Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
//
|
|
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
|
|
|
#include "PassDetail.h"
|
2020-10-07 07:14:37 +08:00
|
|
|
#include "npcomp/RefBackend/RefBackend.h"
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
|
|
|
#include "mlir/Dialect/StandardOps/IR/Ops.h"
|
|
|
|
#include "mlir/IR/StandardTypes.h"
|
|
|
|
#include "mlir/IR/Verifier.h"
|
|
|
|
#include "mlir/Transforms/DialectConversion.h"
|
|
|
|
|
2020-10-08 08:30:10 +08:00
|
|
|
#include "npcomp/Dialect/Refback/IR/RefbackOps.h"
|
2020-10-08 08:12:52 +08:00
|
|
|
#include "npcomp/Dialect/Refbackrt/IR/RefbackrtDialect.h"
|
|
|
|
#include "npcomp/Dialect/Refbackrt/IR/RefbackrtOps.h"
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
|
|
|
using namespace mlir;
|
|
|
|
using namespace mlir::NPCOMP;
|
|
|
|
|
2020-09-17 08:31:40 +08:00
|
|
|
// Get the type used to represent MemRefType `type` on ABI boundaries.
|
|
|
|
// For convenience we do a cast to MemRefType internally.
|
|
|
|
static Type getABIMemrefType(Type type) {
|
|
|
|
return UnrankedMemRefType::get(type.cast<MemRefType>().getElementType(),
|
|
|
|
/*memorySpace=*/0);
|
|
|
|
}
|
|
|
|
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Creating module metadata.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2020-10-08 08:12:52 +08:00
|
|
|
// Returns true if the function signature can be expressed with the refbackrt
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
// ABI.
|
2020-10-08 08:12:52 +08:00
|
|
|
static bool expressibleWithRefbackrtABI(FunctionType type) {
|
|
|
|
// Currently, only memref types can be exposed at refbackrt ABI boundaries.
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
return llvm::all_of(
|
|
|
|
llvm::concat<const Type>(type.getInputs(), type.getResults()),
|
2020-09-17 08:31:40 +08:00
|
|
|
[](Type t) { return t.isa<MemRefType>(); });
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static LogicalResult createModuleMetadata(ModuleOp module) {
|
|
|
|
auto moduleMetadata =
|
|
|
|
OpBuilder::atBlockBegin(module.getBody())
|
2020-10-08 08:12:52 +08:00
|
|
|
.create<refbackrt::ModuleMetadataOp>(module.getLoc());
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
moduleMetadata.metadatas().push_back(new Block);
|
|
|
|
Block &metadatas = moduleMetadata.metadatas().front();
|
|
|
|
OpBuilder::atBlockEnd(&metadatas)
|
2020-10-08 08:12:52 +08:00
|
|
|
.create<refbackrt::ModuleMetadataTerminatorOp>(module.getLoc());
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
|
|
|
SymbolTable symbolTable(module);
|
|
|
|
auto builder = OpBuilder::atBlockBegin(&metadatas);
|
|
|
|
for (auto func : module.getOps<FuncOp>()) {
|
|
|
|
if (symbolTable.getSymbolVisibility(func) !=
|
|
|
|
SymbolTable::Visibility::Public) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
// TODO: Add richer information here such as expected shapes and element
|
|
|
|
// types.
|
2020-10-08 08:12:52 +08:00
|
|
|
builder.create<refbackrt::FuncMetadataOp>(
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
func.getLoc(), builder.getSymbolRefAttr(func.getName()),
|
|
|
|
builder.getI32IntegerAttr(func.getNumArguments()),
|
|
|
|
builder.getI32IntegerAttr(func.getNumResults()));
|
|
|
|
|
2020-10-08 08:12:52 +08:00
|
|
|
if (!expressibleWithRefbackrtABI(func.getType()))
|
|
|
|
return func.emitError() << "func not expressible with refbackrt ABI";
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
}
|
|
|
|
return success();
|
|
|
|
}
|
|
|
|
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Dialect conversion.
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
|
|
|
namespace {
|
2020-09-17 08:31:40 +08:00
|
|
|
class LowerAssertOp : public OpConversionPattern<AssertOp> {
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
public:
|
|
|
|
using OpConversionPattern::OpConversionPattern;
|
|
|
|
LogicalResult
|
2020-09-17 08:31:40 +08:00
|
|
|
matchAndRewrite(AssertOp op, ArrayRef<Value> operands,
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
ConversionPatternRewriter &rewriter) const override {
|
2020-09-17 08:31:40 +08:00
|
|
|
AssertOp::Adaptor adaptor(operands);
|
2020-10-08 08:12:52 +08:00
|
|
|
// The refbackrt runtime function aborts if the argument is true, rather
|
|
|
|
// than when it is false as an `assert` does. So negate the predicate (by
|
|
|
|
// xor'ing with 1).
|
2020-09-17 08:31:40 +08:00
|
|
|
auto c1 = rewriter.create<ConstantOp>(
|
|
|
|
op.getLoc(), rewriter.getIntegerAttr(rewriter.getI1Type(),
|
|
|
|
APInt(/*numBits=*/1, /*val=*/1)));
|
|
|
|
Value assertFailed = rewriter.create<XOrOp>(op.getLoc(), adaptor.arg(), c1);
|
2020-10-08 08:12:52 +08:00
|
|
|
rewriter.replaceOpWithNewOp<refbackrt::AbortIfOp>(op, assertFailed,
|
|
|
|
op.msgAttr());
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
return success();
|
|
|
|
}
|
|
|
|
};
|
|
|
|
} // namespace
|
|
|
|
|
2020-07-11 08:31:24 +08:00
|
|
|
namespace {
|
2020-10-08 08:12:52 +08:00
|
|
|
// At ABI bondaries, use !refbackrt.tensor instead of memref.
|
2020-09-17 08:31:40 +08:00
|
|
|
class FuncOpSignatureConversion : public OpConversionPattern<FuncOp> {
|
2020-07-11 08:31:24 +08:00
|
|
|
public:
|
|
|
|
using OpConversionPattern::OpConversionPattern;
|
|
|
|
LogicalResult
|
2020-09-17 08:31:40 +08:00
|
|
|
matchAndRewrite(FuncOp op, ArrayRef<Value> operands,
|
2020-07-11 08:31:24 +08:00
|
|
|
ConversionPatternRewriter &rewriter) const override {
|
2020-09-17 08:31:40 +08:00
|
|
|
FunctionType type = op.getType();
|
|
|
|
|
|
|
|
TypeConverter::SignatureConversion entryConversion(type.getNumInputs());
|
|
|
|
if (failed(typeConverter->convertSignatureArgs(type.getInputs(),
|
|
|
|
entryConversion)))
|
|
|
|
return rewriter.notifyMatchFailure(op, "could not convert inputs");
|
|
|
|
SmallVector<Type, 1> newResultTypes;
|
|
|
|
if (failed(typeConverter->convertTypes(type.getResults(), newResultTypes)))
|
|
|
|
return rewriter.notifyMatchFailure(op, "could not convert outputs");
|
|
|
|
|
|
|
|
rewriter.updateRootInPlace(op, [&] {
|
|
|
|
// Update the function type.
|
|
|
|
op.setType(FunctionType::get(entryConversion.getConvertedTypes(),
|
|
|
|
newResultTypes, op.getContext()));
|
|
|
|
// Rewrite the entry block.
|
|
|
|
Block &oldEntry = op.getBody().front();
|
|
|
|
Block &newEntry =
|
|
|
|
*rewriter.applySignatureConversion(&op.getBody(), entryConversion);
|
|
|
|
OpBuilder::InsertionGuard guard(rewriter);
|
|
|
|
rewriter.setInsertionPointToStart(&newEntry);
|
|
|
|
BlockArgument newArg, oldArg;
|
|
|
|
for (auto newAndOldArg :
|
|
|
|
llvm::zip(newEntry.getArguments(), oldEntry.getArguments())) {
|
|
|
|
std::tie(newArg, oldArg) = newAndOldArg;
|
2020-10-08 08:12:52 +08:00
|
|
|
auto abiMemref = rewriter.create<refbackrt::ToMemrefOp>(
|
2020-09-17 08:31:40 +08:00
|
|
|
op.getLoc(), getABIMemrefType(oldArg.getType()), newArg);
|
|
|
|
auto memref = rewriter.create<MemRefCastOp>(op.getLoc(), abiMemref,
|
|
|
|
oldArg.getType());
|
|
|
|
rewriter.replaceUsesOfBlockArgument(oldArg, memref);
|
|
|
|
}
|
|
|
|
});
|
2020-07-11 08:31:24 +08:00
|
|
|
return success();
|
|
|
|
}
|
|
|
|
};
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
namespace {
|
2020-10-08 08:12:52 +08:00
|
|
|
// At the return ABI boundaries, convert to !refbackrt.tensor type.
|
2020-09-17 08:31:40 +08:00
|
|
|
// This pattern is needed to trigger the type conversion mechanics to do a
|
|
|
|
// target materialization.
|
|
|
|
class RewriteReturnOp : public OpConversionPattern<ReturnOp> {
|
2020-07-11 08:31:24 +08:00
|
|
|
public:
|
|
|
|
using OpConversionPattern::OpConversionPattern;
|
|
|
|
LogicalResult
|
2020-09-17 08:31:40 +08:00
|
|
|
matchAndRewrite(ReturnOp op, ArrayRef<Value> operands,
|
2020-07-11 08:31:24 +08:00
|
|
|
ConversionPatternRewriter &rewriter) const override {
|
2020-09-17 08:31:40 +08:00
|
|
|
rewriter.replaceOpWithNewOp<ReturnOp>(op, operands);
|
2020-07-11 08:31:24 +08:00
|
|
|
return success();
|
|
|
|
}
|
|
|
|
};
|
|
|
|
} // namespace
|
|
|
|
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
static LogicalResult doDialectConversion(ModuleOp module) {
|
|
|
|
auto *context = module.getContext();
|
|
|
|
|
2020-09-17 08:31:40 +08:00
|
|
|
TypeConverter typeConverter;
|
|
|
|
typeConverter.addConversion([](Type type) { return type; });
|
|
|
|
typeConverter.addConversion([](MemRefType type) {
|
2020-10-08 08:12:52 +08:00
|
|
|
return refbackrt::TensorType::get(type.getContext());
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
});
|
2020-09-17 08:31:40 +08:00
|
|
|
typeConverter.addTargetMaterialization(
|
2020-10-08 08:12:52 +08:00
|
|
|
[](OpBuilder &builder, refbackrt::TensorType type, ValueRange inputs,
|
2020-09-17 08:31:40 +08:00
|
|
|
Location loc) -> Value {
|
|
|
|
assert(inputs.size() == 1);
|
|
|
|
auto abiMemref = builder.create<MemRefCastOp>(
|
|
|
|
loc, inputs[0], getABIMemrefType(inputs[0].getType()));
|
2020-10-08 08:12:52 +08:00
|
|
|
return builder.create<refbackrt::FromMemrefOp>(loc, type, abiMemref);
|
2020-09-17 08:31:40 +08:00
|
|
|
});
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
|
|
|
OwningRewritePatternList patterns;
|
|
|
|
ConversionTarget target(*context);
|
2020-10-08 08:12:52 +08:00
|
|
|
target.addLegalDialect<refbackrt::RefbackrtDialect>();
|
2020-09-17 08:31:40 +08:00
|
|
|
target.addLegalDialect<StandardOpsDialect>();
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
2020-09-17 08:31:40 +08:00
|
|
|
patterns.insert<FuncOpSignatureConversion>(typeConverter, context);
|
|
|
|
target.addDynamicallyLegalOp<FuncOp>(
|
|
|
|
[&](FuncOp op) { return typeConverter.isSignatureLegal(op.getType()); });
|
|
|
|
patterns.insert<RewriteReturnOp>(typeConverter, context);
|
|
|
|
target.addDynamicallyLegalOp<ReturnOp>(
|
|
|
|
[&](ReturnOp op) { return typeConverter.isLegal(op); });
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
|
2020-09-17 08:31:40 +08:00
|
|
|
patterns.insert<LowerAssertOp>(context);
|
|
|
|
target.addIllegalOp<AssertOp>();
|
2020-07-11 08:31:24 +08:00
|
|
|
|
2020-10-30 06:25:55 +08:00
|
|
|
return applyPartialConversion(module, target, std::move(patterns));
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
// This pass lowers the public ABI of the module to the primitives exposed by
|
2020-10-08 08:12:52 +08:00
|
|
|
// the refbackrt dialect.
|
|
|
|
class LowerToRefbackrtABI
|
|
|
|
: public LowerToRefbackrtABIBase<LowerToRefbackrtABI> {
|
2020-09-10 15:23:46 +08:00
|
|
|
void getDependentDialects(DialectRegistry ®istry) const override {
|
2020-10-08 08:12:52 +08:00
|
|
|
registry.insert<refbackrt::RefbackrtDialect>();
|
2020-09-10 15:23:46 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void runOnOperation() override {
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
ModuleOp module = getOperation();
|
|
|
|
|
|
|
|
// Before we lower anything, capture any needed metadata about the argument
|
|
|
|
// lists that will be needed for safely invoking the raw runtime functions
|
|
|
|
// later. (for example, number of expected arguments/results, types,
|
|
|
|
// etc.)
|
|
|
|
if (failed(createModuleMetadata(module)))
|
|
|
|
return signalPassFailure();
|
|
|
|
|
|
|
|
// Now do the actual conversion / lowering.
|
|
|
|
if (failed(doDialectConversion(module)))
|
|
|
|
return signalPassFailure();
|
|
|
|
}
|
|
|
|
};
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
std::unique_ptr<OperationPass<ModuleOp>>
|
2020-10-08 08:12:52 +08:00
|
|
|
mlir::NPCOMP::createLowerToRefbackrtABIPass() {
|
|
|
|
return std::make_unique<LowerToRefbackrtABI>();
|
Rework e2e flow to use new "npcomprt"
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
2020-07-09 08:15:40 +08:00
|
|
|
}
|