chromium/third_party/spirv-tools/src/source/opt/code_sink.h

// Copyright (c) 2019 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#ifndef SOURCE_OPT_CODE_SINK_H_
#define SOURCE_OPT_CODE_SINK_H_

#include <unordered_map>

#include "source/opt/ir_context.h"
#include "source/opt/module.h"
#include "source/opt/pass.h"

namespace spvtools {
namespace opt {

// This pass does code sinking for OpAccessChain and OpLoad on variables in
// uniform storage or in read only memory.  Code sinking is a transformation
// where an instruction is moved into a more deeply nested construct.
//
// The goal is to move these instructions as close as possible to their uses
// without having to execute them more often or to replicate the instruction.
// Moving the instruction in this way can lead to shorter live ranges, which can
// lead to less register pressure.  It can also cause instructions to be
// executed less often because they could be moved into one path of a selection
// construct.
//
// This optimization can cause register pressure to rise if the operands of the
// instructions go dead after the instructions being moved. That is why we only
// move certain OpLoad and OpAccessChain instructions.  They generally have
// constants, loop induction variables, and global pointers as operands.  The
// operands are live for a longer time in most cases.
class CodeSinkingPass : public Pass { … };

}  // namespace opt
}  // namespace spvtools

#endif  // SOURCE_OPT_CODE_SINK_H_