mirror of
https://github.com/official-stockfish/Stockfish.git
synced 2026-07-22 12:47:08 +00:00
anematodeandDisservin
347fcaa392
Passed simp STC LLR: 2.94 (-2.94,2.94) <-1.75,0.25> Total: 29632 W: 7772 L: 7552 D: 14308 Ptnml(0-2): 76, 3155, 8147, 3349, Id expect there to be a small speedup (bc the compiler can skip some stores) but i haven’t tried to measure it closes https://github.com/official-stockfish/Stockfish/pull/6688 No functional change
159 lines
6.6 KiB
C++
159 lines
6.6 KiB
C++
/*
|
|
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
|
|
Copyright (C) 2004-2026 The Stockfish developers (see AUTHORS file)
|
|
|
|
Stockfish is free software: you can redistribute it and/or modify
|
|
it under the terms of the GNU General Public License as published by
|
|
the Free Software Foundation, either version 3 of the License, or
|
|
(at your option) any later version.
|
|
|
|
Stockfish is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
// Input features and network structure used in NNUE evaluation function
|
|
|
|
#ifndef NNUE_ARCHITECTURE_H_INCLUDED
|
|
#define NNUE_ARCHITECTURE_H_INCLUDED
|
|
|
|
#include <cstdint>
|
|
#include <cstring>
|
|
#include <iosfwd>
|
|
|
|
#include "features/half_ka_v2_hm.h"
|
|
#include "features/full_threats.h"
|
|
#include "layers/affine_transform.h"
|
|
#include "layers/affine_transform_sparse_input.h"
|
|
#include "layers/clipped_relu.h"
|
|
#include "layers/sqr_clipped_relu.h"
|
|
#include "nnue_common.h"
|
|
|
|
namespace Stockfish::Eval::NNUE {
|
|
|
|
// Input features used in evaluation function
|
|
using ThreatFeatureSet = Features::FullThreats;
|
|
using PSQFeatureSet = Features::HalfKAv2_hm;
|
|
|
|
// Number of input feature dimensions after conversion
|
|
constexpr IndexType TransformedFeatureDimensionsBig = 1024;
|
|
constexpr int L2Big = 31;
|
|
constexpr int L3Big = 32;
|
|
|
|
constexpr IndexType TransformedFeatureDimensionsSmall = 128;
|
|
constexpr int L2Small = 15;
|
|
constexpr int L3Small = 32;
|
|
|
|
constexpr IndexType PSQTBuckets = 8;
|
|
constexpr IndexType LayerStacks = 8;
|
|
|
|
// If vector instructions are enabled, we update and refresh the
|
|
// accumulator tile by tile such that each tile fits in the CPU's
|
|
// vector registers.
|
|
static_assert(PSQTBuckets % 8 == 0,
|
|
"Per feature PSQT values cannot be processed at granularity lower than 8 at a time.");
|
|
|
|
template<IndexType L1, int L2, int L3>
|
|
struct NetworkArchitecture {
|
|
static constexpr IndexType TransformedFeatureDimensions = L1;
|
|
static constexpr int FC_0_OUTPUTS = L2;
|
|
static constexpr int FC_1_OUTPUTS = L3;
|
|
|
|
Layers::AffineTransformSparseInput<TransformedFeatureDimensions, FC_0_OUTPUTS + 1> fc_0;
|
|
Layers::SqrClippedReLU<FC_0_OUTPUTS + 1> ac_sqr_0;
|
|
Layers::ClippedReLU<FC_0_OUTPUTS + 1> ac_0;
|
|
Layers::AffineTransform<FC_0_OUTPUTS * 2, FC_1_OUTPUTS> fc_1;
|
|
Layers::ClippedReLU<FC_1_OUTPUTS> ac_1;
|
|
Layers::AffineTransform<FC_1_OUTPUTS, 1> fc_2;
|
|
|
|
// Hash value embedded in the evaluation file
|
|
static constexpr std::uint32_t get_hash_value() {
|
|
// input slice hash
|
|
std::uint32_t hashValue = 0xEC42E90Du;
|
|
hashValue ^= TransformedFeatureDimensions * 2;
|
|
|
|
hashValue = decltype(fc_0)::get_hash_value(hashValue);
|
|
hashValue = decltype(ac_0)::get_hash_value(hashValue);
|
|
hashValue = decltype(fc_1)::get_hash_value(hashValue);
|
|
hashValue = decltype(ac_1)::get_hash_value(hashValue);
|
|
hashValue = decltype(fc_2)::get_hash_value(hashValue);
|
|
|
|
return hashValue;
|
|
}
|
|
|
|
// Read network parameters
|
|
bool read_parameters(std::istream& stream) {
|
|
return fc_0.read_parameters(stream) && ac_0.read_parameters(stream)
|
|
&& fc_1.read_parameters(stream) && ac_1.read_parameters(stream)
|
|
&& fc_2.read_parameters(stream);
|
|
}
|
|
|
|
// Write network parameters
|
|
bool write_parameters(std::ostream& stream) const {
|
|
return fc_0.write_parameters(stream) && ac_0.write_parameters(stream)
|
|
&& fc_1.write_parameters(stream) && ac_1.write_parameters(stream)
|
|
&& fc_2.write_parameters(stream);
|
|
}
|
|
|
|
std::int32_t propagate(const TransformedFeatureType* transformedFeatures) const {
|
|
struct alignas(CacheLineSize) Buffer {
|
|
alignas(CacheLineSize) typename decltype(fc_0)::OutputBuffer fc_0_out;
|
|
alignas(CacheLineSize) typename decltype(ac_sqr_0)::OutputType
|
|
ac_sqr_0_out[ceil_to_multiple<IndexType>(FC_0_OUTPUTS * 2, 32)];
|
|
alignas(CacheLineSize) typename decltype(ac_0)::OutputBuffer ac_0_out;
|
|
alignas(CacheLineSize) typename decltype(fc_1)::OutputBuffer fc_1_out;
|
|
alignas(CacheLineSize) typename decltype(ac_1)::OutputBuffer ac_1_out;
|
|
alignas(CacheLineSize) typename decltype(fc_2)::OutputBuffer fc_2_out;
|
|
|
|
Buffer() { std::memset(ac_sqr_0_out, 0, sizeof(ac_sqr_0_out)); }
|
|
};
|
|
|
|
Buffer buffer;
|
|
|
|
fc_0.propagate(transformedFeatures, buffer.fc_0_out);
|
|
ac_sqr_0.propagate(buffer.fc_0_out, buffer.ac_sqr_0_out);
|
|
ac_0.propagate(buffer.fc_0_out, buffer.ac_0_out);
|
|
std::memcpy(buffer.ac_sqr_0_out + FC_0_OUTPUTS, buffer.ac_0_out,
|
|
FC_0_OUTPUTS * sizeof(typename decltype(ac_0)::OutputType));
|
|
fc_1.propagate(buffer.ac_sqr_0_out, buffer.fc_1_out);
|
|
ac_1.propagate(buffer.fc_1_out, buffer.ac_1_out);
|
|
fc_2.propagate(buffer.ac_1_out, buffer.fc_2_out);
|
|
|
|
// buffer.fc_0_out[FC_0_OUTPUTS] is such that 1.0 is equal to 127*(1<<WeightScaleBits) in
|
|
// quantized form, but we want 1.0 to be equal to 600*OutputScale
|
|
std::int32_t fwdOut =
|
|
(buffer.fc_0_out[FC_0_OUTPUTS]) * (600 * OutputScale) / (127 * (1 << WeightScaleBits));
|
|
std::int32_t outputValue = buffer.fc_2_out[0] + fwdOut;
|
|
|
|
return outputValue;
|
|
}
|
|
|
|
std::size_t get_content_hash() const {
|
|
std::size_t h = 0;
|
|
hash_combine(h, fc_0.get_content_hash());
|
|
hash_combine(h, ac_sqr_0.get_content_hash());
|
|
hash_combine(h, ac_0.get_content_hash());
|
|
hash_combine(h, fc_1.get_content_hash());
|
|
hash_combine(h, ac_1.get_content_hash());
|
|
hash_combine(h, fc_2.get_content_hash());
|
|
hash_combine(h, get_hash_value());
|
|
return h;
|
|
}
|
|
};
|
|
|
|
} // namespace Stockfish::Eval::NNUE
|
|
|
|
template<Stockfish::Eval::NNUE::IndexType L1, int L2, int L3>
|
|
struct std::hash<Stockfish::Eval::NNUE::NetworkArchitecture<L1, L2, L3>> {
|
|
std::size_t
|
|
operator()(const Stockfish::Eval::NNUE::NetworkArchitecture<L1, L2, L3>& arch) const noexcept {
|
|
return arch.get_content_hash();
|
|
}
|
|
};
|
|
|
|
#endif // #ifndef NNUE_ARCHITECTURE_H_INCLUDED
|