Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1038. (Read 2347601 times)
does anyone else have problems in a multi-gpu rig with gpu 0 consistently crashing when mining neoscrypt algo? it's only gpu 0, the rest are fine and hash away reliably.
Does anyone have a whirlpoolx working config without crashing? I am using 1.5.55 with (9) 750ti's, (3) 970's, and (1) 960. I am having a hard time getting the miner to run without driver crashes. Quark runs like a champ.
Seem to work here (under Windows), but then I'm only using one card, not 13. 
Does anyone have a whirlpoolx working config without crashing? I am using 1.5.55 with (9) 750ti's, (3) 970's, and (1) 960. I am having a hard time getting the miner to run without driver crashes. Quark runs like a champ.
Thanks - pokeytex
Thanks - pokeytex
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!Compute 5.2 has bether launchconfigurations.
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
If you look in the code you will see that DJM34 made 2 kernals. One for compute 5.0 devices and one for compute 3.5 and 5.2. I have modded both of them and gained around 7% (reduced the codesize and some unroll tweaks)
The deb doesn't include the drivers but it depends on them, and a specific version. That was the problem.
Anyone know if there is a new miner or private miner out right now? All the algos are taking a crap right now.
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
awesome ...
where do you get the 5.2 patch for cuda 6.5? ...
i still have to manually edit the Makefile.am to stop it trying to compile compute 5.2 ...
great work though
...#crysx
pallas:
I prefer using the run file because it gives the option not to install the drivers. With the dist package
you get the included drivers whether you want them or not, and the're usually old.
crysx:
The 5.2 version of cuda is a full package, not a patch, that adds support for compute 5.2. Whether you
choose to use 5.2 it is still determined by Makefile.am. There's no harm in compiling for 5.2 when you
don't need it as long as you also include the compute versions you do need. It just takes longer to compile
and the executable is larger.
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
Will you be selling your private version at any point in the future?
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
awesome ...
where do you get the 5.2 patch for cuda 6.5? ...
i still have to manually edit the Makefile.am to stop it trying to compile compute 5.2 ...
great work though
...#crysx
- tried the deb file and failed
- downloaded the run file for cuda 6.5 9xx cards from nvidia website
- had to manually install the latest driver using the run file, downloaded separately (the one provided in the toolkit isn't compatible with the latest kernels)
- had to remove everything nvidia related on the system and other cleanups
- had to install gcc-4.7 and modify the Makefile.am in order to use it
and probably some other things I forgot....
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
awesome ...
where do you get the 5.2 patch for cuda 6.5? ...
i still have to manually edit the Makefile.am to stop it trying to compile compute 5.2 ...
great work though
...#crysx
- tried the deb file and failed
- downloaded the run file for cuda 6.5 9xx cards from nvidia website
- had to manually install the latest driver using the run file, downloaded separately (the one provided in the toolkit isn't compatible with the latest kernels)
- had to remove everything nvidia related on the system and other cleanups
- had to install gcc-4.7 and modify the Makefile.am in order to use it
and probably some other things I forgot....
yup ... welcome to linux ...
but once you get it running ... you wont have to do much else to it again ...
until the next version of course
...ill stick to manually editing sp's Makefile.am ...
too much work for me at the moment ...
tanx ...
#crysx
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
awesome ...
where do you get the 5.2 patch for cuda 6.5? ...
i still have to manually edit the Makefile.am to stop it trying to compile compute 5.2 ...
great work though
...#crysx
- tried the deb file and failed
- downloaded the run file for cuda 6.5 9xx cards from nvidia website
- had to manually install the latest driver using the run file, downloaded separately (the one provided in the toolkit isn't compatible with the latest kernels)
- had to remove everything nvidia related on the system and other cleanups
- had to install gcc-4.7 and modify the Makefile.am in order to use it
and probably some other things I forgot....
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
awesome ...
where do you get the 5.2 patch for cuda 6.5? ...
i still have to manually edit the Makefile.am to stop it trying to compile compute 5.2 ...
great work though
...#crysx
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
OMG I installed cuda 6.5 with support for compute 5.2, using the run files, what a pain!
but now lyra works and I get about 700 Kh/s more on a single 970 :-)
Is this for lyra only? The other algos appear to be working fine. Currently hashing quark at 15 Mh/s stock clocks.
It should be fixed, because I think the compute5.0 builds will be bether in the summertime. (less power, less heat and lower hashrates)
In the code you have different launch configurations for different compute versions like here in skein:
__global__
#if __CUDA_ARCH__ > 500
__launch_bounds__(480, 3)
#else
__launch_bounds__(240, 6)
#endif
void quark_skein512_gpu_hash_64(uint32_t threads, uint32_t startNounce, uint64_t * const __restrict__ g_hash, const uint32_t *
On the 970/980/980ti I run 480 gputhreads per block, but on 750ti only 240 threads per block. More threads use more power and generates more heat, but sometimes also more hash.
etc.
__global__
#if __CUDA_ARCH__ > 500
__launch_bounds__(480, 3)
#else
__launch_bounds__(240, 6)
#endif
void quark_skein512_gpu_hash_64(uint32_t threads, uint32_t startNounce, uint64_t * const __restrict__ g_hash, const uint32_t *
On the 970/980/980ti I run 480 gputhreads per block, but on 750ti only 240 threads per block. More threads use more power and generates more heat, but sometimes also more hash.
etc.
I get 16.2 MHASH@quark with compute 5.2 build for x86 in the opensource version. (gigabyte windforce 970)
Compute 5.2 has bether launchconfigurations.
Compute 5.2 has bether launchconfigurations.
Ubuntu 15.04 with distribution cuda toolkit (6.5 and compute 5.0)
You need to build for compute 5.2. 5.0 builds won't work on gtx 970 etc.
Is this for lyra only? The other algos appear to be working fine. Currently hashing quark at 15 Mh/s stock clocks.
aaand - there it is ...
http://www.abc.net.au/news/2015-07-15/cern-scientists-discover-new-kind-of-particle-the-pentaquark/6620308 ...
a new algo maybe? ...
...
#crysx
http://www.abc.net.au/news/2015-07-15/cern-scientists-discover-new-kind-of-particle-the-pentaquark/6620308 ...
a new algo maybe? ...
...#crysx
Ubuntu 15.04 with distribution cuda toolkit (6.5 and compute 5.0)
You need to build for compute 5.2. 5.0 builds won't work on gtx 970 etc.
Hmmm Quark hit 70GH...
Jump to: