CUDA 11.6 and cudnn related changes (#5803)

Co-authored-by: Nikita Shulga <nshulga@fb.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

.circleci/config.yml

@@ -1141,6 +1141,12 @@ workflows:
           name: binary_linux_wheel_py3.7_cu115
           python_version: '3.7'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_wheel_py3.7_cu116
+          python_version: '3.7'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           name: binary_linux_wheel_py3.7_rocm4.5.2
@@ -1175,6 +1181,12 @@ workflows:
           name: binary_linux_wheel_py3.8_cu115
           python_version: '3.8'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_wheel_py3.8_cu116
+          python_version: '3.8'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           name: binary_linux_wheel_py3.8_rocm4.5.2
@@ -1209,6 +1221,12 @@ workflows:
           name: binary_linux_wheel_py3.9_cu115
           python_version: '3.9'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_wheel_py3.9_cu116
+          python_version: '3.9'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           name: binary_linux_wheel_py3.9_rocm4.5.2
@@ -1243,6 +1261,12 @@ workflows:
           name: binary_linux_wheel_py3.10_cu115
           python_version: '3.10'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_wheel_py3.10_cu116
+          python_version: '3.10'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           name: binary_linux_wheel_py3.10_rocm4.5.2
@@ -1304,6 +1328,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_wheel_py3.7_cu115
           python_version: '3.7'
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_wheel_py3.7_cu116
+          python_version: '3.7'
       - binary_win_wheel:
           cu_version: cpu
           filters:
@@ -1331,6 +1364,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_wheel_py3.8_cu115
           python_version: '3.8'
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_wheel_py3.8_cu116
+          python_version: '3.8'
       - binary_win_wheel:
           cu_version: cpu
           filters:
@@ -1358,6 +1400,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_wheel_py3.9_cu115
           python_version: '3.9'
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_wheel_py3.9_cu116
+          python_version: '3.9'
       - binary_win_wheel:
           cu_version: cpu
           name: binary_win_wheel_py3.10_cpu
@@ -1373,8 +1424,17 @@ workflows:
           python_version: '3.10'
       - binary_win_wheel:
           cu_version: cu115
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_wheel_py3.10_cu115
           python_version: '3.10'
+      - binary_win_wheel:
+          cu_version: cu116
+          name: binary_win_wheel_py3.10_cu116
+          python_version: '3.10'
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -1399,6 +1459,12 @@ workflows:
           name: binary_linux_conda_py3.7_cu115
           python_version: '3.7'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_conda_py3.7_cu116
+          python_version: '3.7'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -1423,6 +1489,12 @@ workflows:
           name: binary_linux_conda_py3.8_cu115
           python_version: '3.8'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_conda_py3.8_cu116
+          python_version: '3.8'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -1447,6 +1519,12 @@ workflows:
           name: binary_linux_conda_py3.9_cu115
           python_version: '3.9'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_conda_py3.9_cu116
+          python_version: '3.9'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -1471,6 +1549,12 @@ workflows:
           name: binary_linux_conda_py3.10_cu115
           python_version: '3.10'
           wheel_docker_image: pytorch/manylinux-cuda115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          name: binary_linux_conda_py3.10_cu116
+          python_version: '3.10'
+          wheel_docker_image: pytorch/manylinux-cuda116
       - binary_macos_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -1522,6 +1606,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_conda_py3.7_cu115
           python_version: '3.7'
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_conda_py3.7_cu116
+          python_version: '3.7'
       - binary_win_conda:
           cu_version: cpu
           filters:
@@ -1549,6 +1642,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_conda_py3.8_cu115
           python_version: '3.8'
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_conda_py3.8_cu116
+          python_version: '3.8'
       - binary_win_conda:
           cu_version: cpu
           filters:
@@ -1576,6 +1678,15 @@ workflows:
               only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_conda_py3.9_cu115
           python_version: '3.9'
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: binary_win_conda_py3.9_cu116
+          python_version: '3.9'
       - binary_win_conda:
           cu_version: cpu
           name: binary_win_conda_py3.10_cpu
@@ -1591,8 +1702,17 @@ workflows:
           python_version: '3.10'
       - binary_win_conda:
           cu_version: cu115
+          filters:
+            branches:
+              only: main
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
           name: binary_win_conda_py3.10_cu115
           python_version: '3.10'
+      - binary_win_conda:
+          cu_version: cu116
+          name: binary_win_conda_py3.10_cu116
+          python_version: '3.10'
       - build_docs:
           filters:
             branches:
@@ -1897,6 +2017,28 @@ workflows:
           requires:
           - nightly_binary_linux_wheel_py3.7_cu115
           subfolder: cu115/
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.7_cu116
+          python_version: '3.7'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.7_cu116_upload
+          requires:
+          - nightly_binary_linux_wheel_py3.7_cu116
+          subfolder: cu116/
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           filters:
@@ -2027,6 +2169,28 @@ workflows:
           requires:
           - nightly_binary_linux_wheel_py3.8_cu115
           subfolder: cu115/
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.8_cu116
+          python_version: '3.8'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.8_cu116_upload
+          requires:
+          - nightly_binary_linux_wheel_py3.8_cu116
+          subfolder: cu116/
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           filters:
@@ -2157,6 +2321,28 @@ workflows:
           requires:
           - nightly_binary_linux_wheel_py3.9_cu115
           subfolder: cu115/
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.9_cu116
+          python_version: '3.9'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.9_cu116_upload
+          requires:
+          - nightly_binary_linux_wheel_py3.9_cu116
+          subfolder: cu116/
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           filters:
@@ -2287,6 +2473,28 @@ workflows:
           requires:
           - nightly_binary_linux_wheel_py3.10_cu115
           subfolder: cu115/
+      - binary_linux_wheel:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.10_cu116
+          python_version: '3.10'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_wheel_py3.10_cu116_upload
+          requires:
+          - nightly_binary_linux_wheel_py3.10_cu116
+          subfolder: cu116/
       - binary_linux_wheel:
           cu_version: rocm4.5.2
           filters:
@@ -2477,6 +2685,26 @@ workflows:
           requires:
           - nightly_binary_win_wheel_py3.7_cu115
           subfolder: cu115/
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.7_cu116
+          python_version: '3.7'
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.7_cu116_upload
+          requires:
+          - nightly_binary_win_wheel_py3.7_cu116
+          subfolder: cu116/
       - binary_win_wheel:
           cu_version: cpu
           filters:
@@ -2537,6 +2765,26 @@ workflows:
           requires:
           - nightly_binary_win_wheel_py3.8_cu115
           subfolder: cu115/
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.8_cu116
+          python_version: '3.8'
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.8_cu116_upload
+          requires:
+          - nightly_binary_win_wheel_py3.8_cu116
+          subfolder: cu116/
       - binary_win_wheel:
           cu_version: cpu
           filters:
@@ -2597,6 +2845,26 @@ workflows:
           requires:
           - nightly_binary_win_wheel_py3.9_cu115
           subfolder: cu115/
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.9_cu116
+          python_version: '3.9'
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.9_cu116_upload
+          requires:
+          - nightly_binary_win_wheel_py3.9_cu116
+          subfolder: cu116/
       - binary_win_wheel:
           cu_version: cpu
           filters:
@@ -2657,6 +2925,26 @@ workflows:
           requires:
           - nightly_binary_win_wheel_py3.10_cu115
           subfolder: cu115/
+      - binary_win_wheel:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.10_cu116
+          python_version: '3.10'
+      - binary_wheel_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_wheel_py3.10_cu116_upload
+          requires:
+          - nightly_binary_win_wheel_py3.10_cu116
+          subfolder: cu116/
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -2741,6 +3029,27 @@ workflows:
           name: nightly_binary_linux_conda_py3.7_cu115_upload
           requires:
           - nightly_binary_linux_conda_py3.7_cu115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.7_cu116
+          python_version: '3.7'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.7_cu116_upload
+          requires:
+          - nightly_binary_linux_conda_py3.7_cu116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -2825,6 +3134,27 @@ workflows:
           name: nightly_binary_linux_conda_py3.8_cu115_upload
           requires:
           - nightly_binary_linux_conda_py3.8_cu115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.8_cu116
+          python_version: '3.8'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.8_cu116_upload
+          requires:
+          - nightly_binary_linux_conda_py3.8_cu116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -2909,6 +3239,27 @@ workflows:
           name: nightly_binary_linux_conda_py3.9_cu115_upload
           requires:
           - nightly_binary_linux_conda_py3.9_cu115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.9_cu116
+          python_version: '3.9'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.9_cu116_upload
+          requires:
+          - nightly_binary_linux_conda_py3.9_cu116
       - binary_linux_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -2993,6 +3344,27 @@ workflows:
           name: nightly_binary_linux_conda_py3.10_cu115_upload
           requires:
           - nightly_binary_linux_conda_py3.10_cu115
+      - binary_linux_conda:
+          conda_docker_image: pytorch/conda-builder:cuda116
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.10_cu116
+          python_version: '3.10'
+          wheel_docker_image: pytorch/manylinux-cuda116
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_linux_conda_py3.10_cu116_upload
+          requires:
+          - nightly_binary_linux_conda_py3.10_cu116
       - binary_macos_conda:
           conda_docker_image: pytorch/conda-builder:cpu
           cu_version: cpu
@@ -3134,6 +3506,25 @@ workflows:
           name: nightly_binary_win_conda_py3.7_cu115_upload
           requires:
           - nightly_binary_win_conda_py3.7_cu115
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.7_cu116
+          python_version: '3.7'
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.7_cu116_upload
+          requires:
+          - nightly_binary_win_conda_py3.7_cu116
       - binary_win_conda:
           cu_version: cpu
           filters:
@@ -3191,6 +3582,25 @@ workflows:
           name: nightly_binary_win_conda_py3.8_cu115_upload
           requires:
           - nightly_binary_win_conda_py3.8_cu115
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.8_cu116
+          python_version: '3.8'
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.8_cu116_upload
+          requires:
+          - nightly_binary_win_conda_py3.8_cu116
       - binary_win_conda:
           cu_version: cpu
           filters:
@@ -3248,6 +3658,25 @@ workflows:
           name: nightly_binary_win_conda_py3.9_cu115_upload
           requires:
           - nightly_binary_win_conda_py3.9_cu115
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.9_cu116
+          python_version: '3.9'
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.9_cu116_upload
+          requires:
+          - nightly_binary_win_conda_py3.9_cu116
       - binary_win_conda:
           cu_version: cpu
           filters:
@@ -3305,6 +3734,25 @@ workflows:
           name: nightly_binary_win_conda_py3.10_cu115_upload
           requires:
           - nightly_binary_win_conda_py3.10_cu115
+      - binary_win_conda:
+          cu_version: cu116
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.10_cu116
+          python_version: '3.10'
+      - binary_conda_upload:
+          context: org-member
+          filters:
+            branches:
+              only: nightly
+            tags:
+              only: /v[0-9]+(\.[0-9]+)*-rc[0-9]+/
+          name: nightly_binary_win_conda_py3.10_cu116_upload
+          requires:
+          - nightly_binary_win_conda_py3.10_cu116
   docker_build:
     triggers:
       - schedule:

.circleci/regenerate.py

@@ -32,8 +32,8 @@ def build_workflows(prefix="", filter_branch=None, upload=False, indentation=6,
         for os_type in ["linux", "macos", "win"]:
             python_versions = PYTHON_VERSIONS
             cu_versions_dict = {
-                "linux": ["cpu", "cu102", "cu113", "cu115", "rocm4.5.2", "rocm5.0"],
-                "win": ["cpu", "cu113", "cu115"],
+                "linux": ["cpu", "cu102", "cu113", "cu115", "cu116", "rocm4.5.2", "rocm5.0"],
+                "win": ["cpu", "cu113", "cu115", "cu116"],
                 "macos": ["cpu"],
             }
             cu_versions = cu_versions_dict[os_type]
@@ -122,14 +122,10 @@ def upload_doc_job(filter_branch):
 
 
 manylinux_images = {
-    "cu92": "pytorch/manylinux-cuda92",
-    "cu101": "pytorch/manylinux-cuda101",
     "cu102": "pytorch/manylinux-cuda102",
-    "cu110": "pytorch/manylinux-cuda110",
-    "cu111": "pytorch/manylinux-cuda111",
-    "cu112": "pytorch/manylinux-cuda112",
     "cu113": "pytorch/manylinux-cuda113",
     "cu115": "pytorch/manylinux-cuda115",
+    "cu116": "pytorch/manylinux-cuda116",
 }
 
 

packaging/build_conda.sh

@@ -12,9 +12,9 @@ setup_conda_cudatoolkit_constraint
 setup_visual_studio_constraint
 setup_junit_results_folder
 
-# nvidia channel included for cudatoolkit >= 11 however for 11.5 we use conda-forge
+# nvidia channel included for cudatoolkit >= 11 however for 11.5 and 11.6 we use conda-forge
 export CUDATOOLKIT_CHANNEL="nvidia"
-if [[ "$CU_VERSION" == cu115 ]]; then
+if [[ "$CU_VERSION" == cu115 || "$CU_VERSION" == cu116 ]]; then
     export CUDATOOLKIT_CHANNEL="conda-forge"
 fi
 

packaging/pkg_helpers.bash

@@ -46,6 +46,14 @@ setup_cuda() {
 
   # Now work out the CUDA settings
   case "$CU_VERSION" in
+    cu116)
+      if [[ "$OSTYPE" == "msys" ]]; then
+        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.6"
+      else
+        export CUDA_HOME=/usr/local/cuda-11.6/
+      fi
+      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5;8.0;8.6"
+      ;;
     cu115)
       if [[ "$OSTYPE" == "msys" ]]; then
         export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.5"
@@ -62,30 +70,6 @@ setup_cuda() {
       fi
       export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5;8.0;8.6"
       ;;
-    cu112)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.2"
-      else
-        export CUDA_HOME=/usr/local/cuda-11.2/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5;8.0;8.6"
-      ;;
-    cu111)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.1"
-      else
-        export CUDA_HOME=/usr/local/cuda-11.1/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5;8.0;8.6"
-      ;;
-    cu110)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v11.0"
-      else
-        export CUDA_HOME=/usr/local/cuda-11.0/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5;8.0"
-      ;;
     cu102)
       if [[ "$OSTYPE" == "msys" ]]; then
         export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.2"
@@ -94,30 +78,6 @@ setup_cuda() {
       fi
       export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5"
       ;;
-    cu101)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1"
-      else
-        export CUDA_HOME=/usr/local/cuda-10.1/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5"
-      ;;
-    cu100)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.0"
-      else
-        export CUDA_HOME=/usr/local/cuda-10.0/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0;7.5"
-      ;;
-    cu92)
-      if [[ "$OSTYPE" == "msys" ]]; then
-        export CUDA_HOME="C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v9.2"
-      else
-        export CUDA_HOME=/usr/local/cuda-9.2/
-      fi
-      export TORCH_CUDA_ARCH_LIST="3.5;5.0+PTX;6.0;7.0"
-      ;;
     cpu)
       ;;
     rocm*)
@@ -289,33 +249,18 @@ setup_conda_cudatoolkit_constraint() {
     export CONDA_BUILD_VARIANT="cpu"
   else
     case "$CU_VERSION" in
+      cu116)
+        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.6,<11.7 # [not osx]"
+        ;;
       cu115)
         export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.5,<11.6 # [not osx]"
         ;;
       cu113)
         export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.3,<11.4 # [not osx]"
         ;;
-      cu112)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.2,<11.3 # [not osx]"
-        ;;
-      cu111)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.1,<11.2 # [not osx]"
-        ;;
-      cu110)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=11.0,<11.1 # [not osx]"
-        ;;
       cu102)
         export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=10.2,<10.3 # [not osx]"
         ;;
-      cu101)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=10.1,<10.2 # [not osx]"
-        ;;
-      cu100)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=10.0,<10.1 # [not osx]"
-        ;;
-      cu92)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="- cudatoolkit >=9.2,<9.3 # [not osx]"
-        ;;
       cpu)
         export CONDA_CUDATOOLKIT_CONSTRAINT=""
         export CONDA_BUILD_VARIANT="cpu"
@@ -336,30 +281,18 @@ setup_conda_cudatoolkit_plain_constraint() {
     export CMAKE_USE_CUDA=0
   else
     case "$CU_VERSION" in
+      cu116)
+        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=11.6"
+        ;;
       cu115)
         export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=11.5"
         ;;
       cu113)
         export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=11.3"
         ;;
-      cu112)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=11.2"
-        ;;
-      cu111)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=11.1"
-        ;;
       cu102)
         export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=10.2"
         ;;
-      cu101)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=10.1"
-        ;;
-      cu100)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=10.0"
-        ;;
-      cu92)
-        export CONDA_CUDATOOLKIT_CONSTRAINT="cudatoolkit=9.2"
-        ;;
       cpu)
         export CONDA_CUDATOOLKIT_CONSTRAINT=""
         export CONDA_BUILD_VARIANT="cpu"

packaging/windows/internal/cuda_install.bat

@@ -20,72 +20,17 @@ set /a CUDA_VER=%CU_VERSION:cu=%
 set CUDA_VER_MAJOR=%CUDA_VER:~0,-1%
 set CUDA_VER_MINOR=%CUDA_VER:~-1,1%
 set CUDA_VERSION_STR=%CUDA_VER_MAJOR%.%CUDA_VER_MINOR%
+set CUDNN_FOLDER="cuda"
+set CUDNN_LIB_FOLDER="lib\x64"
 
-
-if %CUDA_VER% EQU 92 goto cuda92
-if %CUDA_VER% EQU 100 goto cuda100
-if %CUDA_VER% EQU 101 goto cuda101
 if %CUDA_VER% EQU 102 goto cuda102
-if %CUDA_VER% EQU 110 goto cuda110
-if %CUDA_VER% EQU 111 goto cuda111
-if %CUDA_VER% EQU 112 goto cuda112
 if %CUDA_VER% EQU 113 goto cuda113
 if %CUDA_VER% EQU 115 goto cuda115
-
+if %CUDA_VER% EQU 116 goto cuda116
 
 echo CUDA %CUDA_VERSION_STR% is not supported
 exit /b 1
 
-:cuda92
-if not exist "%SRC_DIR%\temp_build\cuda_9.2.148_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/win2016/cuda_9.2.148_win10.exe --output "%SRC_DIR%\temp_build\cuda_9.2.148_win10.exe"
-    if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_9.2.148_win10.exe"
-    set "ARGS=nvcc_9.2 cuobjdump_9.2 nvprune_9.2 cupti_9.2 cublas_9.2 cublas_dev_9.2 cudart_9.2 cufft_9.2 cufft_dev_9.2 curand_9.2 curand_dev_9.2 cusolver_9.2 cusolver_dev_9.2 cusparse_9.2 cusparse_dev_9.2 nvgraph_9.2 nvgraph_dev_9.2 npp_9.2 npp_dev_9.2 nvrtc_9.2 nvrtc_dev_9.2 nvml_dev_9.2"
-)
-
-if not exist "%SRC_DIR%\temp_build\cudnn-9.2-windows10-x64-v7.2.1.38.zip" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/win2016/cudnn-9.2-windows10-x64-v7.2.1.38.zip --output "%SRC_DIR%\temp_build\cudnn-9.2-windows10-x64-v7.2.1.38.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-9.2-windows10-x64-v7.2.1.38.zip"
-)
-
-goto cuda_common
-
-:cuda100
-
-if not exist "%SRC_DIR%\temp_build\cuda_10.0.130_411.31_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/win2016/cuda_10.0.130_411.31_win10.exe --output "%SRC_DIR%\temp_build\cuda_10.0.130_411.31_win10.exe"
-    if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_10.0.130_411.31_win10.exe"
-    set "ARGS=nvcc_10.0 cuobjdump_10.0 nvprune_10.0 cupti_10.0 cublas_10.0 cublas_dev_10.0 cudart_10.0 cufft_10.0 cufft_dev_10.0 curand_10.0 curand_dev_10.0 cusolver_10.0 cusolver_dev_10.0 cusparse_10.0 cusparse_dev_10.0 nvgraph_10.0 nvgraph_dev_10.0 npp_10.0 npp_dev_10.0 nvrtc_10.0 nvrtc_dev_10.0 nvml_dev_10.0"
-)
-
-if not exist "%SRC_DIR%\temp_build\cudnn-10.0-windows10-x64-v7.4.1.5.zip" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/win2016/cudnn-10.0-windows10-x64-v7.4.1.5.zip --output "%SRC_DIR%\temp_build\cudnn-10.0-windows10-x64-v7.4.1.5.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-10.0-windows10-x64-v7.4.1.5.zip"
-)
-
-goto cuda_common
-
-:cuda101
-
-if not exist "%SRC_DIR%\temp_build\cuda_10.1.243_426.00_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cuda_10.1.243_426.00_win10.exe --output "%SRC_DIR%\temp_build\cuda_10.1.243_426.00_win10.exe"
-    if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_10.1.243_426.00_win10.exe"
-    set "ARGS=nvcc_10.1 cuobjdump_10.1 nvprune_10.1 cupti_10.1 cublas_10.1 cublas_dev_10.1 cudart_10.1 cufft_10.1 cufft_dev_10.1 curand_10.1 curand_dev_10.1 cusolver_10.1 cusolver_dev_10.1 cusparse_10.1 cusparse_dev_10.1 nvgraph_10.1 nvgraph_dev_10.1 npp_10.1 npp_dev_10.1 nvjpeg_10.1 nvjpeg_dev_10.1 nvrtc_10.1 nvrtc_dev_10.1 nvml_dev_10.1"
-)
-
-if not exist "%SRC_DIR%\temp_build\cudnn-10.1-windows10-x64-v7.6.4.38.zip" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cudnn-10.1-windows10-x64-v7.6.4.38.zip --output "%SRC_DIR%\temp_build\cudnn-10.1-windows10-x64-v7.6.4.38.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-10.1-windows10-x64-v7.6.4.38.zip"
-)
-
-goto cuda_common
-
 :cuda102
 
 if not exist "%SRC_DIR%\temp_build\cuda_10.2.89_441.22_win10.exe" (
@@ -112,92 +57,81 @@ echo Installing GPU driver DLLs
 
 goto cuda_common
 
-:cuda110
-
-if not exist "%SRC_DIR%\temp_build\cuda_11.0.2_451.48_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cuda_11.0.2_451.48_win10.exe --output "%SRC_DIR%\temp_build\cuda_11.0.2_451.48_win10.exe"
-    if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_11.0.2_451.48_win10.exe"
-    set "ARGS=nvcc_11.0 cuobjdump_11.0 nvprune_11.0 nvprof_11.0 cupti_11.0 cublas_11.0 cublas_dev_11.0 cudart_11.0 cufft_11.0 cufft_dev_11.0 curand_11.0 curand_dev_11.0 cusolver_11.0 cusolver_dev_11.0 cusparse_11.0 cusparse_dev_11.0 npp_11.0 npp_dev_11.0 nvjpeg_11.0 nvjpeg_dev_11.0 nvrtc_11.0 nvrtc_dev_11.0 nvml_dev_11.0"
-)
-
-if not exist "%SRC_DIR%\temp_build\cudnn-11.0-windows-x64-v8.0.4.30.zip" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cudnn-11.0-windows-x64-v8.0.4.30.zip --output "%SRC_DIR%\temp_build\cudnn-11.0-windows-x64-v8.0.4.30.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-11.0-windows-x64-v8.0.4.30.zip"
-)
-
-goto cuda_common
-
-:cuda111
+:cuda113
 
-if not exist "%SRC_DIR%\temp_build\cuda_11.1.1_456.81_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cuda_11.1.1_456.81_win10.exe --output "%SRC_DIR%\temp_build\cuda_11.1.1_456.81_win10.exe"
+set CUDA_INSTALL_EXE=cuda_11.3.0_465.89_win10.exe
+if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
+    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_11.1.1_456.81_win10.exe"
-    set "ARGS=nvcc_11.1 cuobjdump_11.1 nvprune_11.1 nvprof_11.1 cupti_11.1 cublas_11.1 cublas_dev_11.1 cudart_11.1 cufft_11.1 cufft_dev_11.1 curand_11.1 curand_dev_11.1 cusolver_11.1 cusolver_dev_11.1 cusparse_11.1 cusparse_dev_11.1 npp_11.1 npp_dev_11.1 nvjpeg_11.1 nvjpeg_dev_11.1 nvrtc_11.1 nvrtc_dev_11.1 nvml_dev_11.1"
-)
+    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
+    set "ARGS=thrust_11.3 nvcc_11.3 cuobjdump_11.3 nvprune_11.3 nvprof_11.3 cupti_11.3 cublas_11.3 cublas_dev_11.3 cudart_11.3 cufft_11.3 cufft_dev_11.3 curand_11.3 curand_dev_11.3 cusolver_11.3 cusolver_dev_11.3 cusparse_11.3 cusparse_dev_11.3 npp_11.3 npp_dev_11.3 nvjpeg_11.3 nvjpeg_dev_11.3 nvrtc_11.3 nvrtc_dev_11.3 nvml_dev_11.3"
 
-if not exist "%SRC_DIR%\temp_build\cudnn-11.1-windows-x64-v8.0.5.39.zip" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cudnn-11.1-windows-x64-v8.0.5.39.zip --output "%SRC_DIR%\temp_build\cudnn-11.1-windows-x64-v8.0.5.39.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-11.1-windows-x64-v8.0.5.39.zip"
 )
 
-goto cuda_common
-
-:cuda112
-
-if not exist "%SRC_DIR%\temp_build\cuda_11.2.0_460.89_win10.exe" (
-    curl -k -L https://ossci-windows.s3.amazonaws.com/cuda_11.2.0_460.89_win10.exe --output "%SRC_DIR%\temp_build\cuda_11.2.0_460.89_win10.exe"
+set CUDNN_INSTALL_ZIP=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive.zip
+set CUDNN_FOLDER=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive
+set CUDNN_LIB_FOLDER="lib"
+if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
-    set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\cuda_11.2.0_460.89_win10.exe"
-    set "ARGS=nvcc_11.2 cuobjdump_11.2 nvprune_11.2 nvprof_11.2 cupti_11.2 cublas_11.2 cublas_dev_11.2 cudart_11.2 cufft_11.2 cufft_dev_11.2 curand_11.2 curand_dev_11.2 cusolver_11.2 cusolver_dev_11.2 cusparse_11.2 cusparse_dev_11.2 npp_11.2 npp_dev_11.2 nvjpeg_11.2 nvjpeg_dev_11.2 nvrtc_11.2 nvrtc_dev_11.2 nvml_dev_11.2"
-)
+    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
 
-if not exist "%SRC_DIR%\temp_build\cudnn-11.2-windows-x64-v8.1.0.77.zip" (
-    curl -k -L http://s3.amazonaws.com/ossci-windows/cudnn-11.2-windows-x64-v8.1.0.77.zip --output "%SRC_DIR%\temp_build\cudnn-11.2-windows-x64-v8.1.0.77.zip"
-    if errorlevel 1 exit /b 1
-    set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\cudnn-11.2-windows-x64-v8.1.0.77.zip"
+    rem Make sure windows path contains zlib dll
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/zlib123dllx64.zip" --output "%SRC_DIR%\temp_build\zlib123dllx64.zip"
+    7z x "%SRC_DIR%\temp_build\zlib123dllx64.zip" -o"%SRC_DIR%\temp_build\zlib"
+    xcopy /Y "%SRC_DIR%\temp_build\zlib\dll_x64\*.dll" "C:\Windows\System32"
 )
 
 goto cuda_common
 
-:cuda113
+:cuda115
 
-set CUDA_INSTALL_EXE=cuda_11.3.0_465.89_win10.exe
+set CUDA_INSTALL_EXE=cuda_11.5.0_496.13_win10.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
     curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
-    set "ARGS=thrust_11.3 nvcc_11.3 cuobjdump_11.3 nvprune_11.3 nvprof_11.3 cupti_11.3 cublas_11.3 cublas_dev_11.3 cudart_11.3 cufft_11.3 cufft_dev_11.3 curand_11.3 curand_dev_11.3 cusolver_11.3 cusolver_dev_11.3 cusparse_11.3 cusparse_dev_11.3 npp_11.3 npp_dev_11.3 nvjpeg_11.3 nvjpeg_dev_11.3 nvrtc_11.3 nvrtc_dev_11.3 nvml_dev_11.3"
-
+    set "ARGS=thrust_11.5 nvcc_11.5 cuobjdump_11.5 nvprune_11.5 nvprof_11.5 cupti_11.5 cublas_11.5 cublas_dev_11.5 cudart_11.5 cufft_11.5 cufft_dev_11.5 curand_11.5 curand_dev_11.5 cusolver_11.5 cusolver_dev_11.5 cusparse_11.5 cusparse_dev_11.5 npp_11.5 npp_dev_11.5 nvrtc_11.5 nvrtc_dev_11.5 nvml_dev_11.5"
 )
 
-set CUDNN_INSTALL_ZIP=cudnn-11.3-windows-x64-v8.2.0.53.zip
+set CUDNN_INSTALL_ZIP=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive.zip
+set CUDNN_FOLDER=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive
+set CUDNN_LIB_FOLDER="lib"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
     curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
+
+    rem Make sure windows path contains zlib dll
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/zlib123dllx64.zip" --output "%SRC_DIR%\temp_build\zlib123dllx64.zip"
+    7z x "%SRC_DIR%\temp_build\zlib123dllx64.zip" -o"%SRC_DIR%\temp_build\zlib"
+    xcopy /Y "%SRC_DIR%\temp_build\zlib\dll_x64\*.dll" "C:\Windows\System32"
 )
 
 goto cuda_common
 
-:cuda115
+:cuda116
 
-set CUDA_INSTALL_EXE=cuda_11.5.0_496.13_win10.exe
+set CUDA_INSTALL_EXE=cuda_11.6.0_511.23_windows.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
     curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
-    set "ARGS=thrust_11.5 nvcc_11.5 cuobjdump_11.5 nvprune_11.5 nvprof_11.5 cupti_11.5 cublas_11.5 cublas_dev_11.5 cudart_11.5 cufft_11.5 cufft_dev_11.5 curand_11.5 curand_dev_11.5 cusolver_11.5 cusolver_dev_11.5 cusparse_11.5 cusparse_dev_11.5 npp_11.5 npp_dev_11.5 nvrtc_11.5 nvrtc_dev_11.5 nvml_dev_11.5"
+    set "ARGS=thrust_11.6 nvcc_11.6 cuobjdump_11.6 nvprune_11.6 nvprof_11.6 cupti_11.6 cublas_11.6 cublas_dev_11.6 cudart_11.6 cufft_11.6 cufft_dev_11.6 curand_11.6 curand_dev_11.6 cusolver_11.6 cusolver_dev_11.6 cusparse_11.6 cusparse_dev_11.6 npp_11.6 npp_dev_11.6 nvrtc_11.6 nvrtc_dev_11.6 nvml_dev_11.6"
 )
 
-set CUDNN_INSTALL_ZIP=cudnn-11.3-windows-x64-v8.2.0.53.zip
+set CUDNN_INSTALL_ZIP=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive.zip
+set CUDNN_FOLDER=cudnn-windows-x86_64-8.3.2.44_cuda11.5-archive
+set CUDNN_LIB_FOLDER="lib"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
     curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
+
+    rem Make sure windows path contains zlib dll
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/zlib123dllx64.zip" --output "%SRC_DIR%\temp_build\zlib123dllx64.zip"
+    7z x "%SRC_DIR%\temp_build\zlib123dllx64.zip" -o"%SRC_DIR%\temp_build\zlib"
+    xcopy /Y "%SRC_DIR%\temp_build\zlib\dll_x64\*.dll" "C:\Windows\System32"
 )
 
 goto cuda_common
@@ -256,9 +190,9 @@ if not exist "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_ST
 
 echo Installing cuDNN...
 7z x %CUDNN_SETUP_FILE% -o"%SRC_DIR%\temp_build\cudnn"
-xcopy /Y "%SRC_DIR%\temp_build\cudnn\cuda\bin\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\bin"
-xcopy /Y "%SRC_DIR%\temp_build\cudnn\cuda\lib\x64\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\lib\x64"
-xcopy /Y "%SRC_DIR%\temp_build\cudnn\cuda\include\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\include"
+xcopy /Y "%SRC_DIR%\temp_build\cudnn\%CUDNN_FOLDER%\bin\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\bin"
+xcopy /Y "%SRC_DIR%\temp_build\cudnn\%CUDNN_FOLDER%\%CUDNN_LIB_FOLDER%\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\lib\x64"
+xcopy /Y "%SRC_DIR%\temp_build\cudnn\%CUDNN_FOLDER%\include\*.*" "%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v%CUDA_VERSION_STR%\include"
 
 echo Cleaning temp files
 rd /s /q "%SRC_DIR%\temp_build" || ver > nul