tensorflow 遇到的问题 The TensorFlow library wasn't compiled to use SSE instructions, but these are avail

最新推荐文章于 2023-07-26 10:39:15 发布

原创最新推荐文章于 2023-07-26 10:39:15 发布 · 1.9w 阅读

14 ·

CC 4.0 BY-SA版权

文章标签：

#tensorflow #问题 #SSE

TensorFlow 专栏收录该内容

30 篇文章

订阅专栏

本文解析了TensorFlow中关于SSE指令未被利用的警告信息，并解释了该警告出现的原因及对CPU计算速度的影响。文章指出，此警告源于通过pip安装而非编译安装TensorFlow库，且仅影响CPU计算速度。

____tz_zs

The TensorFlow library wasn't compiled to use SSE instructions, but these are available on your machine and could speed up CPU computations.

W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE instructions, but these are available on your machine and could speed up CPU computations.

大概意思是

警告：你的机器是支持使用SSE指令的，使用SSE可以加速你的CPU计算。但是，你的Tensorflow库没有编译，所以用不了

遇到这个警告的原因是，你是使用例如 "pip install tensorflow" 这种方法安装的Tensorflow，而不是从源码编译安装的。

这只是个警告，只是影响cpu计算速度而已。使用或者不使用SSE你的程序都是可以运行的。

此外，如果你是使用的gpu版本，在gpu上的运行是不受SSE影响的。

stackoverflow：https://stackoverflow.com/questions/43134753/tensorflow-wasnt-compiled-to-use-sse-etc-instructions-but-these-are-availab

cpu版本的警告

2017-07-22 16:25:29.480499: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.480818: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.481114: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.481428: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.481753: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.482047: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.482375: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:25:29.482657: W c:\tf_jenkins\home\workspace\release-win\m\windows\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.

gpu版本的警告

2017-07-22 16:21:50.144021: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.144323: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.144621: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.144905: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.145183: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.145469: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.145750: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:50.146035: W c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
2017-07-22 16:21:51.049124: I c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:940] Found device 0 with properties: 
name: GeForce GTX 965M
major: 5 minor: 2 memoryClockRate (GHz) 1.15
pciBusID 0000:01:00.0
Total memory: 2.00GiB
Free memory: 1.64GiB
2017-07-22 16:21:51.049960: I c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:961] DMA: 0 
2017-07-22 16:21:51.050323: I c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:971] 0:   Y 
2017-07-22 16:21:51.050691: I c:\tf_jenkins\home\workspace\release-win\m\windows-gpu\py\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:1030] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 965M, pci bus id: 0000:01:00.0)