invalid device function with GTX 980

[GoogleCodeExporter]

When i try to run training example of cuda-convnet2 i get this error :

src/nvmatrix.cu(394) : getLastCudaError() CUDA error : kSetupCurand: Kernel 
execution failed : (8) invalid device function .

I have GTX 980 on my machine and it has compute capability 5.2

I tried to modify makefiles in cudaconv3 & cudaconvnet & nvmatrix like this and 
to add 52 instead of 50 tooand i stil have same error. 

GENCODE_SM35    := -gencode arch=compute_35,code=sm_35
GENCODE_FLAGS   := $(GENCODE_SM35)

to

GENCODE_SM35    := -gencode arch=compute_35,code=sm_35
GENCODE_SM50    := -gencode arch=compute_50,code=sm_50
GENCODE_FLAGS   := $(GENCODE_SM50)

Original issue reported on code.google.com by [email protected] on 11 Apr 2015 at 9:21

Attachments:

• log.txt.txt

[GoogleCodeExporter]

I'm also using GTX980. It compiled ok using only the line 
GENCODE_SM50 := -gencode arch=compute_50, code=sm_50. 
Comment all others.
However I getting a error training for CIFAR example. If you succeed training, 
tell me.
The people in Imagenet ask 5 days to approve to download Imagenet examples. 
It's terrible. Alex you should use other example for Convnet2. 

Original comment by [email protected] on 21 Apr 2015 at 6:41

[GoogleCodeExporter]

I'm compiling just fine, but het this eror training on ImageNet data.

Original comment by [email protected] on 21 Apr 2015 at 7:31

[GoogleCodeExporter]

Yes, but this error results of compilation with wrong architecture. Did you try 
compute_50?

Original comment by [email protected] on 22 Apr 2015 at 12:02

[GoogleCodeExporter]

Yes i did, and same thing hapens everey time.....

Original comment by [email protected] on 22 Apr 2015 at 12:26

[GoogleCodeExporter]

I fixed it with this.
GENCODE_SM52 := -gencode arch=compute_52,code=sm_52

Original comment by [email protected] on 28 May 2015 at 9:06