Commit Graph

11 Commits

Author SHA1 Message Date
Nikolay Korovaiko
4aac42cc98 [LT] Add a new backend interface [DUP of the original] (#81662)
This is a dup of https://github.com/pytorch/pytorch/pull/76517 which is failing because Jiewen needs to resign the CLA.

Summary:
This commit introduces a new set of BackendImplInterface: GetDefaultDeviceOrdinal
and SetDefaultDeviceOrdinal. It allows backend to specify their own default
device, e.g, 1 for XLA and 0 for CUDA/CPU.

Test Plan:
./build/bin/test_lazy --gtest_filter=BackendDeviceTest.*

ghstack-source-id: b4adfef49253e51bffbbf40d356188a92c98994d
Pull Request resolved: https://github.com/pytorch/pytorch/pull/76517

Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/81662
Approved by: https://github.com/JackCaoG, https://github.com/wconstab
2022-07-19 01:15:22 +00:00
Michael Suo
30fb2c4aba [lint] autoformat test/cpp and torch/csrc
Let's have some fun.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/78828

Approved by: https://github.com/ezyang
2022-06-11 21:11:16 +00:00
Jiewen Tan
a28b132bc2 Revert D35860266: [pytorch][PR] Update torch::lazy::BackendDevice to have a new default ordinal
Test Plan: revert-hammer

Differential Revision:
D35860266 (f9d07ae644)

Original commit changeset: 554ebe16a068

Original Phabricator Diff: D35860266 (f9d07ae644)

fbshipit-source-id: 325c54aa2e87e51134115213352b3d33a81b7edf
(cherry picked from commit bbd74bf34a534d1b87aadff9790038e3dbbfa9c8)
2022-04-27 18:11:24 +00:00
Wonjoo Lee
f9d07ae644 Update torch::lazy::BackendDevice to have a new default ordinal (#76264)
Summary:
Fixes https://github.com/pytorch/xla/issues/3490. Updates `torch::lazy::BackendDevice` with changes below:

1. Remove the no-op string constructor.
2. Update default ordinal to `-1`.
3. Add a `is_valid` function to check if `ordinal` is valid/non-default (`ordinal >= 0`).

Pull Request resolved: https://github.com/pytorch/pytorch/pull/76264

Reviewed By: mrshenli

Differential Revision: D35860266

Pulled By: alanwaketan

fbshipit-source-id: 554ebe16a0683d37b00270c4f35163bf690bfe28
(cherry picked from commit b941d10e8545dfecfb34e4d5c24a29a1cc49bc4b)
2022-04-25 23:57:18 +00:00
Will Constable
3547f20872 Land remaining parts of Torchscript Lazy Tensor backend (#74111)
Summary:
Also enables bazel build to run lazy codegen.  Bazel (oss) build feeds off the same filelists as cmake/buck (build_variables.bzl), so enabling it is easier than keeping it disabled.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/74111

Test Plan: Run CI and verify test_lazy_ops is running via OSS cmake builds

Reviewed By: bdhirsh

Differential Revision: D34772403

fbshipit-source-id: 8a63f58b9536e6ac1be530667932176ef2549496
(cherry picked from commit e807ffb1918853d10b924fdc24f85ee5b1a39021)
2022-03-22 23:14:03 +00:00
Jiewen Tan
338eb1b2b3 [LTC] Export torch::lazy::GetBackendDevice() (#70963)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/70963

This commit exports torch::lazy::GetBackendDevice().

Test Plan: CI in the lazy_tensor_staging branch.

Reviewed By: wconstab

Differential Revision: D33468938

Pulled By: alanwaketan

fbshipit-source-id: f65599c9238bf6b4f4ffbd5194befdc267272831
2022-01-07 13:13:18 -08:00
Jiewen Tan
ab57f6d12c [LTC] Upstream utils to extract BackendDevice from at::Tensor (#70069)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/70069

This commit upstreams utils to extract BackendDevice from at::Tensor.

Test Plan: ./build/bin/test_lazy --gtest_filter=BackendDeviceTest.GetBackendDevice*

Reviewed By: samdow

Differential Revision: D33293160

Pulled By: alanwaketan

fbshipit-source-id: 78647239f90b4d04adce84ae6022b8983ad30c09
2021-12-23 12:42:03 -08:00
Michael Suo
795af1578c Revert D33172665: [LTC] Upstream utils to extract BackendDevice from at::Tensor
Test Plan: revert-hammer

Differential Revision:
D33172665 (121d067999)

Original commit changeset: b334ee358ea7

Original Phabricator Diff: D33172665 (121d067999)

fbshipit-source-id: 8bff43cddfc5d30483ec5cea8eff037aab9d1cfa
2021-12-22 21:12:49 -08:00
Jiewen Tan
121d067999 [LTC] Upstream utils to extract BackendDevice from at::Tensor (#70069)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/70069

This commit upstreams utils to extract BackendDevice from at::Tensor.

Test Plan: ./build/bin/test_lazy --gtest_filter=BackendDeviceTest.GetBackendDevice*

Reviewed By: wconstab

Differential Revision: D33172665

Pulled By: alanwaketan

fbshipit-source-id: b334ee358ea7b031bbffb0a16fa634715dba83f5
2021-12-22 18:15:45 -08:00
Jiewen Tan
e6c435bf96 [LTC] Upstream helpers for c10::Device <=> BackendDevice (#69064)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/69064

This commit upstreams helpers for converting a c10::Device to
BackendDevice and vice versa.

Test Plan: ./build/bin/test_lazy --gtest_filter=BackendDeviceTest.FromAten:BackendDeviceTest.ToAten

Reviewed By: wconstab

Differential Revision: D32732607

Pulled By: alanwaketan

fbshipit-source-id: 0dd233d37a4a30fc4b22dba322ddd85d4cb3635b
2021-12-01 12:15:32 -08:00
Jiewen Tan
6011c35a79 [LTC] Upstream class BackendDevice (#68027)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/68027

This commit upstreams class BackendDevice to the master, which is a backend
specific representation of the actual hardware, for instances, CPU, GPU, or
TPU.

This concept is important for backend like XLA where it needs to tell the
actual hardware type from the c10::DeviceType::Lazy virtual device during
both IR constructions and lowerings.

Test Plan: ./build/bin/test_lazy --gtest_filter=BackendDeviceTest.*

Reviewed By: wconstab

Differential Revision: D32261838

Pulled By: alanwaketan

fbshipit-source-id: 579c3fc5f9da7847c887a383c6047e8ecb9cc5bc
2021-11-10 07:05:43 -08:00