* Copyright (c) 2025 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
#include <climits>
#include "ascir/Dialect/Asc/IR/Asc.h"
#include "ascir/Dialect/Asc/Transforms/Passes.h"
#include "ascir/Dialect/Utils/ConstantOpBuilder.h"
#include "mlir/Dialect/Func/IR/FuncOps.h"
#include "mlir/IR/Builders.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/IR/Value.h"
#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
namespace mlir {
namespace ascendc {
#define GEN_PASS_DEF_MATERIALIZETENSOR
#include "ascir/Dialect/Asc/Transforms/Passes.h.inc"
}
}
using namespace mlir;
namespace {
struct MaterializeLocalTensor : OpRewritePattern<ascendc::LocalTensorAutoOp> {
using OpRewritePattern::OpRewritePattern;
static ascendc::TPosition getPosition(ascendc::LocalTensorAutoOp op)
{
if (op.getOutput())
return ascendc::TPosition::VECOUT;
if (op.getInput())
return ascendc::TPosition::VECIN;
llvm_unreachable("position is undefined because tensor cannot be enqueued");
}
LogicalResult matchAndRewrite(ascendc::LocalTensorAutoOp op, PatternRewriter& rewriter) const override
{
auto type = op.getType();
auto loc = op.getLoc();
ascir::ConstantOpBuilder consts(rewriter);
Value length;
if (type.hasStaticShape()) {
length = consts.i64(type.getNumElements() * type.getElementTypeBitWidth() / CHAR_BIT);
} else {
assert(op->getNumOperands() != 0 && "must have operands for dynamic shape");
length = consts.i64(type.getElementTypeBitWidth() / CHAR_BIT);
for (auto dim : op.getDynamicShape()) {
length = rewriter.create<arith::MulIOp>(loc, length, dim);
}
}
Value pipe = rewriter.create<ascendc::PipeOp>(loc);
if (!op.getInput() && !op.getOutput()) {
auto bufferTy = ascendc::TBufType::get(op.getContext(), ascendc::TPosition::VECCALC);
Value buffer = rewriter.create<ascendc::TBufOp>(loc, bufferTy);
rewriter.create<ascendc::TPipeInitBufferOp>(loc, pipe, buffer, length);
rewriter.replaceOpWithNewOp<ascendc::TBufGetTensorOp>(op, type, buffer);
return success();
}
auto queueTy = ascendc::QueueType::get(op.getContext(), getPosition(op), 1);
Value queue = rewriter.create<ascendc::QueueOp>(loc, queueTy);
Value num = consts.i32(1);
rewriter.create<ascendc::TPipeInitQueueOp>(loc, pipe, queue, num, length);
auto allocOp = rewriter.replaceOpWithNewOp<ascendc::TQueBindAllocTensorOp>(op, type, queue);
rewriter.setInsertionPoint(allocOp->getBlock()->getTerminator());
rewriter.create<ascendc::TQueBindFreeTensorOp>(allocOp->getLoc(), queue, allocOp.getTensor());
return success();
}
};
class MaterializeTensorPass : public ascendc::impl::MaterializeTensorBase<MaterializeTensorPass> {
void runOnOperation() override
{
func::FuncOp funcOp = getOperation();
if (funcOp.isDeclaration()) {
return;
}
MLIRContext* context = &getContext();
RewritePatternSet patterns(context);
patterns.add<MaterializeLocalTensor>(context);
if (applyPatternsAndFoldGreedily(funcOp, std::move(patterns)).failed()) {
signalPassFailure();
}
}
};
}
namespace mlir {
namespace ascendc {
std::unique_ptr<Pass> createMaterializeTensorPass() { return std::make_unique<MaterializeTensorPass>(); }
}
}