MLPerf training tests put Nvidia ahead, Intel close, and Google well behind

Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by fnbr on Sunday November 12, 2023 @11:49PM (#38247052)

This is dumb- they’re testing against google’s v5e chip, which isn’t the H100 equivalent (the V5P is). Of course Google came out worse here.

[ Parent | on HN ]

Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by kaelinl on Monday November 13, 2023 @12:02AM (#38247112)

MLPerf does not test hardware. Companies submit results from their own hardware. This is what each company chose to submit.

[ Parent | on HN ]

Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by emu on Monday November 13, 2023 @07:14AM (#38249343)

But the interpretation of the journalist is the issue here. Comparing a cost optimized part (TPU v5e) against a performance optimized part (H100) and deciding that this makes Google "behind" is just incorrect.

[ Parent | on HN ]

Re:MLPerf training tests put Nvidia ahead, Intel close, and ... by dgacmu on Monday November 13, 2023 @07:35AM (#38249478)

There is no publicly announced v5p. Source?

[ Parent | on HN ]