GTX 560ti Users: Information Needed


log in

Advanced search

Message boards : Number crunching : GTX 560ti Users: Information Needed

Previous · 1 · 2 · 3 · 4 · 5
Author Message
Profile razamatraz
Send message
Joined: 23 Oct 07
Posts: 142
Credit: 27,807,059
RAC: 0
Canada
Message 1130293 - Posted: 21 Jul 2011, 1:22:34 UTC
Last modified: 21 Jul 2011, 1:24:04 UTC

1. Corsair 1200 AX, Q6600, EVGA GTX 560 Ti DS SC (the dual fan model), Gigabyte GTX 460 OC 1GB, Win 7 x64, 275.33 driver

2. LAtest version of the unified installer

3. I was, but it`s fixed. In my case is was the EVGA utility attempting to synchronize the 560 with the 460 and then never recovering. If I just don`t use the utility it is fine.

4. 460 runs at 70C under full load (24C ambient) 560 Ti at 80C. Fan speed profile is stock though so only 60% speed there.

5. Stock clocks and voltage (from the manufacturer, both are OC versions though 460 is at 715,900,1430,0.987V. 560Ti is at 900,1053,1800,1.05V.

6. This is left on Adaptive.

7. No


This machine is actually really starting to fly...made top 100 today.

Profile M_M
Avatar
Send message
Joined: 20 May 04
Posts: 27
Credit: 15,402,706
RAC: 4,389
Serbia
Message 1130322 - Posted: 21 Jul 2011, 5:05:56 UTC
Last modified: 21 Jul 2011, 5:13:51 UTC

1. Gigabyte GTX560Ti with Nvidia 275.33 drivers, Corsair HX750 PSU, MSI P67-GD65, i7-2600k, Win7x64 Home Premium,
2. x38g, 2 tasks in parallel (to improve efficiency)
3. downclocking from time to time, on average once to twice a day. About 2-3% of WUs finishes with error or completed mark as invalid, not sure is it card related or simply a WU nature. All WUs are processed on GPU.
4. up to 65C (really efficient windforce stock cooler)
5. default, 900Mhz GPU, 2000 MHz Memory. GPU vultage 1.00V
6. Adaptive
7. No.

Never ever had any kind of problem playing any game, and I am pretty sure I didn't experiance this with older version of lunatic app and 266.66 drivers (everything else was the same as above).
____________

Profile Slavac
Volunteer tester
Avatar
Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1130327 - Posted: 21 Jul 2011, 5:35:44 UTC - in response to Message 1130322.

1. Gigabyte GTX560Ti with Nvidia 275.33 drivers, Corsair HX750 PSU, MSI P67-GD65, i7-2600k, Win7x64 Home Premium,
2. x38g, 2 tasks in parallel (to improve efficiency)
3. downclocking from time to time, on average once to twice a day. About 2-3% of WUs finishes with error or completed mark as invalid, not sure is it card related or simply a WU nature. All WUs are processed on GPU.
4. up to 65C (really efficient windforce stock cooler)
5. default, 900Mhz GPU, 2000 MHz Memory. GPU vultage 1.00V
6. Adaptive
7. No.

Never ever had any kind of problem playing any game, and I am pretty sure I didn't experiance this with older version of lunatic app and 266.66 drivers (everything else was the same as above).


Bump your voltage to 1.062 and see if that doesn't fix your downclocking. Worked for me.
____________


Executive Director GPU Users Group Inc. -
brad@gpuug.org

Profile M_M
Avatar
Send message
Joined: 20 May 04
Posts: 27
Credit: 15,402,706
RAC: 4,389
Serbia
Message 1134432 - Posted: 31 Jul 2011, 7:32:19 UTC - in response to Message 1130327.
Last modified: 31 Jul 2011, 7:38:31 UTC

I have raised the voltage to 1.012V (from default 1.000V), so just one step, and in last 4-5 days I have not experianced any downclocks or error WUs (only three completed, marked as invalid WU, 160+ completed and verified WUs, and many still pending results).

So it seems that just the slight voltage increase could help. Seems that on my particular Gigabyte 560Ti-OC factory overclock was a bit overoptimistic in terms of GPU voltage needed for default 900MHz, 1.000V works fine in any game (even 920MHz is ok for games), but not ok for lunatic x38 app, which required at least 1.012V for my particular card/gpu on stock 900MHz.
____________

Profile arkaynProject donor
Volunteer tester
Avatar
Send message
Joined: 14 May 99
Posts: 3641
Credit: 48,594,900
RAC: 5,703
United States
Message 1134686 - Posted: 1 Aug 2011, 2:03:01 UTC

Look like my non-Ti GTX560 is running at 0.987 on the voltage.

I did have to download the beta version of Afterburner to find it out though.
____________

Profile Slavac
Volunteer tester
Avatar
Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1134697 - Posted: 1 Aug 2011, 2:33:54 UTC - in response to Message 1134432.

I have raised the voltage to 1.012V (from default 1.000V), so just one step, and in last 4-5 days I have not experianced any downclocks or error WUs (only three completed, marked as invalid WU, 160+ completed and verified WUs, and many still pending results).

So it seems that just the slight voltage increase could help. Seems that on my particular Gigabyte 560Ti-OC factory overclock was a bit overoptimistic in terms of GPU voltage needed for default 900MHz, 1.000V works fine in any game (even 920MHz is ok for games), but not ok for lunatic x38 app, which required at least 1.012V for my particular card/gpu on stock 900MHz.


Yep. Mine need 1.062V at least or they start throwing errors.

It's interesting to me that these cards need more juice than their stock settings on some machines but on others, no worries.
____________


Executive Director GPU Users Group Inc. -
brad@gpuug.org

Profile BMH
Avatar
Send message
Joined: 27 May 99
Posts: 324
Credit: 90,138,472
RAC: 99,158
United Kingdom
Message 1134958 - Posted: 1 Aug 2011, 21:57:30 UTC

Because the stock frequencies of the 560ti aren't that much above those of the 460 (while the latter has fewer cores - I read 7% in computer magazine I read), manufacturers feel a greater need to play with the clock speeds to tempt consumers with OCd versions, and on top of thise maybe these cards are just a bit more sensitve when overclocked/what voltages are set at or the manufacturers aren't getting the settings quite right... and when you run a GPU, like a CPU, at full load, as when crunching for SETI, those frequencies and voltages have to be just right for things to remain stable.

Fluctuations elsewhere in the system, such as in the power from the PSU can easily prevent an overclock from being stable, and manufacturers of GPUs don't know what our PSUs are like - so personally I don't like pre-overclocked GPUs... but I like the ones with decent coolers and it seem you can't have the latter without the former.

We don't get pre-overclocked CPUs so it annoys me a little that the market is flooded with pre-overclocked GPUs, just to persuade consumers.
____________
Brian.

Pilgrim57
Avatar
Send message
Joined: 28 Jun 02
Posts: 6
Credit: 6,351,765
RAC: 519
United Kingdom
Message 1135590 - Posted: 3 Aug 2011, 18:05:02 UTC - in response to Message 1127468.

I'm noticing a trend that seems to suggest that there's something buggy with the 560ti card as a whole, or at least insofar as its interactions with SETI/Lunatics.

Please take a moment to fill out the information below so we can help the Lunatics builders hammer out this problem. The problem will likely become exaserbated eventually as the 560ti prices continue to become more affordable for more crunchers.

1. Please list your system specs. PSU, GPU's(their make and model), OS, CPU and so forth.
2. Which version of Lunatics are you running?
3. Are you experiencing errors or downclocking? If so, please explain/expand on what issues you are having. For example your cards downclock every 1 hour and in doing so, throw an error message.
4. At what temps are your cards running at full load?
5. What clocks/voltage are you running on your cards?
6. Have you set your NVIDIA power performance to 'max performance' rather than 'adaptive'?
7. Have you disabled hardware acceleration in your web browser?


1.Corsair HX 850W ATX Modular SLI Compliant Power Supply 70amps on 1 x 12v rail
Q9550, Win 7 64 bit SP1, 4GB Cosair defender ram. ASUS P5Q-E mobo
One MSI Twin Frozr 560Ti
2. 0.38
3.Yes. Don't run Seti cuda all the time. Sometimes once or twice in 8 hours drops core to 405MHz. Have looked at error message Boinc say computing error.
4. 60C or less
5. Stock/default 880 core, 1760 Shader, 1050/2100 MHz memory
925/1850/2150 runs 24/7 with F@H but errors with cuda
921/1842/2150 fewer errors once in 8-10 hrs
910/1820/2120 haven't had an error.
Standard volts
6. Adaptive
7. no


Errors can happen when PC is running without other use.
Seem to start with .38 + 27X.xx drivers
Have MSI Afterburner installed with above profiles. Have upped fan speed to keep card cooler. Fan speed a few % below temp eg 51%/55C
I only run one wu at a time. Haven't got round to the edit to make it run 2 [was unsure if t could. GPU usage 70-90% latter being peaks, average about 50%
N.B. F@H runs 98-99% & use more memory.

Thought it would be a driver/optimised client issue rather than power as F@H stresses the card more.


____________

Profile BMH
Avatar
Send message
Joined: 27 May 99
Posts: 324
Credit: 90,138,472
RAC: 99,158
United Kingdom
Message 1135902 - Posted: 4 Aug 2011, 15:08:15 UTC - in response to Message 1135590.

1.Corsair HX 850W ATX Modular SLI Compliant Power Supply 70amps on 1 x 12v rail
Q9550, Win 7 64 bit SP1, 4GB Cosair defender ram. ASUS P5Q-E mobo
One MSI Twin Frozr 560Ti
2. 0.38
3.Yes. Don't run Seti cuda all the time. Sometimes once or twice in 8 hours drops core to 405MHz. Have looked at error message Boinc say computing error.

Errors can happen when PC is running without other use.
Seem to start with .38 + 27X.xx drivers
Have MSI Afterburner installed with above profiles.

I am also using a MSI Twin Frozr 560Ti, with no downclocking issues (unless I overclock). The only difference (besides not using MSI Afterburner) really is I'm using nvidia 270.61 drivers, so you might want to try that instead?
____________
Brian.

Profile Eve - Pyro nl
Send message
Joined: 5 Jul 06
Posts: 57
Credit: 22,550,054
RAC: 5,454
Netherlands
Message 1137714 - Posted: 8 Aug 2011, 20:41:40 UTC - in response to Message 1135902.
Last modified: 8 Aug 2011, 20:44:14 UTC

trowing in a noobquestion ;p

i bought some msi 650 ti hawk cards while running normal boinc without lunatic

what was it too get more units to work on my card as atm ony runs 1 at a time

and should i then go for 2 at once or 3, im thinking 2 as im also playing games while boinc is running

also got 2 gtx 260 working in sli, with atm both 1 unit mayby i should increase that1 aswel
____________
Joined Seti@home on 20 June 1999 (id:797875)
Made This1 to Join the Eve Team

http://www.eveonline.com/seti/

Profile Slavac
Volunteer tester
Avatar
Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1137716 - Posted: 8 Aug 2011, 20:46:32 UTC - in response to Message 1137714.
Last modified: 8 Aug 2011, 20:46:50 UTC

trowing in a noobquestion ;p

i bought some msi 650 ti hawk cards while running normal boinc without lunatic

what was it too get more units to work on my card as atm ony runs 1 at a time

and should i then go for 2 at once or 3, im thinking 2 as im also playing games while boinc is running

also got 2 gtx 260 working in sli, with atm both 1 unit mayby i should increase that1 aswel


3 per card on a 560ti is fine. It will not run 3 AP tasks per card, only 2.

The general rule is 3 MB units per 1 Gig on the card.
____________


Executive Director GPU Users Group Inc. -
brad@gpuug.org

Profile Eve - Pyro nl
Send message
Joined: 5 Jul 06
Posts: 57
Credit: 22,550,054
RAC: 5,454
Netherlands
Message 1139547 - Posted: 12 Aug 2011, 20:13:19 UTC - in response to Message 1137716.

wich file do i need to edit to be able to use more then 1 unit on my gpu ?
____________
Joined Seti@home on 20 June 1999 (id:797875)
Made This1 to Join the Eve Team

http://www.eveonline.com/seti/

Profile perryjay
Volunteer tester
Avatar
Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 15,541,019
RAC: 11,430
United States
Message 1139554 - Posted: 12 Aug 2011, 20:29:18 UTC - in response to Message 1139547.

In your app_info find the count..

<coproc>
<type>CUDA</type>
<count>.5</count>
</coproc>

Yours should read one. Change it to .5 for two or .33 for three at a time. If you are running a 64 bit system then you should find six instances of <count> if you have a 32 bit system there will only be three. Change them all. This is for the MB CUDA work. If you want to try to run AP on your NV GPUs then come back and ask about that and we will show you how to change that.
____________


PROUD MEMBER OF Team Starfire World BOINC

Profile Eve - Pyro nl
Send message
Joined: 5 Jul 06
Posts: 57
Credit: 22,550,054
RAC: 5,454
Netherlands
Message 1139561 - Posted: 12 Aug 2011, 20:57:36 UTC - in response to Message 1139554.

In your app_info find the count..

<coproc>
<type>CUDA</type>
<count>.5</count>
</coproc>

Yours should read one. Change it to .5 for two or .33 for three at a time. If you are running a 64 bit system then you should find six instances of <count> if you have a 32 bit system there will only be three. Change them all. This is for the MB CUDA work. If you want to try to run AP on your NV GPUs then come back and ask about that and we will show you how to change that.



ok thx :)
____________
Joined Seti@home on 20 June 1999 (id:797875)
Made This1 to Join the Eve Team

http://www.eveonline.com/seti/

Previous · 1 · 2 · 3 · 4 · 5

Message boards : Number crunching : GTX 560ti Users: Information Needed

Copyright © 2014 University of California