Overclock.net › Forums › Graphics Cards › NVIDIA › [Official] GTX 580 Overclocking Club
New Posts  All Forums:Forum Nav:

[Official] GTX 580 Overclocking Club - Page 268

post #2671 of 3582
Quote:
Originally Posted by dos659 View Post

I can get 830Mhz core in stock volts from my Asus GTX 580 DCII.

Passed: Batman:Arkham City, Battlefield 3, 3dmark 11, 3d mark vantage.

Max temp: 73c

Tried also at 840 and 850 on stock volts but driver is crashing and recovers while in game.

486

I think im ok with this overclock. What do you think?

Not bad, now give her some voltage. She can handle it.
550D
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-2700K Gigabyte Z77X-UD3H EVGA GTX 970 ACX 2.0+ SSC 4 x 4GB Samsung Green 
Hard DriveHard DriveCoolingOS
Crucial 240gb Samsung EVO 250gb Corsair H80 Win7 Ultimate 64 
MonitorKeyboardPowerCase
Qnix 2560x1440p Quickfire TK MX Brown XFX 750w Black Edition Corsair 550D 
MouseMouse Pad
Lefty Deathadder X Trac Ripper XXL 
  hide details  
Reply
550D
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-2700K Gigabyte Z77X-UD3H EVGA GTX 970 ACX 2.0+ SSC 4 x 4GB Samsung Green 
Hard DriveHard DriveCoolingOS
Crucial 240gb Samsung EVO 250gb Corsair H80 Win7 Ultimate 64 
MonitorKeyboardPowerCase
Qnix 2560x1440p Quickfire TK MX Brown XFX 750w Black Edition Corsair 550D 
MouseMouse Pad
Lefty Deathadder X Trac Ripper XXL 
  hide details  
Reply
post #2672 of 3582
Fixed fan settings:

375

This gives me 65C while gaming and 85C on OCCT 8 shader complexity on 1920x1080 full screen preset for 10mins. Ofcourse its error and artifact free!

If you feel something is wrong please say it biggrin.gif
Edited by dos659 - 1/23/12 at 6:41pm
post #2673 of 3582
Which version of OCCT did you use? I've used 3.1 and 4 and both have always been close to useless but I just noticed there's 4.1 Beta 8 which appears to stress the card (when you tick error checking).

But.. it still doesn't allow you to check it properly (as in I can't find any 100% sure ways to find artifacts). I raised my clocks to 865/2100 and it found no errors / problems after 10mins. Started 3DMark11 and the driver fell over. Backed down to 855/2100 and they passed and a quick game of BF3 showed no issues.
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
post #2674 of 3582
Quote:
Originally Posted by error-id10t View Post

Which version of OCCT did you use? I've used 3.1 and 4 and both have always been close to useless but I just noticed there's 4.1 Beta 8 which appears to stress the card (when you tick error checking).
But.. it still doesn't allow you to check it properly (as in I can't find any 100% sure ways to find artifacts). I raised my clocks to 865/2100 and it found no errors / problems after 10mins. Started 3DMark11 and the driver fell over. Backed down to 855/2100 and they passed and a quick game of BF3 showed no issues.

Ive found Heaven and 3dMark11 useful in artifact and stabliity checking. Furmark and OCCT are hopeless for my GTX 580 now frown.gif

Folding will always trip up an unstable OC on a GPU too
post #2675 of 3582
As im writing in previous posts i tried on 850 and 840 clocks but i was getting driver crash while gaming even if i tried 3d mark 11 and 3d mark vantage and ooct pass. So i seted to 830 and i can play games for 3 hours easily and pass all the previous programs. All this in stock voltages. My max temps while gemming and testing is 65C and when im on OCCT i get 85 for 10 mins with 8 shader complexity 1920x1080. I think my temps are normal so i keep it like that as my 24/7 overclock biggrin.gif
post #2676 of 3582
Quote:
Originally Posted by Sethy666 View Post

Quote:
Originally Posted by error-id10t View Post

Which version of OCCT did you use? I've used 3.1 and 4 and both have always been close to useless but I just noticed there's 4.1 Beta 8 which appears to stress the card (when you tick error checking).
But.. it still doesn't allow you to check it properly (as in I can't find any 100% sure ways to find artifacts). I raised my clocks to 865/2100 and it found no errors / problems after 10mins. Started 3DMark11 and the driver fell over. Backed down to 855/2100 and they passed and a quick game of BF3 showed no issues.

Ive found Heaven and 3dMark11 useful in artifact and stabliity checking. Furmark and OCCT are hopeless for my GTX 580 now frown.gif

Folding will always trip up an unstable OC on a GPU too


I agree with Sethy666. Folding is the best test for GPU stability. Furmark and OCCT are power-limited at the driver level and won't catch your artifacts anymore at high power. 3DMark11 and Heaven will crash if you get very unstable but can have a single frame with a tiny artifact in them that your naked eye may or may not catch.

 

error-id10t:

 

The GPU3 client with the -advmethods flag will catch instability; you're likely to get an "Error: Unstable Machine. NANs detected." or something to that effect if your OC is unstable. It should take less than a few hours to catch more prominent errors but you can be confident that you're close to true stability if you run it for about 48 hours. If you want ultimate stability, run it for 20 Work Units or so, which should take about 5 days on the -advmethods units that are currently available.

 

I had "stable" and "artifact free" clocks of 950 to 968 Mhz on the GPU Core Clock on my 580 and found that I wasn't truly stable and error free until I backed down to 900 MHz on the core. I was passing 3DMark11 and Heaven benchmark with no artifacts, at least no artifacts that I could see with my eyes, and yet my 968 clock got a "NAN" within five minutes of folding. The 950 clock lasted for about five WUs before it got a NAN and I tried a 925 and 918 clock but they had about a 5 to 10% failure rate on WUs over the course of a week or two of testing.


The 900 core has been 24/7 folding stable (on advmethods) since December 27th of last year, which is just a few days short of a month ago, and I can say with confidence that I am truly stable now.

 

The Folding@Home High Performance GPU3 Console Client can be found on this page:

http://folding.stanford.edu/English/DownloadWinOther

 

Here is a direct link to the client.

http://www.stanford.edu/~friedrim/Folding@home-Win32-GPU_XP-641.zip

 

Unzip it to a folder you want to run it in, create a shortcut for it and add the bolded section to the target field of the shortcut properties X:\whateverfolder\Folding@home-Win32-gpu.exe -configonly

 

Run that, leave everything blank or default except for the following fields:

 

Change "Acceptable size of work assignment and work result packets" to big

Change "Change advanced options" to yes

Change "Core Priority" to low (which is higher than idle)

Change "Interval, in minutes, between checkpoints" to 3 (this will make it use more cpu, and if you were folding for performance, setting this to 30 would be better, but since you're error testing, most frequent checkpoints are best)

Change "Set -advmethods flag always, requesting new advanced scientific cores and/or work units if available" to yes

Change "Machine ID (1-16) to 2 for your first GPU  and then follow the exact same procedures above and below, except change machine ID to 3 for your second GPU

Change "Disable CPU affinity lock" to yes (This will not lock the GPU process to one specific CPU core and will allow it to use whatever thread it feels is best. This may have no impact on folding stability testing, but it helped my performance quite a bit.)

Change "Additional client parameters" to -verbosity 9 (the - sign is important there)

 

At this point, the client will exit, as you told it to configure only.

 

Now delete the -configonly flag from your short-cut and run the file as normal. It will download a few things from Stanford, like the current core, and then download a project and begin folding. Ideally, you should get a 7620, 7621, or 7622 Work Unit, as those are the truly stressful units that will catch errors very quickly. If you get a 6800, 6801, 6802, 6803, or 6804 then let it finish and make sure that you have the "Set -advmethods flag always" option in your configuration set to yes (or alternatively, just add -advmethods onto the end of the shortcut where you had the -configonly flag at.

 

You can hit Ctrl-C at any time to close the process and it will resume from its last checkpoint when you start it up again. It will automatically download new Work Units and continue to thrash on your GPU until you stop it. It's best to let it finish a WU and not leave the WU unfinished, so when you get sick of folding and want to stop it, hit Ctrl-C to close the client, and then add the -oneunit flag at the end of the shortcut. That will instruct the client to not download more WU's, finish the WU you're on gracefully, and then shut down.

 

I'd suggest 48 hour minimum to get an idea of stability, once you think that 3DMark11 and Heaven are acceptable. You might be surprised at errors and have to back down. If you're obsessive about stability then a week to a month of 24/7 GPU folding on -advmethods will catch any error you're likely to have.

 

You can game with this running and you can do other things on your PC with this running. Its core priority is low enough that it will only grab idle GPU cycles (and barely above idle CPU cycles) and the -advmethods units use very little, if any, CPU. I play Dirt3 and Skyrim with folding running without any issues. Even BF3 works while folding, although the folding is so slow that it seems like it's stalled if I do that.

 

For maximum stress testing it would be best if you just folded the WUs with nothing else going on, so if you need to use your PC often, what you can do is give it the -oneunit flag and have it fold for 6 to 8 hours while you sleep each night. It won't be as good as continuous stress-testing, but after 10 to 20 nights of single unit completion, you should have a handle on your stability.

 

EDIT: Since you have two GPUs, you'd need to run a separate GPU client (installed to a separate directory) for each GPU to test stability, then use the lower of the two clocks for both GPUs for your SLI mode. You do not need to remove the SLI bridge or disable SLI. Folding will only work on one GPU at a time. It's important that each GPU get a different Machine ID as detailed above, so just follow the exact same procedure as above, but change the machine ID of your second card to a different number than the first card's.

 

According to Stanford's website:

 

 

Quote:
  • SLI or Crossfire does not have to be disabled to run the GPU client. However, SLI and CF cannot be used to make a dual GPU card or cards perform like one super GPU. The SLI and CF link does not have enough bandwidth to support that functionality. Run one FAH GPU client for each GPU chip.
  • If you are running multiple GPU cards with SLI or Crossfire, you may need to extend the desktop for the client to recognize multiple GPUs.
  • If you run Windows Vista or Windows 7, you may need to have each GPU connected to a monitor (a KVM can also work but not all models of KVM will provide a signal when it is not the active monitor). Another alternative is to use VGA dummy plug.

 

I'm not sure about the second two steps, but I know that the machine ID is important. You shouldn't have to mess with your SLI settings at all, as Folding will be single-card specific to each card.


Edited by shad0wfax - 1/24/12 at 4:56pm
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #2677 of 3582
So heres is a question. I have a GX750W psu, its consitered not that good according to JohnnyGuru (down right POS) and to add to that ive runn the heck out of it the last few years, like really hard. Most gaming sessions its screaming and its also starting to shutdown the pc at times. Now ive ordered an new psu and I hope to redo all my old benchmark runs with it. The new psu is consitered a very good one with very stable power. After installing the new one could i expect a better oc and benchmark runs.
My card according to reviews can suck down as much as 650w when overvolted and overclocked (950@1.50v) and ive run it alot harder than that (1027@1.293v). My 1055t is also sitting at 4ghz (also overvolted) and overvolted ram. I have a H100 and H70 (for gpu cooling) about 12 fans, few ssd and couple of hdd and a bunch of stuff hooked to the usb ports (k/m optical drive, pen pad-otherstuffs). Point is i was thinking i was sucking up all the juice and not sure if the card was getting all the power it could get before. The new psu is a seasonic 1000w platinum.


So what do you think, is there room for improvement or what.
post #2678 of 3582
I agree with shadowfax that the advmethods GPU work units are the most stressful thing I've found to run on the 500 series cards. When I was more active in this thread I used to recommend them often as the best and most stressful test for stability..

I do want to add though that you should be fairly confident that your OC is stable before trying to fold on it, if you are constantly failing WUs then you are actively hurting the research being done by folding@home.When you download a work unit and then fail it, it means that nobody else who would have completed that WU can get it for about 2 weeks. For anybody that wants to use folding as a stability testing I would recommend that you successfully complete twice as many work units as you fail so as to not being detrimental to the research being done. If you find yourself failing WUs within a few minutes of starting then you know you need to make the card drastically more stable by either increasing voltage or decreasing clocks by a fair amount, don't fail 10 WUs in a row because you are only trying to lower the clocks by 5 MHz at a time start lower and work your way up rather than the opposite. Feel free to PM me or check out the folding section if you need some help getting folding setup or have any other questions about it.
Main Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 2700k ASUS P8P67 WS Revolution EVGA 980 Ti SC+ Samsung 4x4GB DDR3 1866MHz 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 1TB Samsung Spinpoint F4 2TB Samsung BD Combo Noctua NH-D14 
OSMonitorPowerCase
Windows 10 64 bit Asus PG279Q Kingwin Lazer Platinum 1000W Silverstone Raven RV03 
  hide details  
Reply
Main Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 2700k ASUS P8P67 WS Revolution EVGA 980 Ti SC+ Samsung 4x4GB DDR3 1866MHz 
Hard DriveHard DriveOptical DriveCooling
Samsung 850 Evo 1TB Samsung Spinpoint F4 2TB Samsung BD Combo Noctua NH-D14 
OSMonitorPowerCase
Windows 10 64 bit Asus PG279Q Kingwin Lazer Platinum 1000W Silverstone Raven RV03 
  hide details  
Reply
post #2679 of 3582
Quote:
Originally Posted by juano View Post

I agree with shadowfax that the advmethods GPU work units are the most stressful thing I've found to run on the 500 series cards. When I was more active in this thread I used to recommend them often as the best and most stressful test for stability..

I do want to add though that you should be fairly confident that your OC is stable before trying to fold on it, if you are constantly failing WUs then you are actively hurting the research being done by folding@home.When you download a work unit and then fail it, it means that nobody else who would have completed that WU can get it for about 2 weeks. For anybody that wants to use folding as a stability testing I would recommend that you successfully complete twice as many work units as you fail so as to not being detrimental to the research being done. If you find yourself failing WUs within a few minutes of starting then you know you need to make the card drastically more stable by either increasing voltage or decreasing clocks by a fair amount, don't fail 10 WUs in a row because you are only trying to lower the clocks by 5 MHz at a time start lower and work your way up rather than the opposite. Feel free to PM me or check out the folding section if you need some help getting folding setup or have any other questions about it.



Thank you for emphasizing the point about the quality of the scientific analysis results supplied to the folding team. I should have emphasized that more in my post and I did not. I mentioned my significant reduction in clock speed after the relatively immediate NAN and that I further reduced clock speeds after subsequent (and far less frequent NANs, 1/10 and 1/20 respectively.)

 

To add the emphasis that I should have included:

 

I did not begin folding to test until I was fairly confident in my artifact-free stability in Heaven and in 3DMark11 and when I had a failed WU, I did as you suggest above and I reduced my clock by 24 MHz immediately. I then went about 10 or 12 WUs before I had another failure and reduced my clock another 7 MHz. I went 20 WUs and had another failure and reduced my clock again by 18 MHz and have not failed since. I've completed about 150 -advmethods GPU WUs in total, and had 3 or possibly 4 failures. I've completed 134 WUs (73 of them -advmethods WUs) since then with no failures. Rather than increase my clock by 5-6 MHz until I started failing again, I worked at using my existing clock and reducing core voltage for more thermal efficiency.

 

I took the failures that I had seriously and made a conscious effort to be a far greater benefit to the folding research than I was a detriment.

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #2680 of 3582
Quote:
Originally Posted by cravinmild View Post

So heres is a question. I have a GX750W psu, its consitered not that good according to JohnnyGuru (down right POS) and to add to that ive runn the heck out of it the last few years, like really hard. Most gaming sessions its screaming and its also starting to shutdown the pc at times. Now ive ordered an new psu and I hope to redo all my old benchmark runs with it. The new psu is consitered a very good one with very stable power. After installing the new one could i expect a better oc and benchmark runs.
My card according to reviews can suck down as much as 650w when overvolted and overclocked (950@1.50v) and ive run it alot harder than that (1027@1.293v). My 1055t is also sitting at 4ghz (also overvolted) and overvolted ram. I have a H100 and H70 (for gpu cooling) about 12 fans, few ssd and couple of hdd and a bunch of stuff hooked to the usb ports (k/m optical drive, pen pad-otherstuffs). Point is i was thinking i was sucking up all the juice and not sure if the card was getting all the power it could get before. The new psu is a seasonic 1000w platinum.


So what do you think, is there room for improvement or what.


If you're experiencing random shut-downs from your power supply then it's a good thing that you've ordered a replacement. I suggest that you do not use your PC at all until the new one arrives, as a random shut-down due to under/over-voltage or under/over-current can ruin other components in your system such as motherboards, CPUs, GPUs, RAM, or HDDs. (I once ruined a motherboard and caused mild damage to a CPU due to a catastrophic power supply failure. The CPU would not overclock with stability at all after the failure but performed at stock speeds.)

 

You may or may not expect a better OC and benchmark runs than you currently have; the answer isn't black and white, as you don't know if your previous runs were power supply limited, CPU limited, GPU limited, FSB limited, VRM limited, etc.You can expect is better system stability with no more random shut-downs. I think that you'll be very happy with your new PSU and you might be able to squeeze a bit more out of your system. My guess is that you're going to be limited by motherboard and GPU VRM stability and that it's unlikely that PSU stability will increase performance over what you saw prior to PSU failures.

 

P.S. Your GPU voltage sounds awfully high at 1.50V. Was that a typo and you meant 950@1.150V instead of 950@1.50V ? That would make more sense in comparison to your 1027@1.293V if that were the case.

 

 

EDIT: Shoot, I should have multi-quoted instead of replying separately. Oh well...

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › [Official] GTX 580 Overclocking Club