Merge branch 'master' into patch-3

Author: jordimas

.github/workflows/ci.yml

@@ -109,7 +109,7 @@ jobs:
             -DBUILD_TESTS=ON \
             .
           make -j $(nproc) install
-            
+
       - name: Build Ruy
         if: matrix.backend == 'ruy'
         run: |
@@ -121,7 +121,7 @@ jobs:
             $CMAKE_EXTRA_OPTIONS \
             .
           make -j $(nproc) install
- 
+
       - name: Download test data
         run: |
           wget https://opennmt-models.s3.amazonaws.com/transliteration-aren-all.tar.gz
@@ -229,7 +229,7 @@ jobs:
           ls -l
           find .
           pip install ${{ matrix.wheel_pattern }}
-                
+
       - name: Test Python wheel
         run: |
           pytest -v python/tests/ --ignore=python/tests/test_opennmt_tf.py
@@ -295,6 +295,29 @@ jobs:
         with:
           submodules: recursive
 
+      - name: Show disk and docker usage (before cleanup)
+        run: |
+          df -h
+          echo " -= Docker System =-"
+          docker system df || true
+
+      - name: Free disk space (cleanup heavy preinstalled directories + docker prune)
+        run: |
+          echo " -= Removing big preinstalled directories (shouldn't remove the needed tools) =-"
+          sudo rm -rf /opt/hostedtoolcache || true
+          sudo rm -rf /usr/share/dotnet || true
+          sudo rm -rf /usr/lib/jvm || true
+          sudo rm -rf /usr/local/lib/android || true
+          echo " -= Running docker prune =-"
+          docker system prune -af --volumes || true
+          docker builder prune -af || true
+
+      - name: Show disk and docker usage (after cleanup)
+        run: |
+          df -h
+          echo " -= Docker System =-"
+          docker system df || true
+
       - name: Build Docker images
         run: |
           ./docker/build_all.sh

src/cuda/primitives.cu

@@ -503,7 +503,7 @@ namespace ctranslate2 {
                               beta_ptr,
                               c, CUDA_R_16F, ldc,
                               compute_type,
-                              CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+                              CUBLAS_GEMM_DEFAULT));
   }
 
   template<>
@@ -528,7 +528,7 @@ namespace ctranslate2 {
                               &beta,
                               c, CUDA_R_16BF, ldc,
                               CUDA_R_32F,
-                              CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+                              CUBLAS_GEMM_DEFAULT));
   }
 
   template<>
@@ -556,7 +556,7 @@ namespace ctranslate2 {
                               &beta_i,
                               c, CUDA_R_32I, ldc,
                               CUDA_R_32I,
-                              CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+                              CUBLAS_GEMM_DEFAULT));
   }
 
   template<>
@@ -617,7 +617,7 @@ namespace ctranslate2 {
                                             c, CUDA_R_16F, ldc, stridec,
                                             batch_size,
                                             compute_type,
-                                            CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+                                            CUBLAS_GEMM_DEFAULT));
   }
 
   template<>
@@ -642,7 +642,7 @@ namespace ctranslate2 {
                                             c, CUDA_R_16BF, ldc, stridec,
                                             batch_size,
                                             CUDA_R_32F,
-                                            CUBLAS_GEMM_DEFAULT_TENSOR_OP));
+                                            CUBLAS_GEMM_DEFAULT));
   }
 
   template <typename T>