-
Notifications
You must be signed in to change notification settings - Fork 5.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add more tests and fix bugs for cudnn_norm_conv_test and cudnn_bn_and_relu_test #36314
Conversation
Thanks for your contribution! |
88bcb60
to
2c75558
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. 有一处小疑问,建议后续PR确认下。
@@ -369,10 +381,25 @@ class CudnnNormConvolutionTester { | |||
TensorCopySync(filter_grad, platform::CPUPlace(), cpu_filter_grad); | |||
} | |||
|
|||
bool Support(const platform::CUDADeviceContext &ctx) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
函数名一般用IsSupport
这种形式。
if ((kernel_size_ == 3) || ((kernel_size_ == 1) && (stride_ == 1))) { | ||
return true; | ||
} | ||
} else if (ctx.GetComputeCapability() > 70) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里可能用>=80
比较好,因为还有75
计算能力的GPU卡,不确定是否支持呢。文档中怎么说?另外,这个到底是跟GPU型号有关,还是跟cudnn版本有关呢?
这个检查建议挪到CudnnNormConvolution
的实现里面,并且加一些注释。
PR types
Others
PR changes
Others
Describe
为cudnn_fusion_op增加了单测,更早PR见:
#35557
#35955
#36168
其中: