CUDA Errors with ccminer-sp

[jsj101]

I have two GPUs in my PC; a GTX750 and a GTX660i. The 660ti works fine however I am encountering CUDA errors when trying to mine with the 750 - which should be getting better hash rates than the 660ti but to be honest they are not miles apart.

I encounter problems using the x11, x13 and x15 algorithms in particular. The only way to stop these errors is to set the --intensity below 20 but this results in a very poor hash rate.

The card has 1GB memory and is running at a 1.3ghz clock speed. Drivers are up to date on both cards.

Example error when no intensity is specified. I have downloaded the binary not compiled myself and running on 64bit Win10.

Code:

NiceHashMiner_v1.4.0.2\bin>ccminer_sp.exe -q -a x13 -o stratum+tcp://stratum.nicehash.com:3336 -u 14vZ4DHDzhttzKyNAmLzpRq6VLmCGb16vX -p X -D -d 0
SP-Mod 1.5.79
Compiled with Visual C++ 18 using Nvidia CUDA Toolkit 7.5

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

[2016-05-21 19:15:28] tid(0x00002580) Starting Stratum on stratum+tcp://stratum.nicehash.com:3336
[2016-05-21 19:15:28] tid(0x00002584) CUDA GPU#0 matches NVAPI GPU 0 by busId 1
[2016-05-21 19:15:28] tid(0x00002584) CUDA GPU#1 matches NVAPI GPU 1 by busId 2
[2016-05-21 19:15:28] tid(0x00002584) NVAPI GPU monitoring enabled.
[2016-05-21 19:15:28] tid(0x00002584) 1 miner thread started, using 'x13' algorithm.
[2016-05-21 19:15:29] tid(0x00002580) Stratum difficulty set to 0.04
[2016-05-21 19:15:29] tid(0x00002580) DEBUG: job_id=740a5c1 000000019670e794 xnonce2=000000 time=19:15:29
[2016-05-21 19:15:29] tid(0x00000304) job 740a5c1 000000019670e794 target change: 18ffe70000 (10.2)
[2016-05-21 19:15:29] tid(0x00000304) GPU #0: start=00000000 end=002fffff range=002fffff
Cuda error in func 'x11_simd512_cpu_init' at line 807 : out of memory.

and another example 'invalid device symbol'

Code:

ccminer.exe -o stratum+tcp://stratum.nicehash.com:3336 -u 14vZ4DHDzhttzKyNAmLzpRq6VLmCGb16vX -p x --intensity 20 --diff 0.1
SP-Mod 1.5.79
Compiled with Visual C++ 18 using Nvidia CUDA Toolkit 7.5

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

[2016-05-21 19:18:24] Starting Stratum on stratum+tcp://stratum.nicehash.com:3336
[2016-05-21 19:18:24] NVAPI GPU monitoring enabled.
[2016-05-21 19:18:24] 2 miner threads started, using 'x11' algorithm.
[2016-05-21 19:18:24] Binding thread 0 to cpu 0 (mask 1)
[2016-05-21 19:18:24] Binding thread 1 to cpu 1 (mask 2)
[2016-05-21 19:18:24] ...retry after 5 seconds
[2016-05-21 19:18:29] Stratum difficulty set to 0.04
[2016-05-21 19:18:29] stratum.nicehash.com:3336 x11 block 473391
Cuda error in func 'quark_blake512_cpu_setBlock_80' at line 815 : invalid device symbol.

How can I fix this?

Many thanks

[bathrobehero]

Those two cards are vastly different (different architecture) and your 750 is just that and not a the popular 750 Ti which also has 2 GB of VRAM.

I'd use two instances of the miner for each card (one with -d 0 and the other with -d 1) with different intensities which you have to find yourself if you want maximum speed.

Some miner softwares are also not supporting older (Kepler) cards such as the 660 Ti.

[maxim000]

ccminer_sp work only with maxwell cards, so GTX660 not be worked any way

for non maxwell cards use ccminer from tpruvot

[jsj101]

Both those errors above are from the 750 card, please correct me if I have mis understood - I thought the 750 should be able to work with ccminer-sp without those cuda errors? Thanks