blob: 6519b65cec4651262efa520d08a86973be5063d1 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
|
//===- GlobalIdRewriter.cpp - Implementation of GlobalId rewriting -------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file implements in-dialect rewriting of the global_id op for archs
// where global_id.x = threadId.x + blockId.x * blockDim.x
//
//===----------------------------------------------------------------------===//
#include "mlir/Dialect/GPU/IR/GPUDialect.h"
#include "mlir/Dialect/GPU/Transforms/Passes.h"
#include "mlir/Dialect/Index/IR/IndexOps.h"
#include "mlir/IR/PatternMatch.h"
using namespace mlir;
namespace {
struct GpuGlobalIdRewriter : public OpRewritePattern<gpu::GlobalIdOp> {
using OpRewritePattern<gpu::GlobalIdOp>::OpRewritePattern;
LogicalResult matchAndRewrite(gpu::GlobalIdOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
auto dim = op.getDimension();
auto blockId = gpu::BlockIdOp::create(rewriter, loc, dim);
auto blockDim = gpu::BlockDimOp::create(rewriter, loc, dim);
// Compute blockId.x * blockDim.x
auto tmp = index::MulOp::create(rewriter, op.getLoc(), blockId, blockDim);
auto threadId = gpu::ThreadIdOp::create(rewriter, loc, dim);
// Compute threadId.x + blockId.x * blockDim.x
rewriter.replaceOpWithNewOp<index::AddOp>(op, threadId, tmp);
return success();
}
};
} // namespace
void mlir::populateGpuGlobalIdPatterns(RewritePatternSet &patterns) {
patterns.add<GpuGlobalIdRewriter>(patterns.getContext());
}
|
