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

Add lowering for aten.to.device #1107

Merged
merged 1 commit into from
Aug 10, 2022
Merged

Add lowering for aten.to.device #1107

merged 1 commit into from
Aug 10, 2022

Conversation

gpetters94
Copy link
Collaborator

As per #1075

@gpetters94 gpetters94 requested review from ramiro050 and pashu123 July 25, 2022 17:45
pashu123
pashu123 approved these changes Jul 26, 2022
View reviewed changes
Copy link
Member

@pashu123 pashu123 left a comment

Choose a reason for hiding this comment

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

LGTM.

@pashu123
Copy link
Member

@gpetters94 Could you add an e2e test case?

vivekkhandelwal1
vivekkhandelwal1 reviewed Aug 5, 2022
View reviewed changes
Copy link
Collaborator

@vivekkhandelwal1 vivekkhandelwal1 left a comment •
edited
Loading

Choose a reason for hiding this comment

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

Is the lowering correct? Is it doing what the op is supposed to do? @pashu123 @gpetters94

@pashu123
Copy link
Member

pashu123 commented Aug 5, 2022

We can just fold the aten.to.device op. We don't have distributed/device semantics here. Pinging @ramiro050.

@gpetters94
Copy link
Collaborator Author

I'm not sure if I can add a test for this op. Its name should be to.device but it that's not valid Python syntax, so I'm not sure what I should be calling in the test.

@pashu123
Copy link
Member

pashu123 commented Aug 8, 2022

Please squash all your commits into one.

@ramiro050
Copy link
Collaborator

I'm not sure if I can add a test for this op. Its name should be to.device but it that's not valid Python syntax, so I'm not sure what I should be calling in the test.

You need to specify enough arguments so that the right variant of .to gets used. If you do torch.ops.aten.to(input, device='cpu', dtype=torch.float, non_blocking=False), you should get to.device generated

@ramiro050
Copy link
Collaborator

We can just fold the aten.to.device op. We don't have distributed/device semantics here. Pinging @ramiro050.

This op is a bit strange, because it also has the ability of changing the dtype of the tensor, so we can't just fold it away. An option would be to use a canonicalization pattern, but the pattern would have to be conditional on the input being a value-semantics tensor (see #1075 (comment)), so that it does not get applied too early. I personally think it is simpler to just do a decomposition

@gpetters94 gpetters94 force-pushed the device branch 4 times, most recently from bd9a278 to a6f38fc Compare August 8, 2022 21:46
@pashu123
Copy link
Member

@gpetters94 Please rebase and push.

@gpetters94 gpetters94 force-pushed the device branch 2 times, most recently from 208f567 to 88e48cf Compare August 10, 2022 17:35
@gpetters94 gpetters94 merged commit 79b9cf9 into llvm:main Aug 10, 2022
@tanyokwok tanyokwok mentioned this pull request Sep 21, 2022
Closed
qedawkins pushed a commit to nod-ai/torch-mlir that referenced this pull request Oct 3, 2022
Add -march option to onnx-mlir (llvm#1107)
2dfbc22 
Signed-off-by: Ettore Tiotto <[email protected]>
qedawkins pushed a commit to nod-ai/torch-mlir that referenced this pull request Oct 3, 2022
@YW-Ethan @AlexandreEichenberger
@sstamenova @cjvolzka @gongsu832 @chentong319 @tungld @manbearian
Add verifier for concat op (llvm#1099)
334f3de 
* Add check-onnx-backend to Mac CI. (llvm#1069)

* Add check-onnx-backend to Mac CI.

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Additional Docker help and split README for easier reading (llvm#1084)

* initial docker documentation

Signed-off-by: Alexandre Eichenberger <[email protected]>

* split README with no redundant place for info

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* respond to suggestions

Signed-off-by: Alexandre Eichenberger <[email protected]>

* specify that onnx-mlir.py script generates only code suitable to be exec in Linux and/or Docker env

Signed-off-by: Alexandre Eichenberger <[email protected]>

* fix checkdocs

Signed-off-by: Alexandre Eichenberger <[email protected]>

* responded to review suggestion on onnx-mlir --help

Signed-off-by: Alexandre Eichenberger <[email protected]>

* use ONNX-MLIR everywhere

Signed-off-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* add verify for concat

Signed-off-by: Ethan Wang <[email protected]>

* check all inputs

Signed-off-by: Ethan Wang <[email protected]>

* Support filtering out lit tests based on targets (llvm#1087)

Currently we ignore what targets llvm was built for in the lit tests, but recent changes to onnx-mlir explicitly initialize the available targets.
This makes the corresponding change to the lit configuration, so that we can filter out the lit tests based on the available targets.

Signed-off-by: Stella Stamenova <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Switch URLs to use main instead of master (llvm#1094)

Signed-off-by: Charles Volzka <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Fix MacOS build badge (llvm#1092)

Signed-off-by: Gong Su <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* onnx-mlir.py warning about binary output (.so and .jar) (llvm#1090)

not directly usable if host is not Linux

Signed-off-by: Gong Su <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Make the doc example obey ONNX_MLIR_BUILD_TESTS (llvm#1083)

* Make the doc example obey ONNX_MLIR_BUILD_TESTS

Currently, ONNX_MLIR_BUILD_TESTS controls EXCLUDE_FROM_ALL, however, the targets added through add_executable will always build. We follow the llvm pattern and explicitly set EXCLUDE_FROM_ALL in the add_onnx_mlir_executable function if it is set for the directory, so that add_executable targets don't always build.

Signed-off-by: Stella Stamenova <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Explicitly install into lib on all systems (llvm#1088)

Signed-off-by: Gong Su <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* add check (llvm#1098)

Signed-off-by: Tong Chen <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* fix typos and add ssh-client to dockerfile (llvm#1096)

* fix typos and add ssh-client to dockerfile

Signed-off-by: Ethan Wang <[email protected]>

* sync doc and script

Signed-off-by: Ethan Wang <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Emit print statement only when the verbose option is in effect. (llvm#1097)

Signed-off-by: Ettore Tiotto <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* format & refine code by request

Signed-off-by: Ethan Wang <[email protected]>

* Support older versions 6, 11, 12 for Clip Op (llvm#1100)

Signed-off-by: Tung D. Le <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* using front to get first input

Signed-off-by: Ethan Wang <[email protected]>

* add 3 lit test for concat  verifier

Signed-off-by: Ethan Wang <[email protected]>

* add newline

Signed-off-by: Ethan Wang <[email protected]>

* Add check-onnx-backend to Mac CI. (llvm#1069)

* Add check-onnx-backend to Mac CI.

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Additional Docker help and split README for easier reading (llvm#1084)

* initial docker documentation

Signed-off-by: Alexandre Eichenberger <[email protected]>

* split README with no redundant place for info

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* update

Signed-off-by: Alexandre Eichenberger <[email protected]>

* respond to suggestions

Signed-off-by: Alexandre Eichenberger <[email protected]>

* specify that onnx-mlir.py script generates only code suitable to be exec in Linux and/or Docker env

Signed-off-by: Alexandre Eichenberger <[email protected]>

* fix checkdocs

Signed-off-by: Alexandre Eichenberger <[email protected]>

* responded to review suggestion on onnx-mlir --help

Signed-off-by: Alexandre Eichenberger <[email protected]>

* use ONNX-MLIR everywhere

Signed-off-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Switch URLs to use main instead of master (llvm#1094)

Signed-off-by: Charles Volzka <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Fix MacOS build badge (llvm#1092)

Signed-off-by: Gong Su <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* fix typos and add ssh-client to dockerfile (llvm#1096)

* fix typos and add ssh-client to dockerfile

Signed-off-by: Ethan Wang <[email protected]>

* sync doc and script

Signed-off-by: Ethan Wang <[email protected]>

Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Update document (llvm#1077)

* create

Signed-off-by: Tong Chen <[email protected]>

* delete HowTOAddAnOperation.md

Signed-off-by: Tong Chen <[email protected]>

* modify testing

Signed-off-by: Tong Chen <[email protected]>

* create

Signed-off-by: Tong Chen <[email protected]>

* delete HowTOAddAnOperation.md

Signed-off-by: Tong Chen <[email protected]>

* modify testing

Signed-off-by: Tong Chen <[email protected]>

* fix

Signed-off-by: Tong Chen <[email protected]>

* create

Signed-off-by: Tong Chen <[email protected]>

* add comment

Signed-off-by: Tong Chen <[email protected]>

* delete HowTOAddAnOperation.md

Signed-off-by: Tong Chen <[email protected]>

* modify testing

Signed-off-by: Tong Chen <[email protected]>

* fix

Signed-off-by: Tong Chen <[email protected]>

* create

Signed-off-by: Tong Chen <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Update LLVM level (llvm#1095)

* Update LLVM level to 700997a

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Pass a type converter to all ONNX operations. (llvm#1102)

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Nuke KrnlDummyCastOp now that we use MLIR's UnrealizedConversionCastOp (llvm#1103)

* Nuke KrnlDummyCastOp now that we use MLIR's UnrealizedConversionCastOp

Signed-off-by: Ettore Tiotto <[email protected]>

* Remove a dependency in src/Dialect/Krnl/CMakeList.txt.  Regenerate docs via 'ninja onnx-mlir-docs'.

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Add --emitObj option to onnx-mlir (llvm#1104)

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* fix warnings (llvm#1093)

Signed-off-by: Ian Bearman <[email protected]>

Co-authored-by: Stella Stamenova <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Add -march option to onnx-mlir (llvm#1107)

Signed-off-by: Ettore Tiotto <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Fix Doc spelling and broken links, removed warnings about using main (llvm#1106)

* removed warning about main vs master in CONTRIBUTING, fixed links and spelling mistakes

Signed-off-by: Alexandre Eichenberger <[email protected]>
Signed-off-by: Ethan Wang <[email protected]>

* Update BuildONNX.md

Signed-off-by: Ethan Wang <[email protected]>

Co-authored-by: Ettore Tiotto <[email protected]>
Co-authored-by: Alexandre Eichenberger <[email protected]>
Co-authored-by: Stella Stamenova <[email protected]>
Co-authored-by: Charles Volzka <[email protected]>
Co-authored-by: gongsu832 <[email protected]>
Co-authored-by: chentong319 <[email protected]>
Co-authored-by: Tung D. Le <[email protected]>
Co-authored-by: Ian Bearman <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@vivekkhandelwal1 vivekkhandelwal1 vivekkhandelwal1 left review comments

@ramiro050 ramiro050 ramiro050 left review comments

@pashu123 pashu123 pashu123 approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@gpetters94 @pashu123 @ramiro050 @vivekkhandelwal1