Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

Published by onesixx on

http://it.plusblog.co.kr/221237740617
2020-04-28 03:47:38.631653: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

2020-04-28 03:47:38.638027: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2394260000 Hz

2020-04-28 03:47:38.640102: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55919dbed160 executing computations on platform Host. Devices:

2020-04-28 03:47:38.640130: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Host, Default Version

최신CPU들은 빠른 행렬연산을 도와주는 AVX(Advanced Vector Extensions), FMA( Fused multiply–add SSE 확장판, Streaming SIMD Extensions)같은 instruction(명령모음)을 제공하고, 이는 분석의 계산속도 향상에 도움이 된다.

해결1 – 무시

Error가 아닌 warning 수준의 메세지이므로 무시한다.
아래 해결책과 같은 적극적인 대응을 하면, 더 빠른 연산이 가능하지만 귀찮을땐 무시해도 상관없음.

해결2 – complie

AVX2나 FMA를 사용하여 complie되지 않는 Tensorflow binary에 대해서,
환경에 맞게 compile해서 사용.

Categories: DeepLearning

onesixx

Blog Owner

Subscribe
Notify of
guest

0 Comments
Inline Feedbacks
View all comments
0
Would love your thoughts, please comment.x
()
x