Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Appearance settings

Commit 08723a4

Browse filesBrowse files
rosborosbo
authored
Build CUDA 11 TensorFlow wheels (Kaggle#982)
Starting with CUDA 11.0, toolkit components are versioned independently: https://docs.nvidia.com/cuda/archive/11.0/cuda-toolkit-release-notes/index.html#cuda-major-component-versions http://b/175341540
1 parent 9c870aa commit 08723a4
Copy full SHA for 08723a4

File tree

Expand file treeCollapse file tree

2 files changed

+23
-24
lines changed
Open diff view settings
Filter options
Expand file treeCollapse file tree

2 files changed

+23
-24
lines changed
Open diff view settings
Collapse file

‎tensorflow-whl/CHANGELOG.md‎

Copy file name to clipboardExpand all lines: tensorflow-whl/CHANGELOG.md
+5-4Lines changed: 5 additions & 4 deletions
  • Display the source diff
  • Display the rich diff
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
* `1.11.0-py36`: TensorFlow 1.11.0 wheels built with python 3.6
2-
* `1.12.0-py36`: TensorFlow 1.12.0 wheels with Cuda 9.2
3-
* `1.13.1-py36`: TensorFlow 1.13.1 wheels with Cuda 10.0
4-
* `1.13.1-py36-2`: TensorFlow 1.13.1 wheels with Cuda 10.0 and bump anaconda3 base image version to 5.3.0
2+
* `1.12.0-py36`: TensorFlow 1.12.0 wheels with CUDA 9.2
3+
* `1.13.1-py36`: TensorFlow 1.13.1 wheels with CUDA 10.0
4+
* `1.13.1-py36-2`: TensorFlow 1.13.1 wheels with CUDA 10.0 and bump anaconda3 base image version to 5.3.0
55
* `1.13.1-py37`: TensorFlow 1.13.1 with Python 3.7.0 and bump anaconda3 base image version to 5.3.0
66
* `1.13.1-py37-2`: TensorFlow 1.13.1 with Python 3.7.3
77
* `1.14.0-py36`: TensorFlow 1.14.0 with Python 3.6
@@ -22,4 +22,5 @@
2222
* `2.4.0-py37`: TensorFlow 2.4.0 & Python 3.7 & TFA 0.12
2323
* `2.4.0-py37-2`: TensorFlow 2.4.0 & m61 base image
2424
* `2.4.0-py37-3`: TensorFlow 2.4.0 & 7.0 CUDA compute capability
25-
* `2.4.1-py37`: TensorFlow 2.4.1 & TFA 0.12.1
25+
* `2.4.1-py37`: TensorFlow 2.4.1 & TFA 0.12.1
26+
* `2.4.1-py37-2`: TensorFlow 2.4.1 & CUDA 11.0
Collapse file

‎tensorflow-whl/Dockerfile‎

Copy file name to clipboardExpand all lines: tensorflow-whl/Dockerfile
+18-20Lines changed: 18 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
FROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04 AS nvidia
1+
FROM nvidia/cuda:11.0-cudnn8-devel-ubuntu18.04 AS nvidia
22
FROM gcr.io/deeplearning-platform-release/base-cpu:m66
33

44
# Avoid interactive configuration prompts/dialogs during apt-get.
@@ -17,11 +17,9 @@ RUN sed -i 's/deb https:\/\/developer.download.nvidia.com/deb http:\/\/developer
1717

1818
# Ensure the cuda libraries are compatible with the GPU image.
1919
# TODO(b/120050292): Use templating to keep in sync.
20-
ENV CUDA_MAJOR_VERSION=10
21-
ENV CUDA_MINOR_VERSION=2
22-
ENV CUDA_PATCH_VERSION=89
23-
ENV CUDA_VERSION=$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION.$CUDA_PATCH_VERSION
24-
ENV CUDA_PKG_VERSION=$CUDA_MAJOR_VERSION-$CUDA_MINOR_VERSION=$CUDA_VERSION-1
20+
ENV CUDA_MAJOR_VERSION=11
21+
ENV CUDA_MINOR_VERSION=0
22+
ENV CUDA_VERSION=$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION
2523
LABEL com.nvidia.volumes.needed="nvidia_driver"
2624
LABEL com.nvidia.cuda.version="${CUDA_VERSION}"
2725
ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:${PATH}
@@ -35,19 +33,19 @@ ENV NVIDIA_VISIBLE_DEVICES=all
3533
ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
3634
ENV NVIDIA_REQUIRE_CUDA="cuda>=$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION"
3735
RUN apt-get update && apt-get install -y --no-install-recommends \
38-
cuda-cupti-$CUDA_PKG_VERSION \
39-
cuda-cudart-$CUDA_PKG_VERSION \
40-
cuda-cudart-dev-$CUDA_PKG_VERSION \
41-
cuda-libraries-$CUDA_PKG_VERSION \
42-
cuda-libraries-dev-$CUDA_PKG_VERSION \
43-
cuda-nvml-dev-$CUDA_PKG_VERSION \
44-
cuda-minimal-build-$CUDA_PKG_VERSION \
45-
cuda-command-line-tools-$CUDA_PKG_VERSION \
46-
libcudnn7=7.6.5.32-1+cuda$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION \
47-
libcudnn7-dev=7.6.5.32-1+cuda$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION \
48-
libnccl2=2.5.6-1+cuda$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION \
49-
libnccl-dev=2.5.6-1+cuda$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION && \
50-
ln -s /usr/local/cuda-$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION /usr/local/cuda && \
36+
cuda-cupti-$CUDA_VERSION \
37+
cuda-cudart-$CUDA_VERSION \
38+
cuda-cudart-dev-$CUDA_VERSION \
39+
cuda-libraries-$CUDA_VERSION \
40+
cuda-libraries-dev-$CUDA_VERSION \
41+
cuda-nvml-dev-$CUDA_VERSION \
42+
cuda-minimal-build-$CUDA_VERSION \
43+
cuda-command-line-tools-$CUDA_VERSION \
44+
libcudnn8=8.0.4.30-1+cuda$CUDA_VERSION \
45+
libcudnn8-dev=8.0.4.30-1+cuda$CUDA_VERSION \
46+
libnccl2=2.7.8-1+cuda$CUDA_VERSION \
47+
libnccl-dev=2.7.8-1+cuda$CUDA_VERSION && \
48+
ln -s /usr/local/cuda-$CUDA_VERSION /usr/local/cuda && \
5149
ln -s /usr/local/cuda/lib64/stubs/libcuda.so /usr/local/cuda/lib64/stubs/libcuda.so.1
5250

5351
# Use Bazelisk to ensure the proper bazel version is used.
@@ -99,7 +97,7 @@ ENV TF_NEED_CUDA=1
9997
ENV TF_CUDA_VERSION=$CUDA_MAJOR_VERSION.$CUDA_MINOR_VERSION
10098
# 3.7 (K80), 6.0 (P100), 7.0 (V100), 7.5 (T4): https://developer.nvidia.com/cuda-gpus
10199
ENV TF_CUDA_COMPUTE_CAPABILITIES=3.7,6.0,7.0,7.5
102-
ENV TF_CUDNN_VERSION=7
100+
ENV TF_CUDNN_VERSION=8
103101
ENV TF_NCCL_VERSION=2
104102
ENV NCCL_INSTALL_PATH=/usr/
105103

0 commit comments

Comments
0 (0)
Morty Proxy This is a proxified and sanitized view of the page, visit original site.