"fix gpu init" #7528
Merged
"fix gpu init" #7528
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
chengduoZH
reviewed
Jan 15, 2018
paddle/framework/init.cc
Outdated
| } | ||
| #else | ||
| LOG(WARNING) | ||
| << "'GPU' is not supported, Please re-compile with WITH_GPU option"; | ||
| << "'CUDA' is not supported, Please re-compile with WITH_CUDA option"; |
There was a problem hiding this comment.
Currently, CMakeLists.txt does not have WITH_CUDA option.
chengduoZH
reviewed
Jan 15, 2018
paddle/framework/init.cc
Outdated
| int count = platform::GetCUDADeviceCount(); | ||
| for (int i = 0; i < count; ++i) { | ||
| places.emplace_back(platform::CUDAPlace(i)); | ||
| } |
There was a problem hiding this comment.
The strict way to write is to put line54~56 outside try...catch.
But it's not a big problem here. Please feel free to change it or not.
chengduoZH
reviewed
Jan 23, 2018
python/paddle/v2/fluid/__init__.py
Outdated
| 'WARNING: CUDA_VISIBLE_DEVICES set to {0}, not empty . The computation ' | ||
| 'speed will not be optimized if you use multi-gpu. It will ' | ||
| 'fail if this PaddlePaddle binary is compiled without GPU option' | ||
| .format(gpu_devices), |
There was a problem hiding this comment.
- You want to remind users that currently fluid does not support multi-GPU training, right?
- If the user does not compile
PaddlePaddlewith GPU option, this warning will not show. SoIt will fail if this PaddlePaddle ...is unnecessary.
chengduoZH
reviewed
Jan 23, 2018
python/paddle/v2/fluid/__init__.py
Outdated
| file=sys.stderr) | ||
| else: | ||
| gpu_devices = "0" | ||
| os.environ['CUDA_VISIBLE_DEVICES'] = gpu_devices |
There was a problem hiding this comment.
The logic of line 99~110 maybe has some problem.
chengduoZH
reviewed
Jan 31, 2018
| } | ||
|
|
||
| TEST(InitDevices, CUDA) { | ||
| using paddle::framework::InitDevices; |
There was a problem hiding this comment.
#ifdef PADDLE_WITH_CUDA should be placed in line31.
There was a problem hiding this comment.
This test is compiled in nv_test.
So #ifdef PADDLE_WITH_CUDA is no needed.
paddle/framework/init_test.cc
Outdated
| try { | ||
| count = paddle::platform::GetCUDADeviceCount(); | ||
| } catch (const std::exception& exp) { | ||
| } |
There was a problem hiding this comment.
If #ifdef PADDLE_WITH_CUDA is placed in line 31, I think try ... catch is unnecessary.
This was referenced Jan 31, 2018
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
fix init gpu error.