Optimize update_accumulator_refresh_cache()

mstembera · vondele · commit e608eab8dd9f · 2024-05-13T07:32:32.000+02:00
STC https://tests.stockfishchess.org/tests/view/664105df26ac5f9b286d30e6 LLR: 2.94 (-2.94,2.94) <0.00,2.00> Total: 178528 W: 46235 L: 45750 D: 86543 Ptnml(0-2): 505, 17792, 52142, 18363, 462 Combo of two yellow speedups https://tests.stockfishchess.org/tests/view/6640abf9d163897c63214f5c LLR: -2.93 (-2.94,2.94) <0.00,2.00> Total: 355744 W: 91714 L: 91470 D: 172560 Ptnml(0-2): 913, 36233, 103384, 36381, 961 https://tests.stockfishchess.org/tests/view/6628ce073fe04ce4cefc739c LLR: -2.93 (-2.94,2.94) <0.00,2.00> Total: 627040 W: 162001 L: 161339 D: 303700 Ptnml(0-2): 2268, 72379, 163532, 73105, 2236 closes #5239 No functional change
diff --git a/src/nnue/nnue_feature_transformer.h b/src/nnue/nnue_feature_transformer.h
@@ -664,7 +664,11 @@ class FeatureTransformer {
 
         for (IndexType j = 0; j < HalfDimensions / TileHeight; ++j)
         {
-            auto entryTile = reinterpret_cast<vec_t*>(&entry.accumulation[j * TileHeight]);
+            auto accTile =
+              reinterpret_cast<vec_t*>(&accumulator.accumulation[Perspective][j * TileHeight]);
+            auto entryTile =
+              reinterpret_cast<vec_t*>(&entry.accumulation[j * TileHeight]);
+
             for (IndexType k = 0; k < NumRegs; ++k)
                 acc[k] = entryTile[k];
 
@@ -679,7 +683,7 @@ class FeatureTransformer {
                 auto            columnA = reinterpret_cast<const vec_t*>(&weights[offsetA]);
 
                 for (unsigned k = 0; k < NumRegs; ++k)
-                    acc[k] = vec_add_16(vec_sub_16(acc[k], columnR[k]), columnA[k]);
+                    acc[k] = vec_add_16(acc[k], vec_sub_16(columnA[k], columnR[k]));
             }
             for (; i < int(removed.size()); ++i)
             {
@@ -702,12 +706,17 @@ class FeatureTransformer {
 
             for (IndexType k = 0; k < NumRegs; k++)
                 vec_store(&entryTile[k], acc[k]);
+            for (IndexType k = 0; k < NumRegs; k++)
+                vec_store(&accTile[k], acc[k]);
         }
 
         for (IndexType j = 0; j < PSQTBuckets / PsqtTileHeight; ++j)
         {
-            auto entryTilePsqt =
-              reinterpret_cast<psqt_vec_t*>(&entry.psqtAccumulation[j * PsqtTileHeight]);
+            auto accTilePsqt = reinterpret_cast<psqt_vec_t*>(
+              &accumulator.psqtAccumulation[Perspective][j * PsqtTileHeight]);
+            auto entryTilePsqt = reinterpret_cast<psqt_vec_t*>(
+              &entry.psqtAccumulation[j * PsqtTileHeight]);
+            
             for (std::size_t k = 0; k < NumPsqtRegs; ++k)
                 psqt[k] = entryTilePsqt[k];
 
@@ -732,6 +741,8 @@ class FeatureTransformer {
 
             for (std::size_t k = 0; k < NumPsqtRegs; ++k)
                 vec_store_psqt(&entryTilePsqt[k], psqt[k]);
+            for (std::size_t k = 0; k < NumPsqtRegs; ++k)
+                vec_store_psqt(&accTilePsqt[k], psqt[k]);
         }
 
 #else
@@ -755,8 +766,6 @@ class FeatureTransformer {
                 entry.psqtAccumulation[k] += psqtWeights[index * PSQTBuckets + k];
         }
 
-#endif
-
         // The accumulator of the refresh entry has been updated.
         // Now copy its content to the actual accumulator we were refreshing
 
@@ -765,6 +774,7 @@ class FeatureTransformer {
 
         std::memcpy(accumulator.psqtAccumulation[Perspective], entry.psqtAccumulation,
                     sizeof(int32_t) * PSQTBuckets);
+#endif
 
         for (Color c : {WHITE, BLACK})
             entry.byColorBB[c] = pos.pieces(c);
