Skip to content
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.

Sign up for GitHub

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

Jump to bottom

"fix gpu init" #7528

Merged
merged 7 commits into from
Jan 31, 2018
Merged

"fix gpu init" #7528

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
7d8afc3
"fix gpu init"
dzhwinter Jan 15, 2018
1edd1c0
"set env variable default value for share gpu"
dzhwinter Jan 15, 2018
5a8b530
merge develop branch
dzhwinter Jan 23, 2018
d1a3ba9
"fix ci"
dzhwinter Jan 23, 2018
c17051e
"removed CUDA_VISIBLE_DEVICES default"
dzhwinter Jan 30, 2018
15ae79e
Merge remote-tracking branch 'origin/develop' into fix/initgpu
dzhwinter Jan 30, 2018
44c56cc
"removed"
dzhwinter Jan 31, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 11 additions & 4 deletions paddle/framework/init.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@ See the License for the specific language governing permissions and
limitations under the License. */
#include <string.h> // for strdup
#include <algorithm>
#include <stdexcept>
#include <string>

#include "paddle/framework/init.h"
Expand Down Expand Up @@ -46,17 +47,23 @@ void InitDevices() {

std::vector<platform::Place> places;
places.emplace_back(platform::CPUPlace());
int count = 0;

#ifdef PADDLE_WITH_CUDA
int count = platform::GetCUDADeviceCount();
for (int i = 0; i < count; ++i) {
places.emplace_back(platform::CUDAPlace(i));
try {
count = platform::GetCUDADeviceCount();
} catch (const std::exception &exp) {
LOG(WARNING) << "Compiled with WITH_GPU, but no GPU found in runtime.";
}
#else
LOG(WARNING)
<< "'GPU' is not supported, Please re-compile with WITH_GPU option";
<< "'CUDA' is not supported, Please re-compile with WITH_GPU option";
#endif

for (int i = 0; i < count; ++i) {
places.emplace_back(platform::CUDAPlace(i));
}

platform::DeviceContextPool::Init(places);
}

Expand Down
21 changes: 20 additions & 1 deletion paddle/framework/init_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,26 @@ TEST(InitDevices, CPU) {
using paddle::framework::InitDevices;
using paddle::platform::DeviceContextPool;

#ifndef PADDLE_WITH_CUDA
InitDevices();
DeviceContextPool& pool = DeviceContextPool::Instance();
ASSERT_GE(pool.size(), 1U);
ASSERT_EQ(pool.size(), 1U);
#endif
}

TEST(InitDevices, CUDA) {
using paddle::framework::InitDevices;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

#ifdef PADDLE_WITH_CUDA should be placed in line31.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This test is compiled in nv_test.
So #ifdef PADDLE_WITH_CUDA is no needed.

using paddle::platform::DeviceContextPool;

int count = 0;
try {
count = paddle::platform::GetCUDADeviceCount();
} catch (const std::exception& exp) {
}
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If #ifdef PADDLE_WITH_CUDA is placed in line 31, I think try ... catch is unnecessary.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done


#ifdef PADDLE_WITH_CUDA
InitDevices();
DeviceContextPool& pool = DeviceContextPool::Instance();
ASSERT_EQ(pool.size(), 1U + static_cast<unsigned>(count));
#endif
}