ccminerKlausT yescryptR32 fails on 1080

Reporting here since it's your custom release of it.

Fails on a 1080 founders edition card because it sets the intensity too high. Happens on all of my systems using GTX1080s.


C:\test> Bin\NVIDIA-CcminerKlaustr11\ccminer.exe -b 4068 -N 1 -R 1 -a yescryptR32 -o stratum+tcp://yescryptR32.mine.zpool.ca:6343 -u censored -p ID=GRANTNEW,c=BTC -d 0
ccminer 8.21-KlausT-r11-fix (64bit) for nVidia GPUs
Compiled with Visual Studio 2015 using Nvidia CUDA Toolkit 9.2

Based on pooler cpuminer 2.3.2 and the tpruvot@github fork
CUDA support by Christian Buchner, Christian H. and DJM34
Compiled by nemosminer@github
Includes optimizations implemented by sp-hash, klaust, tpruvot and tsiv.

[2018-10-23 19:38:28] Starting Stratum on stratum+tcp://yescryptR32.mine.zpool.ca:6343
[2018-10-23 19:38:28] NVML GPU monitoring enabled.
[2018-10-23 19:38:28] NVAPI GPU monitoring enabled.
[2018-10-23 19:38:28] 1 miner thread started, using 'yescryptr32' algorithm.
[2018-10-23 19:38:28] Stratum difficulty set to 1
[2018-10-23 19:38:28] Received new yescryptr32 block header
[2018-10-23 19:38:28] block height 98982, 1 transactions
[2018-10-23 19:38:28] Using intensity 12.875 (7680 threads)
GPU #0: Cuda error in func 'scanhash_yescrypt_base' at line 148 : out of memory.
[2018-10-23 19:38:29] stopping 1 threads
[2018-10-23 19:38:29] API bind to port 4068 failed - trying again in 20sec
[2018-10-23 19:38:32] Received new yescryptr32 block header
[2018-10-23 19:38:32] block height 98982, 1 transactions
[2018-10-23 19:38:32] CTRL_C_EVENT received, exiting
[2018-10-23 19:38:34] resetting GPUs

Looking at that out of memory error people say to increase system memory and page file, but these systems have 32GB RAM and 40GB page file, and only a single 1080, so that doesn't seem to be the issue.

Manually specifying an intensity of '-i 12.406' (the value it uses on my 1070) makes it work just fine.

nemosminer / NemosMiner

ccminerKlausT yescryptR32 fails on 1080 #845