frigate/docker/rocm/migraphx/eliminate_contiguous.cpp

196 lines
6.6 KiB
C++
Raw Normal View History

2025-02-03 22:53:47 +03:00
/*
* The MIT License (MIT)
*
* Copyright (c) 2015-2024 Advanced Micro Devices, Inc. All rights reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include <migraphx/eliminate_contiguous.hpp>
#include <migraphx/program.hpp>
#include <migraphx/instruction.hpp>
#include <migraphx/iterator_for.hpp>
#include <migraphx/ranges.hpp>
#include <migraphx/stringutils.hpp>
#include <migraphx/op/contiguous.hpp>
#include <migraphx/op/identity.hpp>
#include <migraphx/par_for.hpp>
#include <utility>
namespace migraphx {
inline namespace MIGRAPHX_INLINE_NS {
MIGRAPHX_DECLARE_ENV_VAR(MIGRAPHX_TRACE_ELIMINATE_CONTIGUOUS)
static bool try_compute_shape(instruction_ref ins,
const std::vector<shape>& inputs,
const std::vector<module_ref>& mods)
{
try
{
shape new_shape = ins->get_operator().compute_shape(inputs, mods);
// Cannot tell if a dynamic shape will need to be made contiguous
if(new_shape.dynamic())
{
return false;
}
// If the output shape is a standard shape, no need to try its output
if(new_shape.standard())
{
return true;
}
// if no changes for the shape, the contiguous can also be removed
if(new_shape == ins->get_shape())
{
return true;
}
auto outputs = ins->outputs();
// If the current instruction has no output, it means it is the last
// instruction and generates a non-standard output shape, and the last
// output shape is different from the case with the contiguous operator
if(outputs.empty())
{
return false;
}
for(auto output : outputs)
{
auto args = output->inputs();
std::vector<shape> input_shapes(args.size());
std::transform(args.begin(), args.end(), input_shapes.begin(), [&](auto& arg) {
return (arg == ins) ? new_shape : arg->get_shape();
});
if(not try_compute_shape(output, input_shapes, output->module_inputs()))
{
return false;
}
}
}
catch(const std::exception& e)
{
if(enabled(MIGRAPHX_TRACE_ELIMINATE_CONTIGUOUS{}))
{
std::cout << "Exception: " << e.what() << std::endl;
}
return false;
}
catch(...)
{
if(enabled(MIGRAPHX_TRACE_ELIMINATE_CONTIGUOUS{}))
{
std::cout << "Unknown exception" << std::endl;
}
return false;
}
return true;
}
static bool try_compute_shape(instruction_ref ins,
const std::vector<instruction_ref>& args,
const std::vector<module_ref>& mods)
{
auto inputs = to_shapes(args);
return try_compute_shape(ins, inputs, mods);
}
template <class F>
static void remove_contiguous(const std::string& op_name, module& m, F f)
{
auto last = std::prev(m.end());
std::vector<instruction_ref> const_instructions;
for(auto ins : iterator_for(m))
{
// return instruction should have inputs with standard shape
if(ins->name() == "@return")
continue;
if(ins != last and ins->outputs().empty())
continue;
if(not f(ins))
continue;
auto args = ins->inputs();
auto mod_args = ins->module_inputs();
for(auto arg : ins->inputs())
{
if(arg->name() != op_name)
continue;
if(enabled(MIGRAPHX_TRACE_ELIMINATE_CONTIGUOUS{}))
{
std::cout << "eliminate_contiguous: ";
m.debug_print(ins);
}
auto prev = arg->inputs().front();
// create copy of args each time as they are modified inside the loop
auto new_args = ins->inputs();
replace(new_args, arg, prev);
if(try_compute_shape(ins, new_args, mod_args))
{
instruction::replace_argument(ins, arg, prev);
}
else if(prev->can_eval())
{
const_instructions.push_back(arg);
}
}
}
// Perform static contiguous evaluations in parallel
std::vector<argument> literals(const_instructions.size());
par_for(const_instructions.size(), 1, [&](const auto i) {
auto c = op::contiguous{};
auto prev = const_instructions[i]->inputs().front();
// compute the output contiguous shape from the previous instruction shape
shape computed_shape = c.compute_shape({prev->get_shape()});
const std::vector<argument>& prev_eval = {prev->eval()};
// prev_eval should not be used in make_compute_output_shape() as computed_shape is static
auto co_shape = make_compute_output_shape(pack(c, computed_shape, prev_eval));
literals[i] = c.compute(co_shape, prev_eval);
});
// Replace static contiguous operations with a literal
for(size_t i = 0; i < const_instructions.size(); i++)
{
auto l = m.add_literal(literals[i].get_shape(), literals[i].data());
m.replace_instruction(const_instructions[i], l);
}
}
void eliminate_contiguous::apply(module& m) const
{
// Skip contiguous from splits first
remove_contiguous(op_name, m, [](auto ins) {
if(ins->name() != "slice")
return true;
return (ins->inputs().front()->outputs().size() == 1);
});
remove_contiguous(op_name, m, [](auto) { return true; });
}
} // namespace MIGRAPHX_INLINE_NS
} // namespace migraphx