undefined | Better HN

0 pointsjohndough4y ago0 comments

Often the limiting factor is memory bandwidth instead of raw FLOPS, so dealing with 4 times larger data types (FP64 vs FP16) is a disadvantage.

0 comments

1 comments · 1 top-level

to clarify: I am comparing FP16 performance, which both the GPU and AMX have native support for.

FP64 is also supported by AMX, making it quite an impressive region of silicon.

j / k navigate · click thread line to collapse