X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=src%2Fnnue%2Fnnue_architecture.h;h=725b40fb43d2e87d9ab484603528f9965a06f59d;hb=cb9c2594fcedc881ae8f8bfbfdf130cf89840e4c;hp=91cdc4bda23158b9a418d535cb14063a3b3c2b90;hpb=84f3e867903f62480c33243dd0ecbffd342796fc;p=stockfish

diff --git a/src/nnue/nnue_architecture.h b/src/nnue/nnue_architecture.h
index 91cdc4bd..725b40fb 100644
--- a/src/nnue/nnue_architecture.h
+++ b/src/nnue/nnue_architecture.h
@@ -1,6 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2020 The Stockfish developers (see AUTHORS file)
+  Copyright (C) 2004-2022 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
@@ -21,18 +21,111 @@
 #ifndef NNUE_ARCHITECTURE_H_INCLUDED
 #define NNUE_ARCHITECTURE_H_INCLUDED
 
-// Defines the network structure
-#include "architectures/halfkp_256x2-32-32.h"
+#include "nnue_common.h"
 
-namespace Eval::NNUE {
+#include "features/half_ka_v2_hm.h"
 
-  static_assert(kTransformedFeatureDimensions % kMaxSimdWidth == 0, "");
-  static_assert(Network::kOutputDimensions == 1, "");
-  static_assert(std::is_same<Network::OutputType, std::int32_t>::value, "");
+#include "layers/affine_transform.h"
+#include "layers/clipped_relu.h"
 
-  // Trigger for full calculation instead of difference calculation
-  constexpr auto kRefreshTriggers = RawFeatures::kRefreshTriggers;
+#include "../misc.h"
 
-}  // namespace Eval::NNUE
+namespace Stockfish::Eval::NNUE {
+
+// Input features used in evaluation function
+using FeatureSet = Features::HalfKAv2_hm;
+
+// Number of input feature dimensions after conversion
+constexpr IndexType TransformedFeatureDimensions = 1024;
+constexpr IndexType PSQTBuckets = 8;
+constexpr IndexType LayerStacks = 8;
+
+struct Network
+{
+  static constexpr int FC_0_OUTPUTS = 15;
+  static constexpr int FC_1_OUTPUTS = 32;
+
+  Layers::AffineTransform<TransformedFeatureDimensions, FC_0_OUTPUTS + 1> fc_0;
+  Layers::ClippedReLU<FC_0_OUTPUTS> ac_0;
+  Layers::AffineTransform<FC_0_OUTPUTS, 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) {
+    if (!fc_0.read_parameters(stream)) return false;
+    if (!ac_0.read_parameters(stream)) return false;
+    if (!fc_1.read_parameters(stream)) return false;
+    if (!ac_1.read_parameters(stream)) return false;
+    if (!fc_2.read_parameters(stream)) return false;
+    return true;
+  }
+
+  // Read network parameters
+  bool write_parameters(std::ostream& stream) const {
+    if (!fc_0.write_parameters(stream)) return false;
+    if (!ac_0.write_parameters(stream)) return false;
+    if (!fc_1.write_parameters(stream)) return false;
+    if (!ac_1.write_parameters(stream)) return false;
+    if (!fc_2.write_parameters(stream)) return false;
+    return true;
+  }
+
+  std::int32_t propagate(const TransformedFeatureType* transformedFeatures)
+  {
+    constexpr uint64_t alignment = CacheLineSize;
+
+    struct Buffer
+    {
+      alignas(CacheLineSize) decltype(fc_0)::OutputBuffer fc_0_out;
+      alignas(CacheLineSize) decltype(ac_0)::OutputBuffer ac_0_out;
+      alignas(CacheLineSize) decltype(fc_1)::OutputBuffer fc_1_out;
+      alignas(CacheLineSize) decltype(ac_1)::OutputBuffer ac_1_out;
+      alignas(CacheLineSize) decltype(fc_2)::OutputBuffer fc_2_out;
+    };
+
+#if defined(ALIGNAS_ON_STACK_VARIABLES_BROKEN)
+    char bufferRaw[sizeof(Buffer) + alignment];
+    char* bufferRawAligned = align_ptr_up<alignment>(&bufferRaw[0]);
+    Buffer& buffer = *(new (bufferRawAligned) Buffer);
+#else
+    alignas(alignment) Buffer buffer;
+#endif
+
+    fc_0.propagate(transformedFeatures, buffer.fc_0_out);
+    ac_0.propagate(buffer.fc_0_out, buffer.ac_0_out);
+    fc_1.propagate(buffer.ac_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 = int(buffer.fc_0_out[FC_0_OUTPUTS]) * (600*OutputScale) / (127*(1<<WeightScaleBits));
+    std::int32_t outputValue = buffer.fc_2_out[0] + fwdOut;
+
+#if defined(ALIGNAS_ON_STACK_VARIABLES_BROKEN)
+    buffer.~Buffer();
+#endif
+
+    return outputValue;
+  }
+};
+
+}  // namespace Stockfish::Eval::NNUE
 
 #endif // #ifndef NNUE_ARCHITECTURE_H_INCLUDED