Move IValues from stack into kernels #19783
Closed
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
Differential Revision: D15091973 Differential Version: 80762472
pytorchbot
added
the
module: internals
dzhulgakov
approved these changes
Apr 30, 2019
dzhulgakov
reviewed
Apr 30, 2019
| static void call_(std::tuple<OutputTypes...>&& output, Stack* stack, guts::index_sequence<indices...>) { | ||
| (void)(stack); // when sizeof...(indices) == 0, this argument would be unused and we have to silence the compiler warning. | ||
| // iterate over all outputs and push them | ||
| (void)std::initializer_list<int>{( |
There was a problem hiding this comment.
why did we have this initializer_list in the first place?
There was a problem hiding this comment.
It is a trick to expand a parameter list into a series of statements. It was calling torch::jit::push(stack, output) on each output. But, actually, inside the implementation of torch::jit::push, there is already an initializer_list doing exactly the same, so instead of calling it on each output individually, we can just pass in torch::jit::push(stack, output, ...) and have that initializer_list take care of it.
|
This pull request has been merged in cb5442b. |
zdevito
pushed a commit
to zdevito/ATen
that referenced
this pull request
May 1, 2019
Summary: Pull Request resolved: pytorch/pytorch#19783 Previously, the IValues were copied into the kernel arguments, which caused a refcount bump if Tensor was taken by value. Now, a kernel can take Tensor by value without any refcount bump because it is moved in. Reviewed By: dzhulgakov Differential Revision: D15091973 fbshipit-source-id: 4c5ff2e3ee86f5934cc84191697f7dbc9c3ee345
zhangguanheng66
pushed a commit
to zhangguanheng66/pytorch
that referenced
this pull request
May 6, 2019
Summary: Pull Request resolved: pytorch#19783 Previously, the IValues were copied into the kernel arguments, which caused a refcount bump if Tensor was taken by value. Now, a kernel can take Tensor by value without any refcount bump because it is moved in. Reviewed By: dzhulgakov Differential Revision: D15091973 fbshipit-source-id: 4c5ff2e3ee86f5934cc84191697f7dbc9c3ee345
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.
Stack:
:black_circle: #19784 The deprecated API allows std::vector arguments 💛
:white_circle: #19783 Move IValues from stack into kernels 💛
Previously, the IValues were copied into the kernel arguments, which caused a refcount bump if Tensor was taken by value.
Now, a kernel can take Tensor by value without any refcount bump because it is moved in.
Differential Revision: D15091973