This folds small version of the tensor-scalar comparison operators as
they are commonly used for shape computations. This includes le, lt, ge,
gt, eq, and ne.
The current padding operation was not functional for dynamic shapes.
Updated and enabled tests so that onnx.pad tests pass.
Work TBD for reflection padding.
The addition of an e2e test is actually provided in the Shark-Testsuite.
This adds 2 test cases for the gridsampler e2e test.
Also as intended there were some items found which needed correction, so
the Gridsampler op has also a change.
Add e2d support for `aten.linalg_norm` by decompose it to
`aten.linalg_vector_norm`.
Lowering to `aten.linalg_matrix_norm` is still unsupported.
To Test:
`python -m e2e_testing.main -v`
---------
Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
Existing lowering via aten.view does not work as well for dynamic shapes
as the lowering to tensor.expand must re-infer dynamic shape matching.
Better to directly lower.
We collapsed and broadcasted scatter indices to a single element
version. We should instead upport `tm_tensor.scatter`s support for
multiple indices and the implicitly broadcasted behavior. This avoids
the serialization and materializing a needlessly large indices tensor.
Even though the reference compiler is not about performance, inlining
the generated sparse helper methods has a rather big positive impact on
performance, leaving a much better first impression. Therefore, we added
this inlining pass (which leaves all other PyTorch modules unaffected,
since they tend to be one big main() method to start with).
testing:
$./tools/e2e_test.sh --config linalg
Summary:
Passed: 1164
Expectedly Failed: 8
$ python -m e2e_testing.main --config=torchdynamo
Summary:
Passed: 976
Expectedly Failed: 162
We can route the torch tests via `onnx` using the `torch.onnx.export`
tooling. We can then reimport, lower to torch, and compile to linalg to
validate the onnx path is working correctly.
The current implementation exposes some failures in the `onnx` path so
we cannot enable the onnx test suite yet due to segmentation faults.
This commit adds the OnnxToTorch lowering for cosh, acosh, asin, asinh,
and atanh op.
This commit also adds the TorchToLinalg lowering for acosh, asin, asinh,
and atanh op.
Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
Some operations include a backend matcher for specialized operations. We
map these back to generics so they appropriately match to the high
performance versions. This is done for the attention operation.
This PR introduces a sparse_jit wrapper that can run simple models with
sparse tensor inputs end-to-end. The implementation shows all required
components on modifying sparse tensor types with a 1:N relation on the
call sites. Two tests shows that the JIT runs end-to-end while computing
the correct results.
More details to follow (generalizing to COO and different ranks, as well
as support for *output* sparse tensors), but the general concepts are
all here now.
**_Update: Thanks to Rob, bump to proper LLVM/MLIR hash is done!_**
_**NOTE that all parameter passing changes are nicely done "downstream"
in MLIR, so very little changes are required in torch-mlir code
proper**_
---------
Co-authored-by: Franz Haniel <77495327+frafranz@users.noreply.github.com>
Co-authored-by: Franz Haniel <franz.haniel@amd.com>
The lowering decomposes AtenTraceOp into an AtenDiagonalOp followed by
AtenSumOp.
The progress is tracked in
https://github.com/nod-ai/SHARK-Turbine/issues/333.
---------
Co-authored-by: Franz Haniel <franz.haniel@amd.com>
There is no lowering support for math::AbsIOp, so if the operand is an
integer type, it will fail to lower to math::AbsFOp since the op operand
#0 must be floating-point-like.
This adds a few passes that will ensure linalg with sparse tensors are
properly lowered to loops and can run using the ExecutionEngine for
testing (a few details on parameter passing from PyTorch still TBD)
Test results:
$ ./tools/e2e_test.sh --config linalg
Summary:
Passed: 1144
Expectedly Failed: 8
$ python -m e2e_testing.main --config=torchdynamo -v
Summary:
Passed: 960
Expectedly Failed: 163
Filed issue:
https://github.com/pytorch/pytorch/issues/119407
Lowering of torch.aten.all.dim to linalg.
Per PyTorch documentation:
> This function matches the behaviour of NumPy in returning output of
dtype bool for all supported dtypes except uint8. For uint8 the dtype of
output is uint8 itself.
Since there is no support for ui8 in torch-mlir currently
(https://github.com/llvm/torch-mlir/pull/1384#issuecomment-1260011334)
implementation returns failure for that case.
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>
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>
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.
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.
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.
This includes custom op matching for decomposed operations and fusing
dequantization into dense operations. As a validation we compare
to the dequant+mm torch implementation.
The logic here is very similar to the conversion for AdaptiveAvgPool1d
#2661 with a few modifications:
1. buffVal = -inf instead of 0
2. the main linalg generic op accumulates a max, instead of a sum, to
the first output tensor
3. avg pooling requires dividing the sum pool by the kernel width, which
we stored as an auxilliary tensor (kSizeTensor). Here, the auxiliary
tensor will be recording the indices. Strangely enough, the only
signature available for this function is to return indices, and it
appears that they must be computed whether the user desires them or not.
See
[pytorch/torch/nn/functional.py](https://github.com/pytorch/pytorch/blob/main/torch/nn/functional.py#L1174).
Before writing other adaptive pooling conversions, the logic of this
decomposition should be rolled into a helper function that will work for
both max and avg pooling ops. Even the auxiliary tensor should likely be
automated. This code was written in a slightly more tedious way than
strictly necessary (often using loops to fill SmallVectors up to rank-2,
which is only two in this case), in order to more easily facilitate the
transition to a helper function.
convolution with [time,batch,channel] ordering, as opposed to the
default [batch, channel, time]. Currently implementing by transposing
the input and output, but may need to get its own implementation in the
future because this is supposed to be an op that gives a speedup. This
is used by fairseq
(https://github.com/facebookresearch/fairseq/issues/172).
(in case you were wondering like me, this is different from transposed
convolution. Transposed convolution has fractional strides).
---------
Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
Co-authored-by: Frederik Harwath <frederik.harwath@amd.com>
Handle both `torch.dequantize` and `torch.quantize_per_tensor` including
the op based quantization parameter tracking. This includes adding
`qint32` to torch types as it was missing during the initial type
inclusion.
For testing we only have `torch.int8` and `torch.float` types on
function boundaries as the `qint8` types require passing the scale
and zero point quantization information which is not supported yet.
Adaptive pooling ops can only be decomposed into their non-adaptive
counterparts in trivial cases.
For example, the current decomposition for AtenAdaptiveAvgPool1dOp in
DecomposeComplexOps.cpp supports outSize = inSize (i.e., do literally
nothing), and outSize = 1 (i.e., do a batched average).
The reason adaptive pooling ops are difficult to lower to linalg is that
they are not constantly strided. They are computed by taking an input
tensor of shape (N, C, Hin), and an output size Hout, and computing the
output tensor at position (n,c, h) in the following way:
1. compute st(h) = (h*Hin)//Hout
2. compute en(h) = 1 + ((h+1)*Hin -1)//Hout
3. apply a computation (max or avg) to the slice: INPUT[n, c,
st(h):en(h)]
The provided sample implementation (for ConvertAtenAdaptiveAvgPool1dOp)
uses tensor.extract to access the input tensor inside the payload of a
linalg generic op. This is likely an unattractive use of linalg generic
ops, which is why I am asking for some more targeted feedback on the
validity of this approach before attempting to support the many other
adaptive pooling ops.
Specifically:
- Is the performance of this implementation bad enough to warrant
targeting different dialects entirely? e.g. TMtensor/linalg ext/ etc.
- If the provided implementation is of acceptable performance to the
community, then is it permissable to remove the Adaptive pooling
decompositions from DecomposeComplexOps.cpp? Based on the current
structure of the -torch-decompose-complex-ops pass, it does not seem
possible to only decompose the adaptive ops in special cases (it seems
to get stuck in an infinite loop on a match failure). I would be happy
to instead incorporate the case logic into the conversion directly, and
remove the decompositions once they are rendered completely obsolete.
As long as this approach is acceptable, I can clean up the
implementation with some helper functions, and quickly add support for
each of the remaining Adaptive pooling ops.
Rob Suderman
Andreas Falkenberg
Ze Zhang
mmakevic
Vivek Khandelwal
ptrifunovic98
Aart Bik
aldesilv
Yuanqiang Liu
Franz Haniel
Avinash Sharma
saienduri
Xida Ren (Cedar)
Ilija Kalinić
Stella Laurenzo
zjgarvey
John Wu
Sungsoon Cho
lonely eagle
lisaliu1
Frederik Harwath