Skip to content

Implement pin_memory() as a NativeFunction #4094

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
Merged
soumith merged 2 commits into from
Dec 18, 2017
Merged

Implement pin_memory() as a NativeFunction #4094

soumith merged 2 commits into from
Dec 18, 2017

Conversation

@colesbury
Copy link
Member

@colesbury colesbury commented Dec 8, 2017 

This adds allocators as a concept in ATen that extends deleters. An
allocator is a subclass of at::Allocator that implements the virtual
methods:

  virtual void* allocate(size_t n);
  virutal void deallocate(void* ptr);

A tensor created with a custom allocator can be resized, unlike a tensor
with a custom deleter.

cc @zdevito

@pytorchbot
Copy link
Collaborator

pytorchbot commented Dec 8, 2017

@colesbury, thanks for your PR! We identified @zdevito to be a potential reviewer.

apaszke
apaszke reviewed Dec 9, 2017
View reviewed changes
aten/src/ATen/Allocator.h Outdated

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

@colesbury
Implement pin_memory() as a NativeFunction
d14d14d 
This adds allocators as a concept in ATen that extends deleters. An
allocator is a subclass of at::Allocator that implements the virtual
methods:

  virtual void* allocate(size_t n);
  virutal void deallocate(void* ptr);

A tensor created with a custom allocator can be resized, unlike a tensor
with a custom deleter.
gchanan
gchanan reviewed Dec 15, 2017
View reviewed changes
aten/src/ATen/templates/StorageDerived.cpp
nullptr,
nullptr,
};
static cudaError_t wrapped_alloc(void * ctx, void** result, size_t size, cudaStream_t stream) {

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

aten/src/ATen/templates/StorageDerived.cpp
nullptr,
call_deleter,
};
static void* wrapped_alloc(void * ctx, ptrdiff_t size) {

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

This comment was marked as off-topic.

Sign in to view

aten/src/ATen/Allocator.h Outdated

This comment was marked as off-topic.

Sign in to view

@soumith soumith merged commit 2c71b67 into pytorch:master Dec 18, 2017
@colesbury colesbury deleted the pin_memory branch December 18, 2017 18:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ezyang ezyang ezyang left review comments

@gchanan gchanan gchanan approved these changes

+1 more reviewer

@apaszke apaszke apaszke left review comments

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@colesbury @pytorchbot @ezyang @gchanan @apaszke @soumith