Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 1217. (Read 2347597 times)
What does this mean, I'm not familiar.
Finally broke 8000 khash with GTX 970 with release 12
Unreal
Dontation coming your way
I recommend others do the same
Unreal
Dontation coming your way
I recommend others do the same
I already got some from you.
But more is welcome. Let's push the 970 to 10MHASH 
I'm on board for that!
Sent you 1.4 DRK, you deserve it
Congrats! You've made it: 10MH/s X11 is now possible.
snip
Cheers
snip
Cheers
ok but that oc is a bit heavy, i would expect about the same without that crazy oc
The new kernals can be overclocked more. +200 core and +400 mem used to be unstable, but now they are fine.
Finally broke 8000 khash with GTX 970 with release 12
Unreal
Dontation coming your way
I recommend others do the same
Unreal
Dontation coming your way
I recommend others do the same
I already got some from you.
But more is welcome. Let's push the 970 to 10MHASH 980GTX Stock X11 => 8,5MH/s
x11:
980 (evga SC acx2) =8800khash
...stock OC again; linux; running on p2 (lower mem speed)
It's OK, your card is overclocked by EVGA via BIOS/factory....mines are 980gtx reference ones.
To go above 1600MHz I think we need BIOS tunning or go for a Galax 980 HOF.
I just fixed x15 hash on github.(970/980)
Finally broke 8000 khash with GTX 970 with release 12
Unreal
Dontation coming your way
I recommend others do the same
Unreal
Dontation coming your way
I recommend others do the same
GPU Clockspeed : 1163 MHz (boost) / 1085 MHz (base)
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
i'm more interested at 980-970http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
i sold all my 750
980GTX Stock X11 => 8,5MH/s
x11:
980 (evga SC acx2) =8800khash
...stock OC again; linux; running on p2 (lower mem speed)
GPU Clockspeed : 1163 MHz (boost) / 1085 MHz (base)
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
i'm more interested at 980-970
i sold all my 750
980GTX Stock X11 => 8,5MH/s
subir fotos gratis
Fixed the makefile now. Also found out that the register configuration was wrong. In the windows version I use 80 regs as default, but the linux version had 128. Performance should be much bether now(linux)
thanks - works
x13 is the bigger suprise for me.
980 (evga SC acx2) =7280khash // 750ti=2300khash
...and i have not the crazy OC like others. 980 is stock OC.
x15 gives an error
Code:
[2014-11-29 19:47:40] GPU #0: GeForce GTX 980, 4589745 kH/s
[2014-11-29 19:47:40] GPU #3: GeForce GTX 980, 3636978 kH/s
Cuda error in func 'quark_blake512_cpu_setBlock_80' at line 309 : an illegal memory access was encountered.
[2014-11-29 19:47:40] GPU #3: GeForce GTX 980, 3636978 kH/s
Cuda error in func 'quark_blake512_cpu_setBlock_80' at line 309 : an illegal memory access was encountered.
GPU Clockspeed : 1163 MHz (boost) / 1085 MHz (base)
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
i'm more interested at 980-970
i sold all my 750
GPU Clockspeed : 1163 MHz (boost) / 1085 MHz (base)
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
http://www.gainward.com/main/vgapro.php?id=926&lang=en
2670-2710 MHASH.
Congrats! You've made it: 10MH/s X11 is now possible.
snip
Cheers
snip
Cheers
ok but that oc is a bit heavy, i would expect about the same without that crazy oc
reliece 12 gave me about 3100 kH/s on X11 with my PALIT 750ti
EDIT: Clocks are +200 core / +400 memory
EDIT: Clocks are +200 core / +400 memory
Release 12 seems to run more stable with higher overclock's. +200 core and +400mem has been running stable 24h here.
how many mega with stock settings? 1250 core default mem?
reliece 12 gave me about 3100 kH/s on X11 with my PALIT 750ti
EDIT: Clocks are +200 core / +400 memory
EDIT: Clocks are +200 core / +400 memory
Release 12 seems to run more stable with higher overclock's. +200 core and +400mem has been running stable 24h here.
Wow, slowed down mem clock and used your core clock without overting, nicely done:
The windforce has a 6pin connector. I think if you modify the bios, the cards will use 65Watt. Not sure if it's worth it though. (Hash/watt) With 65 watt, increased memclock/ higher intensity might boost the hash.
But if you want to make another 750ti world record, it might be worth it
Fixed the makefile now. Also found out that the register configuration was wrong. In the windows version I use 80 regs as default, but the linux version had 128. Performance should be much bether now(linux)
would like to give it a try, but I could not compile you newest source with linux
Code:
...
ptxas info : Function properties for _Z17fugue256_gpu_hashiijPvPj
0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info : Used 65 registers, 368 bytes cmem[0], 4 textures
make[2]: Leaving directory `/opt/miners/ccminer-sp'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/opt/miners/ccminer-sp'
make: *** [all] Error 2
Jump to: