Caffe Windows: pooling_layer.cu:212 Check failed: error==cudaSuccess(8vs.0) invalid device function

[luis.lop...@gmail.com]

Dear all,

I am getting this problem when I run Caffe prebuild binaries "Visual Studio 2015,CUDA 8.0, Python 2.7" from Windows branch. I have a Nvidia GTX 1060 GPU updated to the latest driver available, cuda 8.0 and cuDNN 5.1 installed. I read a lot of information about this issue on Linux OS but nothing to solve it on Windows. Both cuda and caffe device queries runs ok. Here the outputs:

caffe device_query -gpu 0

I0420 10:20:35.530796  1116 caffe.cpp:138] Querying GPUs 0

I0420 10:20:37.060575  1116 common.cpp:187] Device id:                     0

I0420 10:20:37.061079  1116 common.cpp:188] Major revision number:         6

I0420 10:20:37.061578  1116 common.cpp:189] Minor revision number:         1

I0420 10:20:37.061578  1116 common.cpp:190] Name:                          GeForce GTX 1060

I0420 10:20:37.061578  1116 common.cpp:191] Total global memory:           3221225472

I0420 10:20:37.061578  1116 common.cpp:192] Total shared memory per block: 49152

I0420 10:20:37.062592  1116 common.cpp:193] Total registers per block:     65536

I0420 10:20:37.062592  1116 common.cpp:194] Warp size:                     32

I0420 10:20:37.063583  1116 common.cpp:195] Maximum memory pitch:          2147483647

I0420 10:20:37.063583  1116 common.cpp:196] Maximum threads per block:     1024

I0420 10:20:37.064587  1116 common.cpp:197] Maximum dimension of block:    1024, 1024, 64

I0420 10:20:37.064587  1116 common.cpp:200] Maximum dimension of grid:     2147483647, 65535, 65535

I0420 10:20:37.065589  1116 common.cpp:203] Clock rate:                    1670500

I0420 10:20:37.065589  1116 common.cpp:204] Total constant memory:         65536

I0420 10:20:37.066090  1116 common.cpp:205] Texture alignment:             512

I0420 10:20:37.066090  1116 common.cpp:206] Concurrent copy and execution: Yes

I0420 10:20:37.066090  1116 common.cpp:208] Number of multiprocessors:     10

I0420 10:20:37.066591  1116 common.cpp:209] Kernel execution timeout:      Yes

nvidia devicequery:

Device 0: "GeForce GTX 1060"

  CUDA Driver Version / Runtime Version          8.0 / 8.0

  CUDA Capability Major/Minor version number:    6.1

  Total amount of global memory:                 3072 MBytes (3221225472 bytes)

  (10) Multiprocessors, (128) CUDA Cores/MP:     1280 CUDA Cores

  GPU Max Clock rate:                            1671 MHz (1.67 GHz)

  Memory Clock rate:                             4004 Mhz

  Memory Bus Width:                              192-bit

  L2 Cache Size:                                 1572864 bytes

  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)

  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers

  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers

  Total amount of constant memory:               65536 bytes

  Total amount of shared memory per block:       49152 bytes

  Total number of registers available per block: 65536

  Warp size:                                     32

  Maximum number of threads per multiprocessor:  2048

  Maximum number of threads per block:           1024

  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)

  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)

  Maximum memory pitch:                          2147483647 bytes

  Texture alignment:                             512 bytes

  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)

  Run time limit on kernels:                     Yes

  Integrated GPU sharing Host Memory:            No

  Support host page-locked memory mapping:       Yes

  Alignment requirement for Surfaces:            Yes

  Device has ECC support:                        Disabled

  CUDA Device Driver Mode (TCC or WDDM):         WDDM (Windows Display Driver Model)

  Device supports Unified Addressing (UVA):      Yes

  Device PCI Domain ID / Bus ID / location ID:   0 / 1 / 0

  Compute Mode:

     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 8.0, CUDA Runtime Version = 8.0, NumDevs = 1, Device0 = GeForce GTX 1060

Result = PASS

Can some one help me with this issue? Thanks for your support

BR

[Paul Delamusica]

Saw that on windows. Fixed by re-building. Mine is Visual Studio 2013,CUDA 7.5, Python 2.7.