mirror of https://github.com/llvm/torch-mlir
95 lines
3.6 KiB
C++
95 lines
3.6 KiB
C++
//===------------------------------------------------------------*- C++ -*-===//
|
|
//
|
|
// This file is licensed 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 "npcomp/JITRuntime/JITModule.h"
|
|
#include "mlir/ExecutionEngine/CRunnerUtils.h"
|
|
#include "mlir/ExecutionEngine/OptUtils.h"
|
|
#include "npcomp/E2E/E2E.h"
|
|
#include "llvm/Support/InitLLVM.h"
|
|
#include "llvm/Support/TargetSelect.h"
|
|
|
|
using namespace npcomp;
|
|
using namespace mlir;
|
|
using llvm::Error;
|
|
using llvm::Expected;
|
|
using llvm::StringError;
|
|
using llvm::Twine;
|
|
|
|
/// Wrap a string into an llvm::StringError.
|
|
static Error make_string_error(const Twine &message) {
|
|
return llvm::make_error<StringError>(message.str(),
|
|
llvm::inconvertibleErrorCode());
|
|
}
|
|
|
|
static LogicalResult runE2ELowering(ModuleOp module) {
|
|
PassManager pm(module.getContext(), /*verifyPasses=*/true);
|
|
applyPassManagerCLOptions(pm);
|
|
|
|
NPCOMP::E2ELoweringPipelineOptions options;
|
|
NPCOMP::createE2ELoweringPipeline(pm, options);
|
|
return pm.run(module);
|
|
}
|
|
|
|
JITModule::JITModule() {}
|
|
|
|
llvm::Expected<std::unique_ptr<JITModule>>
|
|
JITModule::fromMLIR(mlir::ModuleOp module,
|
|
llvm::ArrayRef<llvm::StringRef> sharedLibs) {
|
|
if (failed(runE2ELowering(module)))
|
|
return make_string_error("could not lower module");
|
|
|
|
auto expectedEngine = ExecutionEngine::create(
|
|
module, [](llvm::Module *) { return Error::success(); },
|
|
/*jitCodeGenOptLevel=*/llvm::None, llvm::to_vector<6>(sharedLibs));
|
|
if (!expectedEngine)
|
|
return expectedEngine.takeError();
|
|
std::unique_ptr<JITModule> ret(new JITModule);
|
|
ret->engine = std::move(*expectedEngine);
|
|
// Here we abuse mlir::ExecutionEngine a bit. It technically returns a
|
|
// function pointer, but here we look up a module descriptor.
|
|
auto expectedAddress = ret->engine->lookup("__npcomp_module_descriptor");
|
|
if (!expectedAddress)
|
|
return expectedAddress.takeError();
|
|
ret->descriptor =
|
|
reinterpret_cast<npcomprt::ModuleDescriptor *>(*expectedAddress);
|
|
return ret;
|
|
}
|
|
|
|
// Converter for bridging to npcomprt llvm-lookalike data structures.
|
|
static npcomprt::StringRef toNpcomprt(llvm::StringRef s) {
|
|
return npcomprt::StringRef(s.data(), s.size());
|
|
}
|
|
|
|
template <typename T>
|
|
static npcomprt::ArrayRef<T> toNpcomprt(llvm::ArrayRef<T> a) {
|
|
return npcomprt::ArrayRef<T>(a.data(), a.size());
|
|
}
|
|
|
|
template <typename T>
|
|
static npcomprt::MutableArrayRef<T> toNpcomprt(llvm::MutableArrayRef<T> a) {
|
|
return npcomprt::MutableArrayRef<T>(a.data(), a.size());
|
|
}
|
|
|
|
llvm::Expected<llvm::SmallVector<npcomprt::Ref<npcomprt::Tensor>, 6>>
|
|
JITModule::invoke(llvm::StringRef functionName,
|
|
llvm::ArrayRef<npcomprt::Ref<npcomprt::Tensor>> inputs) {
|
|
npcomprt::FunctionMetadata metadata;
|
|
if (npcomprt::failed(npcomprt::getMetadata(
|
|
descriptor, toNpcomprt(functionName), metadata)))
|
|
return make_string_error("unknown function: " + Twine(functionName));
|
|
SmallVector<npcomprt::Ref<npcomprt::Tensor>, 6> outputs(metadata.numOutputs);
|
|
if (metadata.numInputs != static_cast<std::int32_t>(inputs.size()))
|
|
return make_string_error("invoking '" + Twine(functionName) +
|
|
"': expected " + Twine(metadata.numInputs) +
|
|
" inputs");
|
|
npcomprt::invoke(
|
|
descriptor, toNpcomprt(functionName), toNpcomprt(inputs),
|
|
toNpcomprt(llvm::makeMutableArrayRef(outputs.data(), outputs.size())));
|
|
return outputs;
|
|
}
|