optimize compilation

OpenNMT · Apr 8, 2024 · d22ca71 · d22ca71
1 parent 845a327
commit d22ca71
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 4 deletions.
diff --git a/include/ctranslate2/ops/flash-attention/softmax.h b/include/ctranslate2/ops/flash-attention/softmax.h
@@ -162,7 +162,7 @@ struct Softmax {
         }
     };
 
-    template<bool Is_dropout=false, bool Split=false, typename Tensor0>
+    template<bool Split=false, typename Tensor0>
     __forceinline__ __device__ TensorT normalize_softmax_lse(Tensor0 &acc_o, float softmax_scale, float rp_dropout=1.0) {
         SumOp<float> sum_op;
         quad_allreduce_(row_sum, row_sum, sum_op);
@@ -174,7 +174,7 @@ struct Softmax {
             float sum = row_sum(mi);
             float inv_sum = (sum == 0.f || sum != sum) ? 1.f : 1.f / sum;
             lse(mi) = (sum == 0.f || sum != sum) ? (Split ? -INFINITY : INFINITY) : row_max(mi) * softmax_scale + __logf(sum);
-            float scale = !Is_dropout ? inv_sum : inv_sum * rp_dropout;
+            float scale = inv_sum;
             #pragma unroll
             for (int ni = 0; ni < size<1>(acc_o_rowcol); ++ni) { acc_o_rowcol(mi, ni) *= scale; }
         }

diff --git a/python/tools/prepare_build_environment_windows.sh b/python/tools/prepare_build_environment_windows.sh
@@ -26,14 +26,14 @@ curl -L -O https://github.com/oneapi-src/oneDNN/archive/refs/tags/v${ONEDNN_VERS
 tar xf *.tar.gz && rm *.tar.gz
 cd oneDNN-*
 cmake -DCMAKE_BUILD_TYPE=Release -DONEDNN_LIBRARY_TYPE=STATIC -DONEDNN_BUILD_EXAMPLES=OFF -DONEDNN_BUILD_TESTS=OFF -DONEDNN_ENABLE_WORKLOAD=INFERENCE -DONEDNN_ENABLE_PRIMITIVE="CONVOLUTION;REORDER" -DONEDNN_BUILD_GRAPH=OFF .
-cmake --build . --config Release --target install --parallel 2
+cmake --build . --config Release --target install --parallel 6
 cd ..
 rm -r oneDNN-*
 
 mkdir build
 cd build
 cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=$CTRANSLATE2_ROOT -DCMAKE_PREFIX_PATH="C:/Program Files (x86)/Intel/oneAPI/compiler/latest/windows/compiler/lib/intel64_win;C:/Program Files (x86)/oneDNN" -DBUILD_CLI=OFF -DWITH_DNNL=ON -DWITH_CUDA=ON -DWITH_CUDNN=ON -DCUDA_TOOLKIT_ROOT_DIR="$CUDA_ROOT" -DCUDA_DYNAMIC_LOADING=ON -DCUDA_NVCC_FLAGS="-Xfatbin=-compress-all" -DCUDA_ARCH_LIST="Common" ..
-cmake --build . --config Release --target install --parallel 2 --verbose
+cmake --build . --config Release --target install --parallel 6 --verbose
 cd ..
 rm -r build