Skip to content

Comment on Linode GPU Instancesparent

Comments

For traditional floating point workloads, the RTX 6000 will probably not be 3x faster. For workloads that can use the tensor ops (integer matrix multiply, basically), the RTX 6000 may be as much as 10-100x faster.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.