Sparse impl of affine_transform_non_ssse3()

[stockfish] / src / nnue / layers / clipped_relu.h

diff --git a/src/nnue/layers/clipped_relu.h b/src/nnue/layers/clipped_relu.h
index ffd2e3b76a94523eec1beb01d6b67bec122a89d2..aab824b357266079c4d7bce15fe4c86b8d684c9b 100644 (file)
--- a/src/nnue/layers/clipped_relu.h
+++ b/src/nnue/layers/clipped_relu.h
@@ -1,6 +1,6 @@
 /*
   Stockfish, a UCI chess playing engine derived from Glaurung 2.1
-  Copyright (C) 2004-2022 The Stockfish developers (see AUTHORS file)
+  Copyright (C) 2004-2023 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
@@ -21,6 +21,10 @@
 #ifndef NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
 #define NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
 
+#include <algorithm>
+#include <cstdint>
+#include <iosfwd>
+
 #include "../nnue_common.h"
 
 namespace Stockfish::Eval::NNUE::Layers {
@@ -59,7 +63,7 @@ namespace Stockfish::Eval::NNUE::Layers {
     }
 
     // Forward propagation
-    const OutputType* propagate(
+    void propagate(
         const InputType* input, OutputType* output) const {
 
   #if defined(USE_AVX2)
@@ -168,18 +172,8 @@ namespace Stockfish::Eval::NNUE::Layers {
 
       for (IndexType i = Start; i < InputDimensions; ++i) {
         output[i] = static_cast<OutputType>(
-            std::max(0, std::min(127, input[i] >> WeightScaleBits)));
-      }
-
-      // Affine transform layers expect that there is at least
-      // ceil_to_multiple(OutputDimensions, 32) initialized values.
-      // We cannot do this in the affine transform because it requires
-      // preallocating space here.
-      for (IndexType i = OutputDimensions; i < PaddedOutputDimensions; ++i) {
-        output[i] = 0;
+            std::clamp(input[i] >> WeightScaleBits, 0, 127));
       }
-
-      return output;
     }
   };