Re: ubuntu 18.10
"J. Andrew Rogers" <[email protected]>
| Newsgroups | gmane.culture.people.rohit-khare |
|---|---|
| Message-ID | <[email protected]> |
Ah, this makes sense, that microarchitecture is pre-Nehalem. Most builds assume you at least have SSE4.2, and performance-optimized software often assumes Haswell or later. People running performance intensive software on ancient microarchitectures are an edge case. > On Apr 11, 2019, at 3:28 PM, Gregory Alan Bolcer <[email protected]> wrote: > > I think the only reason it's on my system is that the default binary for Tensorflow 1.3 needed to be compiled against it. The only reason I had to compile Tensorflow from sources is that my work machine has 2 x 4-core Xeon X5482's which apparently are missing some weird SSE instruction set that all x86 processors now have that mine doesn't. > > Greg > > On 4/11/2019 2:59 PM, J. Andrew Rogers wrote: >> Boost is a disaster of a library with serious versioning problems that unfortunately is commonly used in many legacy C++ code bases. It largely serves no purpose in the modern C++ world but I expect it to continue inflicting pain for many years to come. >>> On Apr 11, 2019, at 1:55 PM, Gregory Alan Bolcer <[email protected]> wrote: >>> >>> After three weeks of apt repo hell, I finally was able to figure out which >>> library (libboost) was causing issues with preventing a completely >>> automated upgrade--that and cuda/nvidia geforce driver had to be installed >>> from a local download instead of the repository. >>> >>> Anyone else? >>> >>> Greg >>> >>> -- >>> [email protected], http://bolcer.org, c: +1.714.928.5476 >>> _______________________________________________ >>> FoRK mailing list >>> http://xent.com/mailman/listinfo/fork >> _______________________________________________ >> FoRK mailing list >> http://xent.com/mailman/listinfo/fork > _______________________________________________ > FoRK mailing list > http://xent.com/mailman/listinfo/fork _______________________________________________ FoRK mailing list http://xent.com/mailman/listinfo/fork