Commit Graph

2510 Commits (484ee82c28acc529781f24b6345385a52590c7fb)
 

Author SHA1 Message Date
dan 484ee82c28 view hack 2024-02-07 03:56:00 +00:00
saienduri bfcf93ea21
Rename torch_mlir.compile APIs and introduce FX based analogs (#2842)
Link to related RFC:
https://discourse.llvm.org/t/rfc-rename-torch-mlir-compile-apis-and-introduce-fx-based-analogs/76646
This commit updates the documentation, tests, CMake files, and API for
the proposed changes in the RFC. There is a new torch_mlir/fx.py for
user level APIs related to importing modules and a corresponding test
for this path can be found at test/python/fx_importer/basic_test.py.

---------

Co-authored-by: MaheshRavishankar <mravisha@amd.com>
2024-02-06 19:07:59 -08:00
Xida Ren (Cedar) cc06391630
AtenSortOp Folder (#2864)
A chunk off

https://github.com/llvm/torch-mlir/pull/2856
https://github.com/llvm/torch-mlir/pull/2860

---------

Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
Co-authored-by: Rob Suderman <rob.suderman@gmail.com>
2024-02-06 21:12:12 +00:00
Daniel Garvey faf7d4aaa5
[fx_importer] Add support for 0D tensors (#2870)
Adds an escape hatch from creating a DenseResourceElementsAttr for
single value tensors into DenseElementsAttr.

For 0d or 1element, splats are better as DenseElementsAttr. Don't use
DenseResourceElementsAttr for it
2024-02-06 00:19:31 -06:00
Dave Liddell 1cb14f6879
Rob's atenTensor folder (#2867)
If a tensor is initialized by a list with a single constant integer,
this folder turns it into a torch.vtensor.literal

---------

Co-authored-by: Dave Liddell <dliddell@xilinx.com>
2024-02-05 17:10:42 -08:00
Rob Suderman 041a54ae0c
[torch] Supporting `torch.aten.mul.float` lowering to `arith` (#2833)
Simple missing scalar operation for multiply floats was missing.
2024-02-05 16:23:04 -08:00
Rob Suderman e3faef5224
[onnx] Convert `onnx.QLinearConv` to `torch` (#2851)
Leaning on the QDQ functionality in torch we can support the QLinearConv
operation by piggybacking through `torch.Convolution`. This includes
some changes such as allowing the `onnx` rewriter to run recursively.
Doing so allows `QLinearConv` to decopmose to `onnx.Convolution` which
is then lowered to `torch`.
2024-02-05 16:09:41 -08:00
Rob Suderman cb52c4b3cc
[onnx] Fix `onnx-to-torch` lowering for flatten shape (#2834)
The existing `flatten` lowering did not define what the intermediate
shape was. This could result in failures to lower further to linalg as
the intermediate shape was unknown. Added a shape refinement section.
2024-02-05 14:23:46 -08:00
Xida Ren (Cedar) b3a56c0711
Update add_ops to mention llvm-project/mlir/utils/generate-test-checks.py (#2862) 2024-02-05 12:13:43 -08:00
Gaurav Shukla f4562a8eaa
[ONNX] Fix the lowering of onnx.expand op (#2861)
Signed-off-by: Gaurav Shukla <gauravshukla789@gmail.com>
2024-02-05 23:46:58 +05:30
Aart Bik d1cd117998
[torch-mlir] remove trailing whitespace from md documentation (#2853) 2024-02-02 11:02:53 -08:00
Xida Ren (Cedar) 24b8c8672a
[torch] Add folders for `torch.fill`, `torch.ones`, `torch.zeros` and `aten.getItem` (#2849)
So that the CumSum Op in OPT can get the constant that it requires to be lowered to TMTensor

---------

Co-authored-by: Rob Suderman <rob.suderman@gmail.com>
Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
2024-02-02 10:46:33 -08:00
Ben Vanik 962d514308
Fixing implicit double->float conversion warning. (#2850)
`[build]
D:\Dev\iree\third_party\torch-mlir\lib\Conversion\TorchOnnxToTorch\DefaultDomainGtoP.cpp(734):
warning C4305: 'argument': truncation from 'double' to 'float'`
2024-02-01 22:02:44 -08:00
Rob Suderman 29baa813bd
[onnx] Fix `pool` lowering for non-symmetric padding (#2837)
`torch` requires that padding be symmetric for pooling operations. To
support non-symmetric pad we need to separately materialize out the
padding operation.

---------

Co-authored-by: James Newling <james.newling@gmail.com>
2024-02-01 14:35:21 -08:00
Sambhav Jain c7d7d7f004
[Bazel] Add TorchToTensor dep to TorchMLIRTorchConversionPasses (#2847)
Fixes bazel build error:
```
ERROR: /root/.cache/bazel/_bazel_root/b89349c08f7224396763d14fe35cba11/external/torch-mlir/BUILD.bazel:547:11: Compiling lib/Dialect/TorchConversion/Transforms/Passes.cpp failed: (Exit 1): clang failed: error executing command /usr/lib/llvm-16/bin/clang -U_FORTIFY_SOURCE -fstack-protector -Wall -Wthread-safety -Wself-assign -Wunused-but-set-parameter -Wno-free-nonheap-object -fcolor-diagnostics -fno-omit-frame-pointer ... (remaining 224 arguments skipped)

Use --sandbox_debug to see verbose messages from the sandbox and retain the sandbox build root for debugging
external/torch-mlir/lib/Dialect/TorchConversion/Transforms/Passes.cpp:23:10: fatal error: 'torch-mlir/Conversion/TorchToTensor/TorchToTensor.h' file not found
#include "torch-mlir/Conversion/TorchToTensor/TorchToTensor.h"
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
Target @torch-mlir//:torch-mlir-opt failed to build
```

Bazel CI:
https://github.com/sjain-stanford/torch-mlir/actions/runs/7735724133/job/21091865352
2024-01-31 22:07:06 -08:00
Dave Liddell 04be6ba773
Make the onnx importer more robust for internal/external and large models (#2794)
Fix for https://github.com/llvm/torch-mlir/issues/2765

The onnx docs say that you can't do shape inference using the in-memory
API for models > 2 GB. This fix replaces that API with the file-based
API. Since the new API generates an intermediate file, also added a
--keep switch to keep that file, which I delete by default.

---------

Co-authored-by: Dave Liddell <dliddell@xilinx.com>
2024-01-31 21:58:43 -08:00
Rob Suderman 34f6948533
[torch] Support `!countIncludePad` when unpadded for average pool (#2836)
We do not support average pool when `countIncludePad is set to false.
However if the input is unpadded then the setting of the boolean is
unneeded. Extended use by checking if padding is zero before rejecting
the lowering.
2024-01-31 15:09:36 -08:00
Rob Suderman 0114a570e3
[torch] Support lowering `torch.item` to `tensor.extract` (#2835)
Extracting scalar values from tensors can be implemented via a lowering
to tensor.extract.
2024-01-31 15:09:12 -08:00
Sambhav Jain 8a17c98b74
Bump stablehlo to openxla/stablehlo@fd52182f76 (#2821)
With the recent LLVM integrate and changes from
https://github.com/llvm/llvm-project/pull/78260, we hit this build error
in Stablehlo (which is quite old).
```
external/stablehlo/stablehlo/transforms/StablehloRefineShapes.cpp:1020:14: error: no member named 'startRootUpdate' in 'mlir::PatternRewriter'
    rewriter.startRootUpdate(op);
    ~~~~~~~~ ^
external/stablehlo/stablehlo/transforms/StablehloRefineShapes.cpp:1026:16: error: no member named 'finalizeRootUpdate' in 'mlir::PatternRewriter'
      rewriter.finalizeRootUpdate(op);
      ~~~~~~~~ ^
external/stablehlo/stablehlo/transforms/StablehloRefineShapes.cpp:1029:16: error: no member named 'cancelRootUpdate' in 'mlir::PatternRewriter'
      rewriter.cancelRootUpdate(op);
      ~~~~~~~~ ^
external/stablehlo/stablehlo/transforms/StablehloRefineShapes.cpp:1108:14: error: no member named 'updateRootInPlace' in 'mlir::PatternRewriter'
    rewriter.updateRootInPlace(op->getParentOp(), [&]() { return; });
    ~~~~~~~~ ^
4 errors generated.
Target @torch-mlir//:torch-mlir-opt failed to build
```

I'm still puzzled as to how this didn't fail with the CMake merge gating
CI (do we not test Stablehlo builds/tests?). In any case, bumping our
submodule to https://github.com/openxla/stablehlo/pull/1918 fixes it.

It exposes a new failing lit test in TorchToStablehlo though, that I
have looped stablehlo developers into
([here](https://discord.com/channels/999073994483433573/999074539138990131/1201235845391331419)).
```
bazel run @torch-mlir//test/Conversion:TorchToStablehlo/scatter.mlir.test 

...external/torch-mlir/test/Conversion/TorchToStablehlo/scatter.mlir
within split at <stdin>:1 offset :33:8: error: unexpected error: Expects non-empty reduction block for type inference                                                                               
  %0 = torch.aten.scatter.src %arg0, %int0, %arg1, %arg2 : !torch.vtensor<[?,?],si64>, !torch.int, !torch.vtensor<[?,?],si64>, !torch.vtensor<[?,?],si64> -> !torch.vtensor<[?,?],si64>             
       ^                                                                                                                                                                                            
LLVM ERROR: Failed to infer result type(s).               
```

Bazel CI:
https://github.com/sjain-stanford/torch-mlir/actions/runs/7732673480/job/21083102228
2024-01-31 14:21:17 -08:00
Rob Suderman 54e258792c
[onnx] Import `onnx` constants as `onnx.Constant` instead of literals (#2831)
To handle the conversion from raw bytes to `DenseElementsAttr` we need
to handle the endianness conversion during `torch-onnx-to-torch`.
Therefore when importing `onnx.Constant` it is better to represent using
the `onnx` constant operation so that only one location requires the
endianness correction.
2024-01-31 11:41:06 -08:00
Rob Suderman 3500523f75
[onnx] Convert resources to denseattr for `onnx.constant` to `torch` (#2830)
`onnx` explicitly specifies that `raw_data` is stored in `little-endian`
layout. While converting
to `torch` we need to convert from a known endian format to an internal
format of consistent
layout. This means endianness must be correct during the import of
`onnx.Constant`.

---------

Co-authored-by: Xida Ren (Cedar) <cedar.ren@gmail.com>
2024-01-31 11:40:53 -08:00
Ilija Kalinić 54ef18c556
Implement lowering of torch.aten.lerp.Scalar (#2773)
Closes nod-ai/SHARK-Turbine#356
2024-01-31 09:39:38 -08:00
Stella Laurenzo 7301aa80fd
Enable -Werror in lib/ and LTC. (#2841)
Required some massaging of LTC to make it warning clean, and I had to
manually disable some warnings on the generated source files (which we
don't control).

The project is warning clean now.

The `-Werror` flag is disabled by default as we can't control everywhere
people will try to build/install. The CI enables it via
-DTORCH_MLIR_ENABLE_WERROR_FLAG=ON.
2024-01-30 23:33:21 -08:00
Stella Laurenzo 943164d797
Fix some spurious `None` values in tests (broken at head). (#2840) 2024-01-30 22:39:22 -08:00
Stella Laurenzo 26c0ecd09c [nfc] Remove unused var causing error downstream 2024-01-30 22:18:13 -08:00
Aart Bik 105aad6f57
[torch-mlir] provide FX traced graph importer for sparse tensors (#2817)
Note that we are waiting for actual FX traced graph support for sparse
tensors. For details see

https://github.com/pytorch/pytorch/issues/117188

Until then, however, we provide this clever importer that builds the FX
traced graph for for the dense case and then puts a sparse annotation
back on the parameters.

With import test.
2024-01-30 21:22:12 -08:00
Ramiro Leal-Cavazos 1a7442e0aa
Add clang-format check to CI (#2816)
This PR adds a check to the CI right after checking out the Torch-MLIR
repository to make sure that the changes in the PR don't require any
`git clang-format` modifications.
2024-01-30 19:59:46 -08:00
Yuanqiang Liu d778950f45
[Torch Dialect] add fold pattern for aten.clone (#2804) 2024-01-31 09:43:21 +08:00
Rob Suderman 25a5a22cbd
[torch] Support `torch.convolution` quantized lowering to `linalg` (#2811)
Linalg has quantized specific operations. We can lower to these
operations when there is a known zeropoint and scale operations. This
allows the `convolution` to occur with lower bitwidth's, improving the
overall performance.
2024-01-30 13:46:47 -08:00
Aaron St George 4c557847bd
Don't fold `aten.detach` if result isn't same type as input. (#2824)
We were seeing some assertion failures after some checks around folders
were tightened up in LLVM:
https://github.com/llvm/llvm-project/pull/75887 . This PR essentially
moves the logic that used to be applied at the LLVM level into the
folder, which seems to be the suggested fix.

I'm not sure if the IR that caused issues for us _should_ be valid?
```
%1 = torch.aten.detach %arg0 : !torch.tensor<[1],f32> -> !torch.tensor
```
A better fix might be to create a verifier ensuring the result of
`aten.detach` has the same type as its operand.

---------

Co-authored-by: aaron-stgeorge <aaron.stgeorge@getcruise.com>
2024-01-30 09:45:51 -08:00
Rob Suderman db67bc555a
Bump LLVM to llvm/llvm-project@70eb0e3 (#2827) 2024-01-30 09:01:42 -08:00
James Newling 9d983161fc
Describe how to get --debug and --debug-only flags in dev notes (#2793)
Change should be visible :
https://github.com/newling/torch-mlir/blob/docs_update/docs/development.md
2024-01-30 08:30:00 -08:00
James Newling 1e882f5803
Additional information in error message (#2783)
See change in test for what the new message looks like.
2024-01-30 08:28:08 -08:00
Vivek Khandelwal e18fcebd3a
[CI] Change Roll PyTorch runner (#2828)
Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-01-30 16:42:18 +05:30
aldesilv eff325abc3
OnnxToTorch ReduceMax lowering (#2768)
Fixes https://github.com/nod-ai/SHARK-Turbine/issues/352
2024-01-30 11:44:48 +05:30
Quinn Dawkins 1d6aca3823
Add .git-blame-ignore-revs to allow ignoring sweeping formatting changes (#2823)
This allows the following command to be used to ignore sweeping
formatting changes.

```
git blame --ignore-revs-file .git-blame-ignore-revs <file_of_interest>
```
2024-01-29 10:29:51 -08:00
Quinn Dawkins 494089d53d
Clang format refresh (#2812)
After noticing a number of commits with unrelated formatting changes,
I think something was changed with clang-format at one point and we're
seeing a number of unrelated changes. Doing a refresh can help avoid
this.

The changes made here came from
```
find lib -iname *.h -o -iname *.cpp  | xargs clang-format -i --style=llvm
find include -iname *.h -o -iname *.cpp  | xargs clang-format -i --style=llvm
find projects -iname *.h -o -iname *.cpp  | xargs clang-format -i --style=llvm
```
2024-01-29 12:59:33 -05:00
Rob Suderman d3fd754b93
[onnx] `onnx.MatMulInteger` lowering to `torch.mm` and `quint*` types (#2761)
Torch does not have an equivalent matmul operation for integers. Instead
it sidechannels the information via its quantized types. For this
lowering we setup these sidechannels then invoke `torch.mm`.
2024-01-29 09:40:21 -08:00
Rob Suderman 67cb2e7341
Fix illegal use of TypeRange (#2815)
TypeRange is an ArrayRef<Type> and therefore cannot be safely
instantiated from a list initializer.
2024-01-29 09:23:05 -08:00
Stella Laurenzo 032f225fa5 [ci] Allow long line in YAML 2024-01-27 19:43:41 -08:00
Stella Laurenzo 6b3ebb237f [ci] Use a different cache key for torch nightly vs stable. 2024-01-27 19:42:29 -08:00
Stella Laurenzo 4513c3ca87
[ci] Add step to run unit tests. (#2820) 2024-01-27 19:35:48 -08:00
Stella Laurenzo 77c14ab22b
[ci] Upgrade to new runners and disable unsupported jobs. (#2818)
Per the RFC and numerous conversations on Discord, this rebuilds the
torch-mlir CI and discontinues the infra and coupling to the binary
releases
(https://discourse.llvm.org/t/rfc-discontinuing-pytorch-1-binary-releases/76371).

I iterated on this to get latency back to about what it was with the old
(much larger and non-ephemeral) runners: About 4m - 4.5m for an
incremental change.

Behind the scenes changes:

* Uses a new runner pool operated by AMD. It is currently set to manual
scaling and has two runners (32-core, 64GiB RAM) while we get some
traction. We can either fiddle with some auto-scaling or use a schedule
to give it an increase during certain high traffic hours.
* Builds are now completely isolated and cannot have run-to-run
interference like we were getting before (i.e. lock file/permissions
stuff).
* The GHA runner is installed directly into a manylinux 2.28 container
with upgraded dev tools. This eliminates the need to do sub-invocations
of docker on Linux in order to run on the same OS that is used to build
wheels.
* While not using it now, this setup was cloned from another project
that posts the built artifacts to the job and fans out testing. Might be
useful here later.
* Uses a special git cache that lets us have ephemeral runners and still
check out the repo and deps (incl. llvm) in ~13s.
* Running in an Azure VM Scale Set.

In-repo changes:

* Disables (but does not yet delete):
  * Old buildAndTest.yml jobs
  * releaseSnapshotPackage.yml
* Adds a new `ci.yml` pipeline and scripts the steps in `build_tools/ci`
(by decomposing the existing `build_linux_packages.sh` for in-tree
builds and modularizing it a bit better).
* Test framework changes:
* Adds a `TORCH_MLIR_TEST_CONCURRENCY` env var that can be used to bound
the multiprocess concurrency. Ended up not using this in the final
version but is useful to have as a knob.
* Changes the default concurrency to `nproc * 0.8 + 1` vs `nproc * 1.1`.
We're running on systems with significantly less virtual memory and I
did a bit of fiddling to find a good tradeoff.
* Changed multiprocess mode to spawn instead of fork. Otherwise, I was
getting instability (as discussed on discord).
* Added MLIR configuration to disable multithreaded contexts globally
for the project. Constantly spawning `nproc * nproc` threads (more than
that actually) was OOM'ing.
* Added a test timeout of 5 minutes. If a multiprocess worker crashes,
the framework can get wedged indefinitely (and then will just be reaped
after multiple hours). We should fix this, but this at least keeps the
CI pool from wedging with stuck jobs.

Functional changes needing followup:

* No matter what I did, I couldn't get the LTC tests to work, and I'm
not 100% sure they were being run in the old setup as the scripts were a
bit twisty. I disabled them and left a comment.
* Dropped out-of-tree build variants. These were not providing much
signal and increase CI needs by 50%.
* Dropped MacOS and Windows builds. Now that we are "just a library" and
not building releases, there is less pressure to test these commit by
commit. Further, since we bump torch-mlir to known good commits on these
platforms, it has been a long time since either of these jobs have
provided much signal (and they take ~an hour+ to run). We can add them
back later post-submit if ever needed.
2024-01-27 18:35:45 -08:00
Stella Laurenzo 4a4d80a6ad
[ci] Add lint job and enable yaml linting of GH files. (#2819) 2024-01-27 15:48:06 -08:00
MaheshRavishankar 28c7051ceb
Bump LLVM to llvm/llvm-project@5fcf907b34 (#2810) 2024-01-26 18:38:44 -08:00
Aart Bik 46a25d7241
[torch-mlir][sparse] preserve sparsity during lowering torch to linalg (#2809)
This preserves sparsity at the most obvious places of lowering TORCH
tensors to MLIR RankedTensorType tensors. Other places are marked for
audit. With some initial lowering tests.
2024-01-26 10:54:59 -08:00
Vivek Khandelwal da7c6d2c16
[MLIR][TORCH] Add support for dynamic shape for Onnx.Transpose op (#2803)
Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-01-26 09:46:54 -08:00
Phaneesh Barwaria 4964977e85
[ONNX][MLIR] support constantOfShape op (#2747) 2024-01-26 09:36:39 -08:00
Yuanqiang Liu e73c5368fb
[FxImporter] make FxImporter to fit python<=3.9 (#2802)
As that torch with py3.9 is also used widely.
2024-01-26 09:01:47 +08:00
Rob Suderman 2ef228328f
[torch] `torch.dequantize` for per channel tensors to` linalg` (#2769)
Support a lowering for dequantization for per channel tensors from
`torch` dialect to a linalg decomposition. Tested via a numerical
`torch` test.
2024-01-25 16:40:21 -08:00