Tensorflow 1.15.0 "The minimum required Cuda capability is 6.0."

01_error.md

The Google official Java build in Maven Central is broken, and has been for quite a while.

Environment: AWS P2.xlarge instance with a K80 GPU and CUDA compute capability of 3.7.

Error: "Ignoring visible gpu device [...] The minimum required Cuda capability is 6.0."

Expected: The minimum CUDA compute capability is 3.5, like the documents state, and as the Python packages require.

https://www.tensorflow.org/install/gpu

Hardware requirements

The following GPU-enabled devices are supported:

• NVIDIA® GPU card with CUDA® Compute Capability 3.5 or higher.

2019-10-26 09:54:23.584569: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1700] Ignoring visible gpu device (device: 0, name: Tesla K80, pci bus id: 0000:00:1e.0, compute capability: 3.7) with Cuda compute capability 3.7. The minimum required Cuda capability is 6.0.

02_full_output.txt

  .   ____          _            __ _ _
 /\\ / ___'_ __ _ _(_)_ __  __ _ \ \ \ \
( ( )\___ | '_ | '_| | '_ \/ _` | \ \ \ \
 \\/  ___)| |_)| | | | | || (_| |  ) ) ) )
  '  |____| .__|_| |_|_| |_\__, | / / / /
 =========|_|==============|___/=/_/_/_/
 :: Spring Boot ::        (v2.1.9.RELEASE)

2019-10-26 09:54:08.482  INFO 2790 --- [           main] io.mikael.poc.Main                       : Starting Main v1.0.0 on ip-172-31-10-103 with PID 2790 (/home/ubuntu/kotlin-tensorflow-segmentation-poc-1.0.0-gpu.jar started by ubuntu in /home/ubuntu)
2019-10-26 09:54:08.486 DEBUG 2790 --- [           main] io.mikael.poc.Main                       : Running with Spring Boot v2.1.9.RELEASE, Spring v5.1.10.RELEASE
2019-10-26 09:54:08.487  INFO 2790 --- [           main] io.mikael.poc.Main                       : No active profile set, falling back to default profiles: default
2019-10-26 09:54:12.349  INFO 2790 --- [           main] i.m.poc.services.SegmentationService     : TensorFlow: 1.15.0
2019-10-26 09:54:12.505 DEBUG 2790 --- [           main] i.m.poc.services.SegmentationService     : GPU enabled
2019-10-26 09:54:12.606511: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-10-26 09:54:12.632309: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300060000 Hz
2019-10-26 09:54:12.632792: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7fbdf21e7390 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2019-10-26 09:54:12.632842: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2019-10-26 09:54:12.634129: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
2019-10-26 09:54:13.661751: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:983] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-10-26 09:54:13.662550: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties: 
name: Tesla K80 major: 3 minor: 7 memoryClockRate(GHz): 0.8235
pciBusID: 0000:00:1e.0
2019-10-26 09:54:13.893752: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
2019-10-26 09:54:15.519264: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
2019-10-26 09:54:16.328755: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
2019-10-26 09:54:16.823456: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
2019-10-26 09:54:18.838846: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
2019-10-26 09:54:19.971950: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
2019-10-26 09:54:23.582661: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
2019-10-26 09:54:23.582863: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:983] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-10-26 09:54:23.583764: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:983] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2019-10-26 09:54:23.584569: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1700] Ignoring visible gpu device (device: 0, name: Tesla K80, pci bus id: 0000:00:1e.0, compute capability: 3.7) with Cuda compute capability 3.7. The minimum required Cuda capability is 6.0.
2019-10-26 09:54:24.063774: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-10-26 09:54:24.063829: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165]      0 
2019-10-26 09:54:24.063850: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0:   N 
2019-10-26 09:54:26.348  INFO 2790 --- [           main] io.mikael.poc.Main                       : Started Main in 19.205 seconds (JVM running for 23.374)