cuda - Caffe using GPU with NVidia Quadro 2200 -


i'm using deep learning framework caffe on ubuntu 14.04 machine. compiled cafe cpu_only option, i.e. disabled gpu , cuda usage. have nvidia quadro k2200 graphics card , cuda version 5.5.

i know if possible use caffe cuda enabled gpu. on nvidia page, written quadro k2200 has compute capability of 5.0. mean can use cuda versions release 5.0? when possible use caffe gpu-enabled quadro k2200, how can choose appropriate cuda version that?

cuda version not same thing compute capability. one, cuda current (7.5 prerelease), while cc @ 5.2. k2200 supports cc 5.0.

the difference:

cuda version means library/toolkit/sdk/etc version. should use highest 1 available.

compute capability gpu's capability perform instructions, etc. every cuda function has minimum cc requirement. when write cuda program, it's cc requirement maximum of requirements of features used.


that said, i've no idea caffe is, quick search shows require cc of 2.0, should go. cc 5.0 pretty recent, few things won't work on it.


Comments

Popular posts from this blog

html - Firefox flex bug applied to buttons? -

html - Missing border-right in select on Firefox -

c# - two queries in same method -