This is intended to explore support for non-structured ops that can't
be modeled by Linalg dialect. `tm_tensor.scan` and `tm_tensor.scatter`
are added as the first such ops. The dialect should aim to be
upstreamed in the future.
A few remain in examples/docs that will be naturally be updated in due
time.
This regresses the list support and the general direction of more widely
supported control flow, lists/dicts/globals that we were going for with
the TorchScript path. The idea is that we are deferring that work to
make torch-mlir a very clean standalone thing. We will reboot it,
probably using some of the tools of iree_pydm to make it simpler, and in
a more natural place (such as an iree-torch repo that depends on IREE and
torch-mlir to build a working PyTorch frontend solution for IREE -- it
was really weird that npcomp depended on IREE).
This moves the bulk of the Python code (including the Torch interop)
from `frontends/pytorch` into `torch-mlir/TorchPlugin`. This also
required reconciling a bunch of other Python-related stuff, like the
`torch` dialects.
As I did this, it was simpler to just remove all the old numpy/basicpy
stuff because we were going to delete it anyway and it was faster than
debugging an intermediate state that would only last O(days) anyway.
torch-mlir has two top-level python packages (built into the
`python_packages` directory):
- `torch_mlir_dialects`: `torch` dialect Python bindings (does not
depend on PyTorch). This also involves building the aggregate CAPI for
`torch-mlir`.
- `torch_mlir`: bindings to the part of the code that links against
PyTorch (or C++ code that transitively does).
Additionally, there remain two more Python packages in npcomp (but
outside `torch-mlir`):
- `npcomp_torch`: Contains the e2e test framework and testing configs
that plug into RefBackend and IREE.
- `npcomp_core`: Contains the low-level interfaces to RefBackend and
IREE that `npcomp_torch` uses, along with its own
`MLIR_PYTHON_PACKAGE_PREFIX=npcomp.` aggregation of the core MLIR
python bindings. (all other functionality has been stripped out)
After all the basicpy/numpy deletions, the `npcomp` C++ code is now very
tiny. It basically just contains RefBackend and the `TorchConversion`
dialect/passes (e.g. `TorchToLinalg.cpp`).
Correspondingly, there are now 4 main testing targets paralleling the
Python layering (which is reflective of the deeper underlying dependency
structure)
- `check-torch-mlir`: checks the `torch-mlir` pure MLIR C++ code.
- `check-torch-mlir-plugin`: checks the code in `TorchPlugin` (e.g.
TorchScript import)
- `check-frontends-pytorch`: Checks the little code we have in
`frontends/pytorch` -- mainly things related to the e2e framework
itself.
- `check-npcomp`: Checks the pure MLIR C++ code inside npcomp.
There is a target `check-npcomp-all` that runs all of them.
The `torch-mlir/build_standalone.sh` script does a standalone build of
`torch-mlir`.
The e2e tests (`tools/torchscript_e2e_test.sh`) are working too.
The update_torch_ods script now lives in
`torch-mlir/build_tools/update_torch_ods.sh` and expects a standalone
build.
This change also required a fix upstream related to cross-shlib Python
dependencies, so we also update llvm-project to
8dca953dd39c0cd8c80decbeb38753f58a4de580 to get
https://reviews.llvm.org/D109776 (no other fixes were needed for the
integrate, thankfully).
This completes most of the large source code changes. Next will be
bringing the CI/packaging/examples back to life.
This creates the `external/torch-mlir` directory as an
LLVM_EXTERNAL_PROJECTS-compatible project (analogous to
`iree-dialects`) and completes movement/rename of all pure MLIR C/C++
compiler code into there. The next step will be to move all the Python
code / code that links/includes PyTorch C++ code (which currently lives
in `frontends/pytorch`) into a subdirectory here.
I call this "earthmoving" because it is mostly mechanical changes and
renames. As a quick summary (we can change this down the road easily)
- C++ `mlir::NPCOMP::Torch -> mlir::torch::Torch`
- CAPI `npcompTorchListTypeGet -> torchMlirTorchListTypeGet`
- preprocessor `#ifndef NPCOMP_ -> #ifndef TORCHMLIR_`
- CMake `NPCOMPFoo -> TorchMLIRFoo`
The goal of this is to create a standalone project creating a center of
mass for entry into the MLIR ecosystem from PyTorch, suitable in scope
for eventual inclusion/ownership in PyTorch. The idea is that
`external/torch-mlir` will some day be pulled out into its own
repository, and then npcomp will simply pull it in as a submodule.
Layering-wise, what lives in `torch-mlir` lowers code from PyTorch
(currently TorchScript, but TorchFX or pytorch/xla-style tracing are
possible extensions) down to what we have been calling the "Torch
backend contract" which is cleaned up IR (inlining, simplifcation,
conversion to value tensors, ...) entirely in the `torch` dialect. This
is the branching off point for further lowering, of which npcomp takes
one opinion (outside `torch-mlir` of course!), namely the
`TorchConversion` dialect/transforms which lower to IR suitable for IREE
and other linalg-on-tensors based lower-level compilers.
Summary of changes:
- move `{include,lib,test}/Dialect/Torch` into `torch-mlir`
- move relevant parts of CAPI into `torch-mlir`.
- leave a few things related to the `torch-mlir` Python build commented
out, which should be resolved in a subsequent change.
This converts a basic list op (torch.prim.ListConstruct) to the IREE
dialect.
```
def forward(self, x: float):
return [x, x]
```
turns into:
```
builtin.func @forward(%arg0: !torch.float) -> !torch.list<!torch.float> {
%0 = torch.prim.ListConstruct %arg0, %arg0 : (!torch.float, !torch.float) -> !torch.list<!torch.float>
return %0 : !torch.list<!torch.float>
}
```
which turns into:
```
builtin.func @forward(%arg0: f64) -> !iree.list<f64> {
%c1 = constant 1 : index
%c0 = constant 0 : index
%c2 = constant 2 : index
%0 = iree.list.create %c2 : !iree.list<f64>
iree.list.set %0[%c0], %arg0 : !iree.list<f64>, f64
iree.list.set %0[%c1], %arg0 : !iree.list<f64>, f64
return %0 : !iree.list<f64>
}
```
As part of doing this, I realized that it was time to formalize the IR
form that we reach right before running TorchTo{Linalg,Std,...}. We now
call it the "Torch backend contract". We then lower the "Torch backend
contract" to the "npcomp backend contract", which involves the new
TorchConversion (`torch_c`) dialect, which holds ops that need to
operate on both the npcomp backend types (e.g. builtin tensors, i1, IREE
list, etc.) and the `!torch` types.
This made more sense, as I realized that if I didn't factor out
`torch_c` then the Torch dialect would have a dependency on IREE
dialect (we previously didn't notice this was an issue because we only
depended on `builtin` types), which seemed wrong to me.
Recommended review order:
- TorchToIREE.cpp / `TorchToIREE/basic.mlir`
- Look at the new structure of createTorchScriptToNpcompBackendPipeline.
It now lives in TorchConversion/Transforms/Passes.cpp and cleanly
calls into `Torch::createTorchScriptToTorchBackendPipeline` for the
frontend lowering to the Torch backend contract.
- Mechanical change extracting
`torch_c.{to,from}_{i1,i64,f64,builtin_tensor,iree_list}` into a new
TorchConversion dialect, and a few passes specific to the lowering
from the Torch backend contract to the npcomp backend contract.
- Minor fixes to TorchToLinalg.cpp to use unconverted operands (now that
we convert lists as part of operand materialization, we need to use
the original operands). Also added test for AtenMaxPool2dOp and fixed
m_TorchConstantIntList.
- TmpDeleteDeadIREELists pass. Temporary pass for deleting dead IREE lists that
are created as part of operand materialization for conv/max pool/avg pool ops
in TorchToLinalg.
These were legacy concepts that are now superceded by direct Torch to
linalg-on-tensors lowering. These were based on some very early thinking
related to the layering of frontends vs codegen, which is now obsolete
because:
- We expected a lot more centralization at the frontend (TCF) level. It
turns out that frontend needs really vary a lot, and there is no grand
unifying TCF dialect plausible. The additional layer isn't worth it.
- Linalg-on-tensors obsoletes the primary need for TCP. There are still
a few things not representable with linalg-on-tensors, but the support
is growing and the whole "not included in linalg-on-tensors" direction
needs to be rethought. Our TCP dialect didn't cover any of the
actually important things in this space (such as sort, FFT, top-k,
etc.).
See historical [slides](https://drive.google.com/file/d/1iljcpTQ5NPaMfGpoPDFml1XkYxjK_6A4/view) / [recording](https://drive.google.com/file/d/1jSPa8TwPKUt0WuLquGc8OgSUVYJHMvWZ/view)
for more details on the origin story here.
Their presence was confusing users too
[bug](https://github.com/llvm/mlir-npcomp/issues/248).
Also,
- Trim down npcomp-run-mlir testing. It was testing TCF to TCP
lowering for the most part. The essential stuff is retained and
rephrased with linalg-on-tensors. (we should probably rename it
"refback-run" or something, as it is just a way to invoke RefBackend)
- test/Python/Backend/RefJIT/simple_invoke_numpy.py is XFAIL'ed. Our
"anti-framework" direction seems to be the likely future path.
This is a really major and invasive restructuring of the way we get
torch operators (`torch::jit::Operator` / `c10::OperatorHandle`) into
MLIR. Please forgive the challenging review, but due to the sheer
invasiveness, it wasn't really practical do do it in sane smaller
pieces.
This fully replaces everything that was already working on the
TorchScript path (actually, more -- we added tanh support to
TorchToLinalg in order to delete the older code paths). Additionally,
I've kept the lights on for the acap path too, including what little e2e
stuff was working before (for expediency I made a few tiny compromises
along the way that will be easy to undo when we give that path proper
attention).
Overview of the new design:
- The torch operator `somens::someunqualname.someoverloadname` is
imported as `torch.somens.someunqualname.someoverloadname` (skip the
last dotted part if the overload name is empty), OR, if we don't have
such an op registered, it is imported as
`torch.operator "somens.someunqualname.someoverloadname" (...) : ...`.
- The addition of the "overload name" is a critical element here, as
the `(ns,unqual,overload)` triple is unique, which solves a lot of
problems we were having.
- This involves having separate MLIR ops for the `trailing_` and
`.out` variants and all the different overloads. This seemed
necessary, because the set of overloads is so wild and varied and
unstructured. The previous design was leaning into some underlying
structure that just isn't there -- the default situation is
the "random overload that we want to manage on the MLIR side",
rather than that being an exception. E.g. `aten::ne` (not-equal)
has 21 overloads, only 4 of which are c10 dispatcher ops see
[gist](https://gist.github.com/silvasean/190ba918c550c956260e21254e1b8aa1),
and the "out" variant is really called `.Tensor_out` instead of
`.out` as it frequently is for other ops.
- Rationale for all being in `torch` namespace: the set of operators
are so varied and unstructured that "dialect per namespace"
doesn't result in anything resembling the typical MLIR dialect
boundary expectations. We could maybe draw the boundary at
dispatcher ops vs non-dispatcher ops, but that doesn't seem to
really result in very much useful structure at this point in time.
- Note: within the torch operator registry, we effectively have a
mini-basicpy subdialect (already type-resolved), which is reasonably
structured.
- The existing Torch op interfaces are also removed -- now that we
track the overload name, we can losslessly find the original
operator.
- Instead of `ATenRecognizeKernelsPass`, we now have a
`ReduceOpVariantsPass` that keys off certain traits (and perhaps
eventually interfaces) to reduce variants of ops to a smaller set,
ideally operating on immutable tensors and using surrounding ops to
model the mutability/aliasing aspects.
- Note: `torch.ns.unqual.overload` ops allow both immutable and
mutable tensors (unlike the previous hard distinction in the common
case). This is a premonition for a future change that will introduce a
bona fide `!torch.tensor` type that will clean up a bunch of stuff.
- `TorchToLinalg` / `TorchToStd` supercede the existing
"ATen->TCF->TCP->Linalg" path.
- The new `torch_ods_gen.py` supercedes `torch_signature_ods_gen.py`.
It should look somewhat familiar, but the benefit of hindsight has
allowed a lot of simplifications.
The overall trend seems to be to make the `torch` dialect a nice layer
independent of anything else. It feels like as a natural result of
various future changes we will be removing the reliance on basicpy+numpy
dialects and have a nice self-contained type system too that properly
models the TorchScript type system (including proper subtyping,
mutable/immutable tensors, optional dtype, etc.).
Recommended review order:
- Start at some of the new import IR, e.g. in
`frontends/pytorch/test/node_import/prim.py`,
`frontends/pytorch/test/acap_export/test_export_add3.py`, and other
tests.
- `frontends/pytorch/python/torch_mlir_utils/codegen/torch_ods_gen.py`
and associated generated files:
- `include/npcomp/Dialect/Torch/IR/GeneratedAtenOps.td`
- `include/npcomp/Dialect/Torch/IR/GeneratedPrimOps.td`
- Inspect `ReduceOpVariants.cpp` / `reduce-op-variants.mlir` and the new
traits in `include/npcomp/Dialect/Torch/IR/TorchTraits.h`
- Various code changes in the import path in
`frontends/pytorch/csrc/builder`. Probably most interesting is the new
code in `torch_to_mlir_utils.cpp` that has the logic to create the
`torch.operator` ops or `torch.ns.unqual.overload` ops.
This is the [new ResNet IR](https://gist.github.com/silvasean/5407aafb710d07612b7b5b92eabecebe),
just to be able to look at a substantial sample of IR in the new style.
This trait lets us model the semantics of various aten/torch/numpy ops
that are insensitive to type refinements. This replaces
hardcoded/inconsistent checks for this property.
To show usage of this new trait, we fix up some old uses, and improve
RefineTypes to be smarter about rewriting with this trait.
This pass verifies that a given module satisfies the contract that we
have for backends. This is phrased as an "allowlist", because we want to
keep this interface tight. Also, this gives much better diagnostics than
a backend randomly crashing or failing to compile would (though they
could still be improved).
This was especially painful because if we had
`tensor<?x!numpy.any_dtype>` slip through, at some point RefBackend
would convert it to a memref type and trip the "verify type invariants"
assertion which gives no location or anything and crashed the process,
which was very unpleasant.
We implement this with the dialect conversion framework, which works
reasonably well and was quick to put together and familiar, but is still
very "op oriented". We probably want to make this hand-rolled
eventually, especially the error reporting (the most useful kind of
error for a dialect conversion user is not necessarily the best for this
use case). Also, in production, these error will go to users, and need
to be surfaced carefully such as "the compiler needs a type annotation
on this function parameter" which in general requires some special
analysis, wordsmithing, and overall awareness of the e2e use case (such
as how much we can lean into certain source locations) to provide a
meaningful user-level diagnostic.
Also, add `inline` to the current frontend lowering pass pipeline to
allow slightly more complicated programs that otherwise would fail on
shape inference.
Recommended review order:
- Changes in frontends/pytorch/examples/
- Changes in python/npcomp/compiler/pytorch/backend/
- Boilerplate for the `npcomp-iree-backend-lower-linkage` pass.
This change separates out a
`npcomp.compiler.pytorch.backend.frontend_lowering` module that does the
common lowering for all backends. The individual compiler backends
`npcomp.compiler.pytorch.backend.{refjit,iree}` now accept a loosely
defined "TCP + scalar code" IR mix that will be formalized in the
future as the interface to codegen backends.
This also required adding a small pass
`npcomp-iree-backend-lower-linkage` which adds `iree.module.export` onto
functions, and layering that into the frontend flow. The pass doesn't
require a C++-level dependency on IREE, which is nice for now. TBD how
we are going to handle lists (we hope we can get away with sneakerneting
some td files and relying on loose IR compatibility).
Running through IREE requires the ability to import `iree.compiler` and
`iree.runtime`, which can be obtained as follows:
```
python3 -m pip install iree-compiler-snapshot iree-runtime-snapshot -f https://github.com/google/iree/releases/tag/snapshot-20210406.200
PYTHONPATH="${PYTHONPATH}:${MY_IREE_BUILD}/bindings/python/"
```
This patch makes it painfully clear that we don't have any e2e testing
harness to really plug into, and also don't have a usable Python API to
our compiler stack (something usable in a jupyter notebook).
That will be addressed in subsequent commits. We've been flying by the
seat of our pants with this `examples` directory that isn't subject to
any kind of testing or real usability concerns.
* Most updates are mechanical except:
* python/npcomp/__init__.py and python/NpcompModule.cpp: New init/registration bits to replace some automatic things being done in the old bindings. Also an annoying linkage hack that I'll need to triage next.
* NpcompModule.cpp: New python helpers for custom types and other hard to reach items (for the new bindings).
* PybindUtils.h: Extended type casting so that the local extension can directly exchange Mlir* C types.
* python/npcomp/dialects/*: Build support and ODS bindings for local dialects.
* mlir_utils.py: Defines an ImportContext to replace the old/bad "Helper" class that tracked locations, and insertion points. This has a number of methods on it that would be good candidates to think about better ways to do them upstream.
* Also hoisted a few stand-alone samples to dedicated unit tests as they covered important things.
* More cleanup can be done, but keeping this patch as mechanical as possible to stay in NFC land (this is big enough).
* IREE doesn't have proper install support, so there is some temporary hoaky hacking in our CMakeLists.txt to shuttle some symlinks around.
* Reworked the original numpy e2e with IREE test to pipe through iree-translate.
* Removed all of the C++-level dependencies.
* Will generalize and apply to the PyTorch backend in a followup.
* Need to have a dag of shared library deps in order to interop across python extensions (as presented in ODM).
* Introduced add_npcomp_library and friends to mirror the MLIR setup.
* Adds a libNPCOMP.so shared library.
* Redirects tools and extensions to link against libNPCOMP.so (instead of static libs).
* Moves all libraries to lib/, all binaries to bin/ and all python extensions to python/. The invariant is that the rpaths are setup to have a one level directory structure.
* Reworks the _torch_mlir extension to build like the others (still need to come up with a consolidated rule to do this instead of open coded).
* Includes an upstream version bump to pick up needed changes.
Sizes with dynamic linking (stripped, release, asserts enabled):
libNPCOMP.so: 43M (includes much of the underlying LLVM codegen deps)
libMLIR.so: 31M
_npcomp.so: 1.6M (python extension)
_torch_mlir.so: 670K (python extension)
npcomp-capi-ir-test: 6.3K
npcomp-opt: 351K
npcomp-run-mlir: 461K
mnist-playground: 530K
Still more can be done to normalize and optimize but this gets us structurally to the starting point.
I now realize that VerboseCamelCase is not the best choice for dialect
directory/file names and C++ identifiers (take e.g. "Linalg", "Basicpy",
etc. as prior art here; not LinearAlgebra or BasicPython). If I had to
name the convention it seems to be "Shortword" (or of course just
acronym dialects like LLVM, SCF, etc.).
This rename also has the side benefit of differentiating RefBackend
directories, which now refer to the actual backend itself, from
Refback/Refbackrt, which are the dialects which happen to be used by
that backend.
Other than the dialect definitions (which will live in standard Dialect/
subdirectory), the goal here is to keep RefBackend-related code nested
in {include/npcomp,lib,test}/RefBackend.
This is the first in a patch series that is refactoring the
constellation of things variously called or associated with "E2E",
"RefE2E", "npcomprt", and "TCP" into a more cleanly layered result.
Concretely, this first patch fixes the fact that TCP was basically
acting like a dumping ground needed by the reference backend. This
splits it out, which is fairly mechanical, but touches a lot of lines of
code (basically replacing `tcp` with `refback` and `TCP` with
`RefBackend).
Now, the RefBackend dialect is that dumping ground, which
is slighly better, as it starts allowing TCP to become a nice clean
middle layer that is not related per se to the reference backend.
The previous name RefE2E or "reference e2e flow" was super confusing.
Now that we are seeing more clearly where the "backend" distinction
lies, the [RefBackend] commit tag is born :)
* Uses the MLIR-C API since that will save us a lot of grief down the road (i.e. will give PyTorch and libMLIR/libNPCOMP the ability to skew version-wise).
* Quite a few TODOs and not yet populating the function in any way.
* llvm-project: b5924a8e27536d19dd5c4d302db29fb6163d5faa
* mhlo: 848ca244d20f045b7921da55a98a04d95ef94f0e
* Multiple breakages that need to be fixed.
Fixes:
* Refactor dialect registration
* Remove all kindof methods (Casting functionality has been added upstream and is implicitly
available, see https://llvm.discourse.group/t/removing-kinds-from-attributes-and-types/1547.)
* Update dialect registration to comply with https://reviews.llvm.org/D85495.
* Remove type kinds and update some changed dialect signatures.
* Upgrade ATen dialect to match upstream needs.
* Move dialect registration to tablegen.
* Register the ListType in tablegen.
* Change dialect initialization signature.
* Use TypeSwitch in MlirIr location printer.
* Remove global registry depends from npcomp-opt.
* Change LowerToLLVM to pass an MLIRContext vs an LLVMDialect for type creation.
* Remove dep on MLIREDSCInterface that is removed upstream.
* Thread through the DialectRegistry for opt and python-like tools.
* Modernize pass registration (This was forced because the GEN_PASS_REGISTRATION code now generates inline functions vs literal pass registration statements)
Co-authored-by: Marius Brehler <marius.brehler@iml.fraunhofer.de>
This patch adds a dialect intended to be used as a frontend dialect
to facilitate lowering from "A Tensor Library" in torch/pytorch.
This patch includes several passes that are useful in conjuction with the
dialect:
--aten-layer-name: Generates layer names for each operation, which are not
present in the original pytorch.
--aten-to-std: Lower the ATen dialect into standard dialect function calls.
--return-elimination-pass: convert functions (primarily the toplevel function)
to pass return values by reference. This simplifies pytorch integration.
--aten-op-report: generate a textual report about the model
--liveness-report
Future patches will implement actual integration with the pytorch jit to
intercept and generates MLIR in this dialect, then lower the resulting MLIR
into function calls through aten-layer-name -> aten-to-std ->
return-elimination -> std-to-llvm. The result would then jitted using the LLVM
jit, linked against a runtime library which makes calls back into pytorch to
implement all the layers.
Co-authored-by: Jeff Fifield <jeff.fifield@xilinx.com>
Co-authored-by: Jeff Fifield <jeff.fifield@xilinx.com>
Mostly this is CMake cleanup. Several library dependencies are missing, which
is often revealed with shared library builds. Also, it's generally bad to
link directly against LLVM libraries because it fails when using
LLVM_LINK_LLVM_DYLIB. MLIR will pull in libLLVM.so, and there will be
duplicate linkage with the the explicit libraries. There may need to be more
refactoring here.
* Rewrites public function signatures to operate on tensors (vs ndarray).
* Most of our backends presume immutable tensors at public function boundaries.
This ~totally reworks the existing "runtime" stuff to be more
principled and usable, such as from Python. It's still not fully
production-quality, mainly in the department of memory management (e.g.
it currently leaks memory; we need to figure out "who frees memrefs" +
the analysis and transformation needed to do that (maybe use upstream
buffer allocation pass?)).
The user API is in include/npcomp/runtime/UserAPI.h, though
include/npcomp/JITRuntime/JITModule.h is a friendlier wrapper.
The stuff under {include,lib}/runtime is totally firewalled from the
compiler and tiny (<6kB, though no attention has gone into optimizing
that size). For example, we don't link in libSupport into the runtime,
instead having our own bare bones replacements for basics like ArrayRef
(the JITRuntime helps with bridging that gap, since it *can* depend on
all common LLVM utilities).
The overall features of npcomprt is that it exposes a module that
with multiple function entry points. Each function has arguments and
results that are tensor-valued, and npcomprt::Tensor is the runtime type
that is used to interact with that (and a npcomprt::Ref<T>
reference-counting wrapper is provided to wrap npcomprt::Tensor in the
common case).
From an implementation perspective, an npcomprt module at the
LLVM/object/binary level exposes a single module descriptor struct that
has pointers to other metadata (currently just a list of function
metadata descriptors). All interactions with the npcomp runtime are
keyed off of that module descriptor, including function lookups and
dispatching. This is done to dodge platform ABI issues and also allow
enough reflection to e.g. verify provided arguments.
Most of the compiler-side work here was in LowerToNpcomprtABI and
LowerToLLVM.
Also,
- Rename npcomp_rt/NpcompRt to npcomprt/Npcomprt; it was getting
annoying to type the underscores/caps.
- misc improvements to bash_helpers.sh
Yi Zhang
Sean Silva
Stella Laurenzo
stephenneuendorffer