SPIRV-Cross/spirv_cfg.hpp
Hans-Kristian Arntzen a489ba7fd1 Reduce pressure on global allocation.
- Replace ostringstream with custom implementation.
  ~30% performance uplift on vector-shuffle-oom test.
  Allocations are measurably reduced in Valgrind.

- Replace std::vector with SmallVector.
  Classic malloc optimization, small vectors are backed by inline data.
  ~ 7-8% gain on vector-shuffle-oom on GCC 8 on Linux.

- Use an object pool for IVariant type.
  We generally allocate a lot of SPIR* objects. We can amortize these
  allocations neatly by pooling them.

- ~15% overall uplift on ./test_shaders.py --iterations 10000 shaders/.
2019-04-09 15:09:44 +02:00

150 lines
3.2 KiB
C++

/*
* Copyright 2016-2019 Arm Limited
*
* 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 SPIRV_CROSS_CFG_HPP
#define SPIRV_CROSS_CFG_HPP
#include "spirv_common.hpp"
#include <assert.h>
namespace SPIRV_CROSS_NAMESPACE
{
class Compiler;
class CFG
{
public:
CFG(Compiler &compiler, const SPIRFunction &function);
Compiler &get_compiler()
{
return compiler;
}
const Compiler &get_compiler() const
{
return compiler;
}
const SPIRFunction &get_function() const
{
return func;
}
uint32_t get_immediate_dominator(uint32_t block) const
{
auto itr = immediate_dominators.find(block);
if (itr != std::end(immediate_dominators))
return itr->second;
else
return 0;
}
uint32_t get_visit_order(uint32_t block) const
{
auto itr = visit_order.find(block);
assert(itr != std::end(visit_order));
int v = itr->second.get();
assert(v > 0);
return uint32_t(v);
}
uint32_t find_common_dominator(uint32_t a, uint32_t b) const;
const SmallVector<uint32_t> &get_preceding_edges(uint32_t block) const
{
auto itr = preceding_edges.find(block);
if (itr != std::end(preceding_edges))
return itr->second;
else
return empty_vector;
}
const SmallVector<uint32_t> &get_succeeding_edges(uint32_t block) const
{
auto itr = succeeding_edges.find(block);
if (itr != std::end(succeeding_edges))
return itr->second;
else
return empty_vector;
}
template <typename Op>
void walk_from(std::unordered_set<uint32_t> &seen_blocks, uint32_t block, const Op &op) const
{
if (seen_blocks.count(block))
return;
seen_blocks.insert(block);
op(block);
for (auto b : get_succeeding_edges(block))
walk_from(seen_blocks, b, op);
}
private:
struct VisitOrder
{
int &get()
{
return v;
}
const int &get() const
{
return v;
}
int v = -1;
};
Compiler &compiler;
const SPIRFunction &func;
std::unordered_map<uint32_t, SmallVector<uint32_t>> preceding_edges;
std::unordered_map<uint32_t, SmallVector<uint32_t>> succeeding_edges;
std::unordered_map<uint32_t, uint32_t> immediate_dominators;
std::unordered_map<uint32_t, VisitOrder> visit_order;
SmallVector<uint32_t> post_order;
SmallVector<uint32_t> empty_vector;
void add_branch(uint32_t from, uint32_t to);
void build_post_order_visit_order();
void build_immediate_dominators();
bool post_order_visit(uint32_t block);
uint32_t visit_count = 0;
bool is_back_edge(uint32_t to) const;
};
class DominatorBuilder
{
public:
DominatorBuilder(const CFG &cfg);
void add_block(uint32_t block);
uint32_t get_dominator() const
{
return dominator;
}
void lift_continue_block_dominator();
private:
const CFG &cfg;
uint32_t dominator = 0;
};
} // namespace SPIRV_CROSS_NAMESPACE
#endif