CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1060.

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: chrysophylax on June 16, 2015, 05:32:01 AM

what parameters do you use for the commandline? ...

do you have spmod results? ...

tanx ...

#crysx

In case that was aimed at me I used -i 20 because higher figures like 25 used the same amount of memory and didn't increase the speed at all so I figured that was the max.

I don't think sp_ added zr5 into his fork.

chrysophylax

legendary

Activity: 2912

Merit: 1091

--- ChainWorks Industries ---

Quote from: ?? on ??

Quote from: bathrobehero on June 16, 2015, 04:23:11 AM

Quote from: ?? on ??

Either. Both, if you can.

There you go (Tpruvot's latest release: ccminer-rel1.6.4-vc2013):

750 Ti (GV-N75TOC-2GI):
stock 1.02 mh/s at 58.6W
oc - 1.14 mh/s at 65.4W (+140/0 - 1293-1328mhz)

970 (GV-N970WF3OC-4GD - 250w OC edition instead of 145w):
stock - 2.75 mh/s at 187W
oc - 3.0 mh/s at 208W (+185/0 - 1501mhz)

780 Ti (GV-N78TOC-3GD):
stock - 2.0 mh/s at 237W
oc - 2.26 mh/s at 275W (+120/0 - 1204mhz)

Measured at the wall and while the cards are on a 1300W 80+ gold PSU, it's barely utilized so its efficiency is probably pretty terrible. Used my go to OC settings but it could probably go higher.

Thanks - I assumed Nvidia was doing far better than AMD on this one, but I wanted to check; good thing I did.

what parameters do you use for the commandline? ...

do you have spmod results? ...

tanx ...

#crysx

rednoW

legendary

Activity: 1510

Merit: 1003

Quote from: djm34 on June 16, 2015, 04:35:06 AM

no it isn't a bug... what do you think the intensity is Roll Eyes

(and in most of the case it is already too high)

True! For my gtx750 1gb intensity 22.9 gives slightly better result then default one that is higher.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bensam1231 on June 16, 2015, 03:23:07 AM

Quote from: djm34 on June 15, 2015, 02:28:37 PM

Quote from: bensam1231 on June 15, 2015, 11:57:41 AM

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

you need more memory... or pagefile... (which is required to allocate that memory to the gpu...)
alternatively you can lower the intensity, that should free up some memory
other alternative, try to start 1 or 2 ccminer session

As mentioned running more then one instance does not fix this, system memory is fine. It doesn't get maxed out or anywhere close to it while monitoring it. I'll throw some more memory into the system and see what happens though. Currently it's at 4GB.

Haven't tried a lower intensity, this seems more like a bug then anything if that's the case.

no it isn't a bug... what do you think the intensity is Roll Eyes

(and in most of the case it is already too high)

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: ?? on ??

Either. Both, if you can.

There you go (Tpruvot's latest release: ccminer-rel1.6.4-vc2013):

750 Ti (GV-N75TOC-2GI):
stock 1.02 mh/s at 58.6W
oc - 1.14 mh/s at 65.4W (+140/0 - 1293-1328mhz)

970 (GV-N970WF3OC-4GD - 250w OC edition instead of 145w):
stock - 2.75 mh/s at 187W
oc - 3.0 mh/s at 208W (+185/0 - 1501mhz)

780 Ti (GV-N78TOC-3GD):
stock - 2.0 mh/s at 237W
oc - 2.26 mh/s at 275W (+120/0 - 1204mhz)

Measured at the wall and while the cards are on a 1300W 80+ gold PSU, it's barely utilized so its efficiency is probably pretty terrible. Used my go to OC settings but it could probably go higher.

rednoW

legendary

Activity: 1510

Merit: 1003

Quote from: sp_ on June 16, 2015, 03:30:43 AM

Tanguy Pruvot has optimized jh-512. I have added it into my fork now. Boost in quark and x11

hell ya! + 30-40 khs in quark!

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: bensam1231 on June 16, 2015, 03:23:07 AM

Quote from: djm34 on June 15, 2015, 02:28:37 PM

Quote from: bensam1231 on June 15, 2015, 11:57:41 AM

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

you need more memory... or pagefile... (which is required to allocate that memory to the gpu...)
alternatively you can lower the intensity, that should free up some memory
other alternative, try to start 1 or 2 ccminer session

As mentioned running more then one instance does not fix this, system memory is fine. It doesn't get maxed out or anywhere close to it while monitoring it. I'll throw some more memory into the system and see what happens though. Currently it's at 4GB.

Haven't tried a lower intensity, this seems more like a bug then anything if that's the case.

Memory is not necessary, pagefile will most likely work. I have 14GB (!) of pagefile on my rigs because I think there was a cudaminer algo that only ran if I had tons of pagefile but it never used more than a few megabytes.
Same goes for sgminer to a certain degree, although I was never able to run it with more than 4 cards enabled.

Quote from: ?? on ??

Can I get some ZR5 (Ziftr) hashrates from people with 750Ti/960/970/980/980Ti? Power use in W, too, if you've got it.

Stock or OC?

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Tanguy Pruvot has optimized jh-512. I have added it into my fork now. Boost in quark and x11

bensam1231

legendary

Activity: 1764

Merit: 1024

Quote from: djm34 on June 15, 2015, 02:28:37 PM

Quote from: bensam1231 on June 15, 2015, 11:57:41 AM

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

you need more memory... or pagefile... (which is required to allocate that memory to the gpu...)
alternatively you can lower the intensity, that should free up some memory
other alternative, try to start 1 or 2 ccminer session

As mentioned running more then one instance does not fix this, system memory is fine. It doesn't get maxed out or anywhere close to it while monitoring it. I'll throw some more memory into the system and see what happens though. Currently it's at 4GB.

Haven't tried a lower intensity, this seems more like a bug then anything if that's the case.

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: rednoW on June 15, 2015, 11:36:27 AM

2 sp_
Some testing your own built #52 vs. fresh github clone (my own build) vs one of the best "old" #47
gtx 750 best results
x11      2958 vs 2964 vs 2950
lyra2re    912 vs 909 vs 904
quark    6000 vs 5965 vs 5980
neoscrypt 190 vs 190 vs n/a
Please, return back quark and lyra performance for compute 5.0 ))

Try the last commit. Should give a small boost on the 750ti at least.

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

Quote from: Angora on June 16, 2015, 02:25:25 AM

did some testing with Lyra2 n Quark on 3 versions of modded ccminer
DJM34 VTC0.1 Lyra2: 477K-540K Quark 2147K-2482K
SP 1.5.1 Lyra2: 500K-580K Quark 2230K-2273K
sp release 52: Lyra2 only 1 accept for 590K Quark 2116K-2366K
testing time 5 minutes for each on the donate settings
Card is EVGA 750ti clock n memory speed 1268M PCIE 2.0 OS Windows 8.1 64bit
V52 was giving me lots of rejects from extranonce.. Now I recall reading in a sgminer thread that extranonce can cause the miner to show hash as failed but it most likely was accepted on the pool end. The solution was to use a command line argument to turn off extranonce. something like --noextranonce.

your quark numbers are way off. they should be around 6MHASH on the EVGA 750ti 300% faster than the numbers you get.

chrysophylax

legendary

Activity: 2912

Merit: 1091

--- ChainWorks Industries ---

Quote from: Angora on June 16, 2015, 02:25:25 AM

did some testing with Lyra2 n Quark on 3 versions of modded ccminer

DJM34 VTC0.1 Lyra2: 477K-540K Quark 2147K-2482K

SP 1.5.1 Lyra2: 500K-580K Quark 2230K-2273K

sp release 52: Lyra2 only 1 accept for 590K Quark 2116K-2366K

Testing time 5 minutes for each on the donate settings

Card is EVGA 750ti clock n memory speed 1268M PCIE 2.0 OS Windows 8.1 64bit

V52 was giving me lots of rejects from extranonce.. Now I recall reading in a sgminer thread that extranonce can cause the miner to show hash as failed but it most likely was accepted on the pool end. The solution was to use a command line argument to turn off extranonce. something like --noextranonce.

thats strange how in v52 spmod lyra2 was bombing out ...

we had the tests run about an hour ago - and lyra was fine - same with quark and neoscrypt - x11 and x13 ...

x15 was running fine - but the nicehash stratum dropped the link - as it usually does on the us stratum ...

what settings did you use for lyra2 on v52? ... i have only the intensity parameter active -i 16.5 ( -i 22.5 is max but many cpu validation errors ) ...

#crysx

Angora

full member

Activity: 241

Merit: 100

did some testing with Lyra2 n Quark on 3 versions of modded ccminer

DJM34 VTC0.1 Lyra2: 477K-540K Quark 2147K-2482K

SP 1.5.1 Lyra2: 500K-580K Quark 2230K-2273K

sp release 52: Lyra2 only 1 accept for 590K Quark 2116K-2366K

Testing time 5 minutes for each on the donate settings

Card is EVGA 750ti clock n memory speed 1268M PCIE 2.0 OS Windows 8.1 64bit

V52 was giving me lots of rejects from extranonce.. Now I recall reading in a sgminer thread that extranonce can cause the miner to show hash as failed but it most likely was accepted on the pool end. The solution was to use a command line argument to turn off extranonce. something like --noextranonce.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bensam1231 on June 15, 2015, 11:57:41 AM

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

you need more memory... or pagefile... (which is required to allocate that memory to the gpu...)
alternatively you can lower the intensity, that should free up some memory
other alternative, try to start 1 or 2 ccminer session

rednoW

legendary

Activity: 1510

Merit: 1003

Quote from: bensam1231 on June 15, 2015, 11:57:41 AM

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

try to use -i param with lower intensity, e.g. 14.3

bensam1231

legendary

Activity: 1764

Merit: 1024

Bugreport for Neoscrypt. Running 5 970s on the same system causes it to spit out "Cuda error in func 'scanhash-neoscrypt' at line 67 : out of memory."

Limiting it to any 4 devices with -d eliminates the problem. Tried running separate instances with different devices, doesn't work. System memory is fine. DJM's miner has the same problem.

rednoW

legendary

Activity: 1510

Merit: 1003

2 sp_

Some testing your own built #52 vs. fresh github clone (my own build) vs one of the best "old" #47
gtx 750 best results
x11      2958 vs 2964 vs 2950
lyra2re    912 vs 909 vs 904
quark    6000 vs 5965 vs 5980
neoscrypt 190 vs 190 vs n/a
Please, return back quark and lyra performance for compute 5.0 ))

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

If you build it yourself you need cuda 6.5. This fork has issues with cuda 7.

Angora

full member

Activity: 241

Merit: 100

Quote from: chrysophylax on June 14, 2015, 10:45:42 PM

Quote from: Angora on June 14, 2015, 09:54:34 PM

I just tried testing version 1.5.52 with the donation settings....
ONLY quark worked. All others were giving 99% reject. I tried the Lyra2 on 2 pools I use (IPO Miner & Verters) and there again 99% reject.

However, when I tried with version 06.15, Lyra2,quark & neoscrypt worked. None of the X series did. Got 99% reject.

Card is EGA GTX 750ti running @ stock

i see that - and the mining on quark continues ...

what we seems to find with nicehash is the extranonce2 size issue - and that issue is a regular thing ...

for the pool to accept shares - the miner needs to mine for a little bit in order for the nicehash stratums to do what they do ...

but again - this seems to be an issue with most of the algos at the moment ... this does not seem to be much of an issue with most other pools - with exception of yaamp on occasion ...

GKarB5 - is mining quite nicely of quark at the moment - but the initial 30seconds or so were rejected shares ...

https://www.westhash.com/index.jsp?p=miners&a=12&addr=1CTiNJyoUmbdMRACtteRWXhGqtSETYd6Vd ...

im not sure if nicehash force that to happen while the stratum adjusts or it takes longer with more shares with teh other algos ... same happens with yaamp - though not as much for some reason ...

all the donation algos are connected to nicehash on the US stratum - except x15 which is completely down on the US stratum altogether ...

today ill be testing the other algos over a small period of time with the test miner we have here ... 2 x gigabyte 750ti oc cards ( one card is the lp edition ) and see how they react ...

it does seem strange that the previous versions 'worked' on the other algos while the latest didnt for you ...

we will be testing with the latest git clone today ...

#crysx

The error I get from new version is "share above target".
I ran quark on suprnova for 6 hrs and error rate was just under 5%. Was getting some nice hash rates. About 1 minute out of gate, 1.2M on the pool.

Does it make a difference as to which version of Cuda is installed? I have v7 installed.

OS win8.1 64bit

sp_

legendary

Activity: 2954

Merit: 1087

Team Black developer

I rewrote cubehash and saved some registers. Look's like x11 is +5khash on the 750ti, so practically no gain... Sad

but no more insane tables like this one:

x[0][0][0][0][0] ^= in[0];

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1060. (Read 2347601 times)