@stasbekman : semianalysis.com/p/the-memory-w… This is an important study as if you're not aware the new accelerators' efficiency keeps on going down because the compute speed up is happening faster than HBM speed up. So for example H1500 is less efficient than H200, even though the compute is the • TwiCopy

Stas Bekman

@stasbekman

+ Follow

Toolmaker. Software creator, optimizer and harmonizer.

Makes ML systems work and fly @ Snowflake.

ID: 1068360975898660864

linkhttps://stasosphere.com/machine-learning/ calendar_today30-11-2018 04:28:00

2,2K Tweet

8,8K Followers

282 Following

Stas Bekman

@stasbekman

10 months ago

semianalysis.com/p/the-memory-w… This is an important study as if you're not aware the new accelerators' efficiency keeps on going down because the compute speed up is happening faster than HBM speed up. So for example H100 is less efficient than H200, even though the compute is the

thumb_up_off_alt157

chat_bubble_outline4

repeat33

shareShare