Hacker News Reader

1.6.0 does not appear to be installable via pip yet, though. (you can install tensorflow==1.6.0rc1 if you really want it now)

The inclusion of AVX instructions should give model training a notable speed boost at lower numbers of CPUs.

View on HN

Looks like it's on pypi now https://pypi.python.org/pypi/tensorflow-gpu

I thought the plan was to have common code for the generic, non-critical parts like graph handling, while operators would be shipped in multiple variants, keyed and loaded at runtime by CPUID. You can even delegate that to glibc if they are packaged as shared libraries and it's a recent enough version. Maybe in 1.7?

Do you have any evidence that the TF developers are going to do that?

I heard it from someone at Google at a public event, but there was no date attached to it. I should have asked.

Reply on news.ycombinator.com