Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Fix build issue with USE_CUDNN=0 #11470

Merged
merged 23 commits into from
Jul 12, 2018
Merged

Fix build issue with USE_CUDNN=0 #11470

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
23 commits
Select commit Hold shift + click to select a range
6961a20
Fix build issue with CUDNN=0
anirudh2290 Jun 28, 2018
7bdf6c7
Fix nocudnn func name
anirudh2290 Jun 28, 2018
06815b4
Remove python2 tests
anirudh2290 Jun 28, 2018
def0842
Remove CPP package test
anirudh2290 Jun 28, 2018
e7e6084
Check assert raises when cudnn disabled for op tests on gpu
anirudh2290 Jul 2, 2018
a488e43
Merge with master
anirudh2290 Jul 2, 2018
0631682
Add line
anirudh2290 Jul 2, 2018
2d11ade
Remove whitespace
anirudh2290 Jul 2, 2018
f338866
add decorator for other ops
anirudh2290 Jul 2, 2018
2c7453a
Fix tests
anirudh2290 Jul 5, 2018
a5c2c79
Add and remove assert
anirudh2290 Jul 5, 2018
71d7d72
Fix op and common
anirudh2290 Jul 5, 2018
65d01e0
Fix merge issue
anirudh2290 Jul 5, 2018
4ce776c
Remove C API
anirudh2290 Jul 10, 2018
556dee1
Fix
anirudh2290 Jul 10, 2018
67d00a3
Fix lint
anirudh2290 Jul 10, 2018
266ba86
Add init git
anirudh2290 Jul 10, 2018
e837777
Rename CUDNN_DISABLED env variable
anirudh2290 Jul 10, 2018
2f590ee
Add a runtime function for nocudnn
anirudh2290 Jul 10, 2018
a2fc5f5
Resolve conflicts
anirudh2290 Jul 11, 2018
749981d
Remove MXCudnnIsenabled
anirudh2290 Jul 11, 2018
9573913
Add comment for disabled test
anirudh2290 Jul 11, 2018
e8d87dd
Add full link in comment
anirudh2290 Jul 11, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 39 additions & 0 deletions Jenkinsfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -281,6 +281,17 @@ try {
}
}
},
'GPU: MKLDNN_CUDNNOFF': {
node('mxnetlinux-cpu') {
ws('workspace/build-mkldnn-gpu-nocudnn') {
timeout(time: max_time, unit: 'MINUTES') {
init_git()
docker_run('ubuntu_build_cuda', 'build_ubuntu_gpu_mkldnn_cuda', false)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

supposed to be build_ubuntu_gpu_mkldnn_nocudnn?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

good catch! will fix.

pack_lib('mkldnn_gpu_nocudnn', mx_mkldnn_lib)
}
}
}
},
'GPU: CUDA9.1+cuDNN7': {
node('mxnetlinux-cpu') {
ws('workspace/build-gpu') {
Expand Down Expand Up @@ -645,6 +656,34 @@ try {
}
}
},
'Python2: MKLDNN-GPU-NOCUDNN': {
node('mxnetlinux-gpu') {
ws('workspace/ut-python2-mkldnn-gpu-nocudnn') {
try {
init_git()
unpack_lib('mkldnn_gpu_nocudnn', mx_mkldnn_lib)
python2_gpu_ut('ubuntu_gpu')
publish_test_coverage()
} finally {
collect_test_results_unix('nosetests_gpu.xml', 'nosetests_python2_mkldnn_gpu_nocudnn.xml')
}
}
}
},
'Python3: MKLDNN-GPU-NOCUDNN': {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it needed to test python2 and python3 or is one enough?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

one is enough , i will remove python2.

node('mxnetlinux-gpu') {
ws('workspace/ut-python3-mkldnn-gpu-nocudnn') {
try {
init_git()
unpack_lib('mkldnn_gpu_nocudnn', mx_mkldnn_lib)
python3_gpu_ut('ubuntu_gpu')
publish_test_coverage()
} finally {
collect_test_results_unix('nosetests_gpu.xml', 'nosetests_python3_mkldnn_gpu_nocudnn.xml')
}
}
}
},
'Python3: CentOS 7 CPU': {
node('mxnetlinux-cpu') {
ws('workspace/build-centos7-cpu') {
Expand Down
18 changes: 18 additions & 0 deletions ci/docker/runtime_functions.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -457,6 +457,24 @@ build_ubuntu_gpu_mkldnn() {
report_ccache_usage
}

build_ubuntu_gpu_mkldnn_nocudnn() {
set -ex

build_ccache_wrappers

make \
DEV=1 \
USE_CPP_PACKAGE=1 \
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No need for the cpp package here since we don't run the cpp tests

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

not sure about this . shouldnt we test if it builds fine with CUDNN with CPP_PACKAGE just in case cpp package depends on the flag to use cudnn specific functionality (unlikely) ?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The CPP package just generates wrappers around the C-API and should not rely on backend-specific things like CUDA or CUDNN. Thus, I'd say that we should keep it separate and only build the CPP package if we're actually going to run it (which we do in another test)

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

okay, removed!

USE_BLAS=openblas \
USE_MKLDNN=1 \
USE_CUDA=1 \
USE_CUDA_PATH=/usr/local/cuda \
USE_CUDNN=0 \
-j$(nproc)

report_ccache_usage
}

build_ubuntu_gpu_cuda91_cudnn7() {
set -ex
# unfortunately this build has problems in 3rdparty dependencies with ccache and make
Expand Down
12 changes: 6 additions & 6 deletions src/operator/nn/convolution.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,9 +90,9 @@ void ConvolutionCompute<gpu>(const nnvm::NodeAttrs& attrs,
int dtype = inputs[conv::kData].type_flag_;

#if CUDNN_MAJOR < 5
if (param_.layout.value() != kNCW &&
param_.layout.value() != kNCHW &&
param_.layout.value() != kNCDHW) {
if (param.layout.value() != kNCW &&
param.layout.value() != kNCHW &&
param.layout.value() != kNCDHW) {
// Need CuDNN > 5.0 for layout support. use MXNet implementation
MSHADOW_REAL_TYPE_SWITCH(dtype, DType, {
ConvolutionOp<gpu, DType> op;
Expand Down Expand Up @@ -168,9 +168,9 @@ void ConvolutionGradCompute<gpu>(const nnvm::NodeAttrs& attrs,
int dtype = out_grad.type_flag_;

#if CUDNN_MAJOR < 5
if (param_.layout.value() != kNCW &&
param_.layout.value() != kNCHW &&
param_.layout.value() != kNCDHW) {
if (param.layout.value() != kNCW &&
param.layout.value() != kNCHW &&
param.layout.value() != kNCDHW) {
// Need CuDNN > 5.0 for layout support. use MXNet implementation
MSHADOW_REAL_TYPE_SWITCH(dtype, DType, {
ConvolutionOp<gpu, DType> op;
Expand Down