Simple tests for packages I maintain on Arch Linux
C++
Some very simple tests for Arch Linux packages
- DISCLAIMER: This is in no way or form official and should not be viewed as representing Arch Linux
Currently supported
pytorch
C++ API
Python API
tensorflow
C++ API
Python API
Keras API
cuDNN
C++ API
Running the tests
make
Example Output
$ make # on CPU
make -C pytorch/ATan run
make[1]: Entering directory '/home/gizdov/Tests/arch-package-tests/pytorch/ATan'
./atan
CUDA device is not available.
1 0
1 1
[ CPUIntType{2,2} ]
make[1]: Leaving directory '/home/gizdov/Tests/arch-package-tests/pytorch/ATan'
make -C pytorch/autograd run
make[1]: Entering directory '/home/gizdov/Tests/arch-package-tests/pytorch/autograd'
./autograd
CUDA device is not available.
We are using CPU
1.0782 1.3101
1.5989 1.8083
[ Variable[CPUFloatType]{2,2} ]
make[1]: Leaving directory '/home/gizdov/Tests/arch-package-tests/pytorch/autograd'
make -C tensorflow/basic run
make[1]: Entering directory '/home/gizdov/Tests/arch-package-tests/tensorflow/basic'
./basic
2019-04-16 15:20:08.215189: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2593000000 Hz
2019-04-16 15:20:08.215431: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x55fac13fa840 executing computations on platform Host. Devices:
2019-04-16 15:20:08.215451: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): <undefined>, <undefined>
2019-04-16 15:20:08.217658: I basic.cpp:22] 19
-3
make[1]: Leaving directory '/home/gizdov/Tests/arch-package-tests/tensorflow/basic'######################################################## COMPLETED ########################################################
$ make # on GPU
make -C pytorch/ATan run
make[1]: Entering directory '/home/gizdov/Git/arch-package-tests/pytorch/ATan'
g++ atan.cpp -I/usr/include/torch/csrc/api/include -I/usr/include/python3.7m -L/usr/lib/pytorch -L/opt/cuda/lib -lc10 -ltorch -lcaffe2 -lnvrtc -lcuda -o atan
./atan
CUDA device is available.
0 1
-1 1
[ CUDAIntType{2,2} ]
make[1]: Leaving directory '/home/gizdov/Git/arch-package-tests/pytorch/ATan'
make -C pytorch/autograd run
make[1]: Entering directory '/home/gizdov/Git/arch-package-tests/pytorch/autograd'
g++ autograd.cpp -I/usr/include/torch/csrc/api/include -I/usr/include/python3.7m -L/usr/lib/pytorch -L/opt/cuda/lib -lc10 -ltorch -lcaffe2 -lnvrtc -lcuda -o autograd
./autograd
CUDA device is available.
We are using CUDA
1.5140 0.8996
1.2108 0.9279
[ Variable[CUDAFloatType]{2,2} ]
make[1]: Leaving directory '/home/gizdov/Git/arch-package-tests/pytorch/autograd'
make -C tensorflow/basic run
make[1]: Entering directory '/home/gizdov/Git/arch-package-tests/tensorflow/basic'
g++ basic.cpp -I/usr/lib/python3.7/site-packages/tensorflow/include -ltensorflow_cc -ltensorflow_framework -o basic
./basic
2019-04-16 13:21:42.975410: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3401190000 Hz
2019-04-16 13:21:42.976343: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x559ffc6dd700 executing computations on platform Host. Devices:
2019-04-16 13:21:42.976364: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): <undefined>, <undefined>
2019-04-16 13:21:43.026726: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:998] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-04-16 13:21:43.027210: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x559ffc9f4160 executing computations on platform CUDA. Devices:
2019-04-16 13:21:43.027225: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): GeForce GTX 980, Compute Capability 5.2
2019-04-16 13:21:43.027947: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 0 with properties:
name: GeForce GTX 980 major: 5 minor: 2 memoryClockRate(GHz): 1.329
pciBusID: 0000:01:00.0
totalMemory: 3.95GiB freeMemory: 3.69GiB
2019-04-16 13:21:43.027961: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1512] Adding visible gpu devices: 0
2019-04-16 13:21:44.977808: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-04-16 13:21:44.977843: I tensorflow/core/common_runtime/gpu/gpu_device.cc:990] 0
2019-04-16 13:21:44.977851: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 0: N
2019-04-16 13:21:44.978213: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 3406 MB memory) -> physical GPU (device: 0, name: GeForce GTX 980, pci bus id: 0000:01:00.0, compute capability: 5.2)
2019-04-16 13:21:45.040475: I basic.cpp:22] 19
-3
make[1]: Leaving directory '/home/gizdov/Git/arch-package-tests/tensorflow/basic'######################################################## COMPLETED ########################################################