[not ready for review yet] torch.compile support for parseSemiStructuredTensor #104974
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/104974
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Unrelated FailureAs of commit bb20c9b with merge base 7921243 ( NEW FAILURE - The following job has failed:
FLAKY - The following job failed but was likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Jul 11, 2023
Closed
github-actions
bot
requested review from
SherlockNoMad,
albanD,
antoniojkim,
ezyang,
jbschlosser,
miladm,
voznesenskym and
wconstab
bdhirsh
removed request for
SherlockNoMad,
albanD,
antoniojkim,
ezyang,
jbschlosser,
miladm,
voznesenskym and
wconstab
cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
This was referenced Jul 14, 2023
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
bdhirsh
force-pushed
the
gh/bdhirsh/430/head
branch
from
e755cfa to
7aa167d
Compare
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
…SemiStructuredTensor" cc jcaip, this is an E2E test of compiling a small model with a `SparseSemiStructuredTensor` subclass tensor used as one of the parameters. The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it. A few things to note: (1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both). (1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode. [ghstack-poisoned]
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
cc @jcaip, this is an E2E test of compiling a small model with a
SparseSemiStructuredTensorsubclass tensor used as one of the parameters.The generated inductor code looks like this (P788647425): you can see that the subclass desugars the matmul into a sparse_mm() + contiguous() call, and inductor is able to fuse the contiguous() call into the relu() that follows it.
A few things to note:
(1) the test actually... fails. I haven't figured out why, but the results don't match the non-sparse version. FWIW, the code that inductor outputs looks reasonable, so it's not immediately clear if it's a compile() bug, or something to do with sparsity giving less accurate results (or both).
(1) inference mode is a bit broken with torch.compile still: in the test, i needed to make sure that the model was instantiated, compiled and run all inside of inference_mode.
Stack from ghstack (oldest at bottom):