Convert another FDIV to FMUL

marian-nmt · Sep 17, 2024 · d4d66c8 · d4d66c8
1 parent 471fe24
commit d4d66c8
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/src/tensors/cpu/tensor_operators.cpp b/src/tensors/cpu/tensor_operators.cpp
@@ -1295,11 +1295,11 @@ void RMSNormalizationImpl(float* out,
       sqSum += sp[i] * sp[i];
     }
 
-    float rms = std::sqrt(sqSum / cols + eps);
+    float invRms = 1.0/std::sqrt(sqSum / cols + eps);
 
     #pragma omp simd
     for(int i = 0; i < cols; ++i) {
-      float t = alpha[alphaStride * i] * (sp[i] / rms);
+      float t = alpha[alphaStride * i] * (sp[i] * invRms);
       if(hasBeta)
         t += beta[betaStride * i];