AMD has excellent hardware and has had years now to get anything going. They didn't, and they don't so our best guess is, they won't.