Does this make any difference on nvidia cards? I noticed a 1% increase in speed but that could be something else.
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 1053. (Read 3426921 times)
December 20, 2013, 01:08:28 PM
December 20, 2013, 01:02:28 PM
Hi,
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs
) 
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs
) I haven't solved this yet, but I'm seeing it too. I worked around it by running one instance of cudaminer per device in the card instead of letting cudaminer handle both devices. Might work for you.
Are you using SLI?!
If you have SLI enabled mining will NOT work.
Disable SLI, and see if that helps.
If SLI is disabled try re-enabling it, restarting, then disabling it again.
December 20, 2013, 12:44:23 PM
Code for 18 Dec release to run on GTS 450 please.
Change -l 32x4 to -l auto, remove -C (it's ignored now).
December 20, 2013, 12:19:57 PM
Hi,
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs)
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs
) I haven't solved this yet, but I'm seeing it too. I worked around it by running one instance of cudaminer per device in the card instead of letting cudaminer handle both devices. Might work for you.
December 20, 2013, 11:51:52 AM
Wow! This new version rocks, you are the best! 
200 kh/s (instead of 125) @ GTX660
+ My computer doesn't lag
200 kh/s (instead of 125) @ GTX660
+ My computer doesn't lag
December 20, 2013, 10:24:50 AM
Hi,
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs)
BTW: As expected there is no speed up on GTX 470, 480, 570 or 580, but much more heat (and roaring fans) with the new release.
getting 255 kh/s from a GTX 660 TI and on a GTX 660 (without TI) 195 khs (improvement about 40%)
Impressive speed up
But I am unable to start two cards (no SLI) with the new release (2013-12-18 (beta)) neither on a rig with two GTX 650 Ti nor on a rig with two GTX 660.
(These teams are still working with an older version 2013-11-15 but at very slow rates compared to the present release)
I think I got every possible error. segmentation fault, floating point expection, result does not validate, unreasonable khs.
But everything is fine, when I state a single device with -d.
Any ideas whats going wrong ? (A loss of 300 khs
) BTW: As expected there is no speed up on GTX 470, 480, 570 or 580, but much more heat (and roaring fans) with the new release.
December 20, 2013, 09:54:48 AM
Code for 18 Dec release to run on GTS 450 please.
Used .bat for older release to run on
New release won't start with this code
Used .bat for older release to run on
Code:
cudaminer.exe -i 0 -C 2 -l 32x4 -m 1
New release won't start with this code
December 20, 2013, 09:44:10 AM
Alright, I managed to tweak my settings and keep almost the same hash rate but below 100% TDP and well below on the temps too. I love this 660 ti so much haha
And anyone with Mine-Litecoin that hasn't heard, site is back up and running beautifully
And anyone with Mine-Litecoin that hasn't heard, site is back up and running beautifully
December 20, 2013, 09:38:11 AM
getting 105-107 kHash/s from aGT 640 (GDDR5 model), overclocked. Wow.
Has anyone figured out yet which nVidia cards give
- best kHash/s per USD (or per EUR) ?
- best kHash/s per Watt (not just the GPU's TDP, but actual power draw including memory chips)?
I like the overclockability of GTX 780. Despite having just 12 SMX it seems to reach the same kHash/s levels as my 15 SMX wielding 780 Ti's. I can only configure 106% TDP, whereas the 780 seems to allow much more.
Christian
Has anyone figured out yet which nVidia cards give
- best kHash/s per USD (or per EUR) ?
- best kHash/s per Watt (not just the GPU's TDP, but actual power draw including memory chips)?
I like the overclockability of GTX 780. Despite having just 12 SMX it seems to reach the same kHash/s levels as my 15 SMX wielding 780 Ti's. I can only configure 106% TDP, whereas the 780 seems to allow much more.
Christian
You could try skyn3ts bioses (unlocks pretty much everything you'd want) http://www.overclock.net/t/1438886/official-nvidia-gtx-780-ti-owners-club (scroll down and read ^^)
December 20, 2013, 09:11:37 AM
Get 480 khash/s with Geforce GTX 780 but cannot use cudaminer because the pool don't detect the work.
How are you getting 480? I'm getting 310 on the 780
Something else: there's no way of setting an intensity for cudaminer or is there?
Edit: updated to version 12-18 and getting 480 khash/s now
December 20, 2013, 08:12:07 AM
Increase from dec 10 to dec 18 release:
GTX 780 (no overclock, capability 3.5): From 350 to 435 khash/s [-H 2 -d 0 -i 0 -l T6x22 -m 1 -C 0]
GTX 670 (no overclock, capability 3.0): From 188 to 235 khas/s [-H 2 -d 0 -i 0 -l K7x10 -C 0 -m 1]
GTX 570 (no overclock, capability 2.0): 236 khash/s unchanged [-H 2 -d 0 -i 0 -l F15x16]
Really nice improvement!
GTX 780 (no overclock, capability 3.5): From 350 to 435 khash/s [-H 2 -d 0 -i 0 -l T6x22 -m 1 -C 0]
GTX 670 (no overclock, capability 3.0): From 188 to 235 khas/s [-H 2 -d 0 -i 0 -l K7x10 -C 0 -m 1]
GTX 570 (no overclock, capability 2.0): 236 khash/s unchanged [-H 2 -d 0 -i 0 -l F15x16]
Really nice improvement!
December 20, 2013, 08:10:05 AM
Get 480 khash/s with Geforce GTX 780 but cannot use cudaminer because the pool don't detect the work.
December 20, 2013, 05:54:10 AM
getting 105-107 kHash/s from aGT 640 (GDDR5 model), overclocked. Wow.
Has anyone figured out yet which nVidia cards give
- best kHash/s per USD (or per EUR) ?
- best kHash/s per Watt (not just the GPU's TDP, but actual power draw including memory chips)?
I like the overclockability of GTX 780. Despite having just 12 SMX it seems to reach the same kHash/s levels as my 15 SMX wielding 780 Ti's. I can only configure 106% TDP, whereas the 780 seems to allow much more.
Christian
Has anyone figured out yet which nVidia cards give
- best kHash/s per USD (or per EUR) ?
- best kHash/s per Watt (not just the GPU's TDP, but actual power draw including memory chips)?
I like the overclockability of GTX 780. Despite having just 12 SMX it seems to reach the same kHash/s levels as my 15 SMX wielding 780 Ti's. I can only configure 106% TDP, whereas the 780 seems to allow much more.
Christian
December 20, 2013, 05:11:31 AM
2x 670 ftw are now getting 648 kh/s with K7x32
December 20, 2013, 05:09:51 AM
any body help me?
i using laptop. i set many line commond for cudaminer but dont working:
i using laptop. i set many line commond for cudaminer but dont working:
I wouldn't even bother trying to figure it out, to be honest. What is an 8400M going to get you in an ideal scenario, like 15 khash? It won't even come close to paying for the power, and to make it worse you're probably reducing the lifetime of a part that's already old and well used by now because it's in a laptop.
You are right!
ok thnaks bro December 20, 2013, 04:55:53 AM
any body help me?
i using laptop. i set many line commond for cudaminer but dont working:
i using laptop. i set many line commond for cudaminer but dont working:
I wouldn't even bother trying to figure it out, to be honest. What is an 8400M going to get you in an ideal scenario, like 15 khash? It won't even come close to paying for the power, and to make it worse you're probably reducing the lifetime of a part that's already old and well used by now because it's in a laptop.
December 20, 2013, 04:39:41 AM
any body help me?
i using laptop. i set many line commond for cudaminer but dont working:
i using laptop. i set many line commond for cudaminer but dont working:
December 20, 2013, 02:34:03 AM
hello! My gtx670 with settings -i 0 -H 1 -l K14x16 -C 2 looks not bad. About 280-290khash/s. GPU temp 85C, its not too high?
And one more question. another desktop with Quadro 600. Does anyone know how correctly configure it? Autosetting gives only about 20-30khashes((
And one more question. another desktop with Quadro 600. Does anyone know how correctly configure it? Autosetting gives only about 20-30khashes((
December 20, 2013, 01:40:13 AM
Wish we could get something going specifically for cudaminer that is better than the current google docs spreadsheet. It is constantly being screwed up.
December 20, 2013, 12:57:08 AM
My 770GTX produces 330kH/s at -i 0 -C 2 -m 1 -H 2 -l K16x16 (shitty CPU does not like -H 1 
)
However for 24h operation I had to lower voltage and underclock to get temps stabilize to 79 degrees of Celsius.(I will optimize this later) With underclock I get 300kH/s.
)However for 24h operation I had to lower voltage and underclock to get temps stabilize to 79 degrees of Celsius.(I will optimize this later) With underclock I get 300kH/s.
Jump to: