quark scales well with gpu clock
1500 => 1510mhz gives 6030 => 6070 khs
Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1041. (Read 2347601 times)
42mh/s at stock?
i think not, he likes to flash oc'ed gigabyte bioses )))but if the clocks are at stock I don't think it would boost past factory clocks, unless I'm mistaken
if this was the case then I don't really see how it would utilize the higher TDP limit.
at the top sp says standard clocks , this would indicate stock/factory clocks
42mh/s at stock?
i think not, he likes to flash oc'ed gigabyte bioses )))
compiled your latest commit Sp looking to get that 42+ Mh/s from my 750 ti's
just testing with -i 22.9 I got a 70 kh/s increase total, cards overclocked average 1345 mhz 39+ Mh/s .. way off your 42 at stock clocks.
I tried testing with the -g2 3,4,5 but the miner would crash tested down to -i 20 still no dice.
I would love to know your command line for the above run and nvidia driver used.
42mh/s at stock?
DO WANT
just testing with -i 22.9 I got a 70 kh/s increase total, cards overclocked average 1345 mhz 39+ Mh/s .. way off your 42 at stock clocks.
I tried testing with the -g2 3,4,5 but the miner would crash tested down to -i 20 still no dice.
I would love to know your command line for the above run and nvidia driver used.
42mh/s at stock?
DO WANT
just few results for new compiled git from sp_ on my poor gtx750 1gb
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
nice hashes ...
i dont get that with the 2GB cards ...
but then - i dont overclock either ...
...#crysx
Bro, if you don't OC, you ain't taking much advantage of your system!
One of my rigs:
6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
That's a very good speed on a bunch of 750Ti 6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
One of my rigs:
6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
nice ...
we dont get anywhere near that ... if that is quark ...
unless you have modded the cards firmware as well ... which i think you gave instructions a while back ...
no oc and stock ( as in - pull out of box - plug in - and run compiled ccminer ) we get much less than that ...
which algo is that sp ( just to be sure ) ? ...
#crysx
If you compile the latest version or wait for release 55 you will get a small increase in the quark hash. My gigabyte 970 went from 15.8 Mhash to 16.1 Mhash on the standard clocks.
Any quick guide to compile latest version? 
Or when will release 55 come out?
If you compile the latest version or wait for release 55 you will get a small increase in the quark hash. My gigabyte 970 went from 15.8 Mhash to 16.1 Mhash on the standard clocks.
sp - for us non-programming users - is there a guide on how to compile a latest commit for windows? I tried researching a couple nights ago but got way in over my head. Please advise.
thanks - pokeytex
nice speeds there you have noted the PhysX driver number from the 2-jun-2014 ?
There is a GEFORCE 340.43 DRIVER (BETA) driver released on the 17th of june 2014
but this contains Physx driver number version 9.13.1220
not sure what geforce driver you are using here, would like to test though.
which driver you using again sp?
also any -g used for this test?
you have noted the PhysX driver number from the 2-jun-2014 ? There is a GEFORCE 340.43 DRIVER (BETA) driver released on the 17th of june 2014
but this contains Physx driver number version 9.13.1220
not sure what geforce driver you are using here, would like to test though.
which driver you using again sp?
also any -g used for this test?
One of my rigs:
6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
6x750ti's The driver version is from 2-jun-2014. standard clocks. release 54 +
Driver: 9.13.2
The results are verified at the pool.
just few results for new compiled git from sp_ on my poor gtx750 1gb
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
nice hashes ...
i dont get that with the 2GB cards ...
but then - i dont overclock either ...
...#crysx
just few results for new compiled git from sp_ on my poor gtx750 1gb
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
lyra 1500/1600 1130khs
quark 1500/1575 6030khs
Other algos without improvements comparing to my old records.
win7 64bit, ccminer 32bit, cuda 6.5, driver 353.30
quark on nicehash gives 7-8% rejects with ccminer and 1-2% on sgminer.
If you compile the latest version or wait for release 55 you will get a small increase in the quark hash. My gigabyte 970 went from 15.8 Mhash to 16.1 Mhash on the standard clocks.
Currently I'm using ccMiner 1.5.54-git SP-MOD on 2x GTX980 and I achieved around 18.5MH/s per card running quark algo.
All default setting except +100MHz on GPU Core clock - with default power limit since I want to keep both temperature and wattage low.
Is the hashrate I'm getting good?
All default setting except +100MHz on GPU Core clock - with default power limit since I want to keep both temperature and wattage low.
Is the hashrate I'm getting good?
what are you guys speeds with the popular algos with cuda 7.5 under linux?
sp52 SDK 6.5 x11 GTX960 4972
sp54 SDK 7.5 x11 GTX960 4846
tested with :
ccminer-sp52 -a x11 -o stratum+tcp://drk.coinmine.pl:6090 -u tpruvot.x11 -p x -b 0.0.0.0
(it seems to submit shares)
sp54 SDK 7.5 x11 GTX960 4846
tested with :
ccminer-sp52 -a x11 -o stratum+tcp://drk.coinmine.pl:6090 -u tpruvot.x11 -p x -b 0.0.0.0
(it seems to submit shares)
BITCOIN--
Bitcoin is up over $285 on several exchanges. WooHoo!
I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!
--scryptr
Bitcoin is up over $285 on several exchanges. WooHoo!
I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!
--scryptr
reminder: you can safely install cuda 7.5 on linux if you answer no to the symlink, the only thing to change is in configure.sh (to build)
--with-cuda=/usr/local/cuda-7.5
and to execute, add the line /usr/local/cuda-7.5/lib64 in a file named /etc/ld.so.conf.d/cuda.conf and type ldconfig
UBUNTU--
If I do a fresh install of UBUNTU, and while using only the default drivers, I download and install CUDA TK 7.5 as a *.DEB file, I should get the pre-packaged drivers that install with the ToolKit. The *.DEB file should install pathways, also. Am I right?
After wading thru too many tutorials, and reading the manual itself, I think this is true. --scryptr
@sp_
I changed your latest from git as djm34 suggested, removing just enough #pragma unrolls to allow compile to succeed. I then compiled fresh with both cuda 6.5 and 7.5. This is the Quark performance I get on Windows / 980:
git-6.5: 19.2MH/s
git-7.5: 18.9MH/s
All in all, looks like a minor drop for me. I'm not convinced that these figures are entirely accurate, as it would take me much longer to do a reliable benchmark, time which I can't afford at the moment. Other algos might be broken, I only briefly checked Quark.
Ha! I missed that one
Linux ftw then
I changed your latest from git as djm34 suggested, removing just enough #pragma unrolls to allow compile to succeed. I then compiled fresh with both cuda 6.5 and 7.5. This is the Quark performance I get on Windows / 980:
git-6.5: 19.2MH/s
git-7.5: 18.9MH/s
All in all, looks like a minor drop for me. I'm not convinced that these figures are entirely accurate, as it would take me much longer to do a reliable benchmark, time which I can't afford at the moment. Other algos might be broken, I only briefly checked Quark.
linux doesnt have the heavycoin hefty problem
Ha! I missed that one
Linux ftw then 
BITCOIN--
Bitcoin is up over $285 on several exchanges. WooHoo!
I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!
--scryptr
Bitcoin is up over $285 on several exchanges. WooHoo!
I'm looking forward to moving to CUDA 7.5 sometime soon. As soon as Los Jefes sort it out, that is!
--scryptr
reminder: you can safely install cuda 7.5 on linux if you answer no to the symlink, the only thing to change is in configure.sh (to build)
--with-cuda=/usr/local/cuda-7.5
and to execute, add the line /usr/local/cuda-7.5/lib64 in a file named /etc/ld.so.conf.d/cuda.conf and type ldconfig
UBUNTU--
If I do a fresh install of UBUNTU, and while using only the default drivers, I download and install CUDA TK 7.5 as a *.DEB file, I should get the pre-packaged drivers that install with the ToolKit. The *.DEB file should install pathways, also. Am I right?
After wading thru too many tutorials, and reading the manual itself, I think this is true. --scryptr
Jump to: