CCMINER tpruvot cuda8 branch on nvda 367.57 Cuda error in func 'neoscrypt_hash_k

[aurigae]

CCMINER tpruvot cuda8 branch on nvda 367.57 Cuda error in func 'neoscrypt_hash_k on xubuntu 16.10
The error occured after upgrading from the previous nvda prop driver.

Rebuild error:

1. Default installation of Xubuntu 16.10
2. install nvidia prop driver 367.57
3. Clone and compile CCminer tpruvot cuda 8 branch with gcc 4.8 (http://askubuntu.com/questions/724872/downgrade-gcc-from-5-2-1-to-4-9-ubuntu-15-10) and the Cflags / symlinks for the prop driver https://github.com/tpruvot/ccminer/blob/linux/INSTALL.
4. run ./build.sh from the git repository directory
5. Run ./ccminer -a neoscrypt

Code:

[2016-11-07 01:52:42] Starting on stratum+tcp://MININGPOOLADDRESS:PORT
[2016-11-07 01:52:42] NVML GPU monitoring enabled.
[2016-11-07 01:52:42] 1 miner thread started, using 'neoscrypt' algorithm.
[2016-11-07 01:52:42] Stratum difficulty set to 256
[2016-11-07 01:52:43] GPU #0: Using x cuda threads
Cuda error in func 'neoscrypt_hash_k4' at line 1517 : misaligned address.

Any idea how to fix that without downgrading to the previous nvda driver?

[1715535994]

1715535994

[1715535994]

1715535994

[aurigae]

This is wired, error remains after downgrading back to 367.48 where it previously did work.

[aurigae]

Wrong algo mined - if you receive this error check the -a option or port ....

explain: CUDA_SAFE_CALL(cudaMemcpy(resNonces, d_NNonce[thr_id], 2 * sizeof(uint32_t), cudaMemcpyDeviceToHost));
reference was correct so i had to assume i made a mistake....

CLOSED