Don't initialize the current device in CUDAGenerator::CUDAGenerator #7392
Conversation
This file contains hidden or 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
colesbury
requested review from
apaszke,
ezyang,
gchanan,
soumith and
zdevito
as code owners
ezyang
approved these changes
May 8, 2018
Contributor
|
Can we add a test that checks for this unwanted init? It's not the first time we get this regression. |
Previously, CUDAGenerator::CUDAGenerator would initialize the random number generator on the current device. This would usually be device 0. This is undesirable because initialize the CUDA context allocates a few 100 MBs due to all the kernels in libTHC.so. This avoids the unecessary call to THCRandom_getGenerator() in the CUDAGenerator constructor. Fixes pytorch#7320
Member
Author
onnxbot
added a commit
to onnxbot/onnx-fb-universe
that referenced
this pull request
May 13, 2018
…e in CUDAGenerator::CUDAGenerator (pytorch/pytorch#7392) pytorch/pytorch@976b1d5
Collaborator
|
@colesbury unfortunately cuda runtime API ignore the existence of contexts alltogether, so there's no way to query context using them. |
weiyangfb
pushed a commit
to weiyangfb/pytorch
that referenced
this pull request
Jun 11, 2018
…ytorch#7392) Previously, CUDAGenerator::CUDAGenerator would initialize the random number generator on the current device. This would usually be device 0. This is undesirable because initialize the CUDA context allocates a few 100 MBs due to all the kernels in libTHC.so. This avoids the unecessary call to THCRandom_getGenerator() in the CUDAGenerator constructor. Fixes pytorch#7320 Previously, CUDAGenerator::CUDAGenerator would initialize the random number generator on the current device. This would usually be device 0. This is undesirable because initialize the CUDA context allocates a few 100 MBs due to all the kernels in libTHC.so. This avoids the unecessary call to THCRandom_getGenerator() in the CUDAGenerator constructor. Fixes pytorch#7320 * Fix call to get THCState
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
Previously, CUDAGenerator::CUDAGenerator would initialize the random
number generator on the current device. This would usually be device 0.
This is undesirable because initialize the CUDA context allocates a few
100 MBs due to all the kernels in libTHC.so.
This avoids the unecessary call to THCRandom_getGenerator() in the
CUDAGenerator constructor.
Fixes #7320