Author

Topic: Cuda compilation failed calling "cudaFree" (Read 382 times)

newbie
Activity: 1
Merit: 0
September 08, 2017, 11:40:41 AM
#1
I compiled succesfully the CudaMiner (https://github.com/EntropyFactory/CudaMiner) in a Intel Haswell Machine with the NVIDIA Tesla K80. I'm compiling with Ubuntu 14.04, but i receive this error. How to fix it?


Error:
Code:
[2017-09-08 16:22:28] NVML GPU monitoring is not available.
[2017-09-08 16:22:28] 1 miner threads started, using 'keccakc' algorithm.
[2017-09-08 16:22:28] Starting Stratum
[2017-09-08 16:22:28] Stratum detected new block
[2017-09-08 16:22:28] GPU #0: cudaError 78 (a PTX JIT compilation failed) calling 'cudaFree(0)' (salsa_kernel.cu line 240)

[2017-09-08 16:22:28] GPU #0: Tesla K80 with compute capability 3.7
[2017-09-08 16:22:28] GPU #0: cudaError 78 (a PTX JIT compilation failed) calling 'cudaDeviceSetCacheConfig(kernel->cache_config())' (salsa_kernel.cu line 429)

[2017-09-08 16:22:29] GPU #0: cudaError 78 (a PTX JIT compilation failed) calling 'cudaDeviceSetSharedMemConfig(kernel->shared_mem_config())' (salsa_kernel.cu line 430)

[2017-09-08 16:22:29] GPU #0: interactive: 0, tex-cache: 0 , single-alloc: 0
[2017-09-08 16:22:29] GPU #0: 32 hashes / 4.0 MB per warp.
[2017-09-08 16:22:29] GPU #0: Performing auto-tuning (Patience...)
[2017-09-08 16:22:31] GPU #0: maximum total warps (BxW): 0
[2017-09-08 16:22:31] GPU #0:    0.00 hash/s with configuration T0x0
[2017-09-08 16:22:31] GPU #0: using launch configuration T0x0
*** stack smashing detected ***: ./cudaminer terminated


Here you can find the cuda-memcheck:

https://pastebin.com/yYuFd0Us

My system

gcc (Ubuntu 4.8.4-2ubuntu1~14.04.3) 4.8.4

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2013 NVIDIA Corporation
Built on Wed_Jul_17_18:36:13_PDT_2013
Cuda compilation tools, release 5.5, V5.5.0
I installed drivers using those commands

$ sudo apt-get install ocl-icd-opencl-dev
$ sudo apt-get install nvcc
Jump to: