Overclock.net › Forums › Graphics Cards › NVIDIA › GTX 570 overclocking thread
New Posts  All Forums:Forum Nav:

GTX 570 overclocking thread - Page 144

post #1431 of 2146
Quote:
Originally Posted by Cpt.Teacup View Post

Make sure you unlock voltage control in the settings.

I found the problem, it was my mistake though I was raising 1 voltage per time lol.

Gear mentioned in this thread:

post #1432 of 2146
Quote:
Originally Posted by Cpt.Teacup View Post

I got my PNY GTX 570 to 900/2200 @ 1.1v, but I have some questions before I commit to it.
I ran an hour of OCCT with shader complexity on 8 and got 17 errors. Should I lower my clocks until there's 0 errors, or will it be fine for gaming? When I tested an hour with GPU Tool before on 1.075v there were 0 errors.

I would personally run Unigine overnight to see if it is game stable. For things other than gaming, dropping the core clock 5MHz should do it.

post #1433 of 2146
Quote:
Originally Posted by Rops84 View Post

@ .Griff.
My best guess is to back down on memory OC...sounds to me like a memory related issue...

The card was RMA'd a couple of weeks ago. It was indeed faulty vram.
Gaming Rig
(16 items)
 
Laptop
(6 items)
 
 
CPUMotherboardGraphicsGraphics
i5 4670K @ 4.6GHz MSI Z87 GD65 Gaming MSI R9 280x Gaming MSI R9 280x Gaming 
RAMHard DriveCoolingOS
16GB Patriot Viper Samsung 840 Pro Corsair H100 Windows 8 x64  
MonitorKeyboardPowerCase
Achieva Shimian QH270-IPSB Logitech G19 Corsair HX 750 Corsair 600T SE 
MouseMouse PadAudio
Razer Salmosa Alioth 320 Creative Sound Blaster Z 
CPURAMHard DriveOptical Drive
Intel® Pentium® Dual-Core Processor T2060  DDR2 RAM (533 MHz)  5,400 rpm DVD Super Multi (Double Layer) drive  
OSMonitor
Windows 7 x86 Ultimate size : 15.4 " type : WXGA TFT display interna... 
  hide details  
Reply
Gaming Rig
(16 items)
 
Laptop
(6 items)
 
 
CPUMotherboardGraphicsGraphics
i5 4670K @ 4.6GHz MSI Z87 GD65 Gaming MSI R9 280x Gaming MSI R9 280x Gaming 
RAMHard DriveCoolingOS
16GB Patriot Viper Samsung 840 Pro Corsair H100 Windows 8 x64  
MonitorKeyboardPowerCase
Achieva Shimian QH270-IPSB Logitech G19 Corsair HX 750 Corsair 600T SE 
MouseMouse PadAudio
Razer Salmosa Alioth 320 Creative Sound Blaster Z 
CPURAMHard DriveOptical Drive
Intel® Pentium® Dual-Core Processor T2060  DDR2 RAM (533 MHz)  5,400 rpm DVD Super Multi (Double Layer) drive  
OSMonitor
Windows 7 x86 Ultimate size : 15.4 " type : WXGA TFT display interna... 
  hide details  
Reply
post #1434 of 2146
Quote:
Originally Posted by Cpt.Teacup View Post

I got my PNY GTX 570 to 900/2200 @ 1.1v, but I have some questions before I commit to it.
I ran an hour of OCCT with shader complexity on 8 and got 17 errors. Should I lower my clocks until there's 0 errors, or will it be fine for gaming? When I tested an hour with GPU Tool before on 1.075v there were 0 errors.
This is from a test with OCCT:
330
There's a difference between the set voltage in MSI Afterburner and the reported voltage in GPU-Z, what exactly does this mean?
I've heard that running at 1.1v can be unsafe, should I be worried? GPU-Z doesn't even report it going up to 1.1v so I'm not really sure what I should follow.
Edit:
I've been doing some testing with MSI Kombustor 2.2 and found some interesting results.
In every test running at 900/2200 with 1.1v had a few fps decrease over 1.075v. I haven't done a long test with 1.075v in OCCT, but for the short time that I did it had a similar number of errors as 1.1v.
The other odd thing was that in the tessellation tests the standard clocks performed better than the OC, but the OC performed better in the combined test.
I'm also worried about the card's performance overall. When I tested with 3DMark11 with the standard settings on the card, the score was way lower than it should be for the card. I just tested again with the OC and it was still below the target score for standard clocks.
Standard clocks (732/1900 0.975v)
OC (900/2200 1.1v)
Any ideas what the problem is and if it's fixable?

732/1900 0.975v (stock)
700

900/2200 1.075v
700

(Nvidia's settings were at default.)
So is this card slow like 3DMark says?
I'm going to do some more messing around with the voltage to see how it runs. How many errors can I get in OCCT and still consider it stable for gaming and general use?

Edit:
Quote:
Originally Posted by nawon72 View Post

I would personally run Unigine overnight to see if it is game stable. For things other than gaming, dropping the core clock 5MHz should do it.

Thanks, I'll be sure to do that once I find a good voltage.
Any ideas why less voltage increased the performance in Kombustor?
Edited by Cpt.Teacup - 12/4/11 at 12:20pm
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
post #1435 of 2146
(See above post for more information.)

So I've been trying all kinds of settings and then testing them in Furmark 1.9.1 and am confused by the results.
I benched the stock clocks (732/1900 0.975v) and got a score of 1216.
Then I benched with 900/2200 1.075v and the score dropped to 888
I did each bench twice to be sure.
Now, on almost all of the clocks I tried the fps varied a lot and caused the animation to stutter constantly (is this normal?) However, when I ran the bench using 870/1900 0.988v the fps stayed rock steady at 19 for nearly the whole test, so there was no stuttering. It performed the same using 1.013v and 1.025v.
This doesn't seem right. The stock clocks, although stuttering, provided an average fps of 20, while every other clocks and voltages I tried gave an average of 19 or less. Except for the second test that scored 888, all of the tests were run with the memory at the stock 1900MHz, I only changed the core speed and the voltage.
I'm going to play with it some more, but I'm not really sure what's going on here. Other benchmarks like Heaven and Kombustor had a clear increase between stock clocks and the OC, but Furmark shows the opposite. Am I doing something wrong, or is there something wrong with my card? If it matters, I'm using a reference card from PNY, I've been using it at stock clocks for almost a year.
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
post #1436 of 2146
Did you increase voltage on the memory for that 2200? While its possible to run 2200 without its abit on the side of luck. I run 2200 or 2300 around +30 to +50 mv.

But yeah unstable card will get a worse score. 2200 Memory on stock voltage is easily unstable. If you ran OCCT after a few minutes you would register plenty of errors that your not seeing. If an error occurs the graphics card does not necessarily render it. If it realizes and error has occurred it can try to rerender a frame. This will slow the framerate, performance, score down but still provide a satisfactory visual result.

Its odd furmark would be running into memory issues that Unigine is not but still possible.
RainONE
(13 items)
 
ASUS G50VT
(13 items)
 
 
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5Ghz ASUS P8Z68 V PRO GTX 570 MSI TF III @ 950/1900/2300 16GB Corsair Vengeance 
Hard DriveOSKeyboardPower
2TB Seagate 5900 RPM Windows 7 x64 Ultimate Ideazon Z Board Seasonic X750 
CaseMouse
Thermaltake V9 BlacX Logitech G700 
CPUGraphicsRAMOS
Intel Core 2 Duo T9600 2.8Ghz @ 3.1Ghz Nvidia 9800M GS (OC @ 640/1600/800) 4GB DDR2 800Mhz Windows 7 
Mouse
Logitech G700 
  hide details  
Reply
RainONE
(13 items)
 
ASUS G50VT
(13 items)
 
 
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5Ghz ASUS P8Z68 V PRO GTX 570 MSI TF III @ 950/1900/2300 16GB Corsair Vengeance 
Hard DriveOSKeyboardPower
2TB Seagate 5900 RPM Windows 7 x64 Ultimate Ideazon Z Board Seasonic X750 
CaseMouse
Thermaltake V9 BlacX Logitech G700 
CPUGraphicsRAMOS
Intel Core 2 Duo T9600 2.8Ghz @ 3.1Ghz Nvidia 9800M GS (OC @ 640/1600/800) 4GB DDR2 800Mhz Windows 7 
Mouse
Logitech G700 
  hide details  
Reply
post #1437 of 2146
How would I change the memory voltage with MSI Afterburner?

I got it running 900/2200 1.013v stable in Furmark and it scored better than when I used 1.075v but it still scored less than the stock settings. It also performed exactly the same as when I ran it before with 900/1900 1.013v, so it doesn't appear to be an issue with memory. I will try some other benchmarks to see how they compare.

Edit:
I just did some tests with OCCT. The first time I got tons of errors popping up constantly for some reason, I think I had the wrong voltage set, so I ended that quickly and rechecked my settings. Once I made sure the card was running at 900/2200 1.013v, I ran a 5 minute test and got 0 errors, I then ran a 5 minute test with stock clocks and had no errors, then I ran another 5 minute test with the OC and once again had 0 errors. I will do an hour long test later to be sure, but usually when I use OCCT, if there are errors, they start showing up within the first 5 minutes.

When I ran the OC tests the fps stayed at 103 for the entire 5 minutes, but when I tested with the stock clocks the fps was constantly changing anywhere from 85 to 135. Even though there's no clear fps increase with the OC, it seems to greatly increase the stability.

I also ran 10 passes of the memtest and there were no errors.

I'll test with Heaven next.

Edit 2:
Well, I ran Heaven with the OC twice and it crashed both times. It was on stage 3 where the camera is below the airship. I also noticed some artifacts in the stages before it crashed.
I'll try increasing the voltage and see what happens.

Edit 3:
I got Heaven to run a benchmark at max settings without noticeable artifacts with the card running at 900/2200 1.063v.
Now to see if OCCT still likes it.

Edit 4:
As expected from my crappy luck, 5 minutes of OCCT with 900/2200 1.063v had 47 errors. I ran 10 passes of the memtest again though and it had 0 errors.
After that failure, I tried it with 1.013v again since it had 0 errors before, but this time there were instantly a bunch of errors. I have no idea why it worked the first time, but it's obviously not stable.

I'm going to take a break for tonight so this should be the last edit for this post. Please let me know what you guys think and what I should do next, I'm just playing with it by now, I don't really know what the deal is anymore.
Make sure you read my above posts too before you respond, there's a lot more information in them!
Edited by Cpt.Teacup - 12/4/11 at 6:56pm
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
Wannabe
(15 items)
 
  
CPUMotherboardGraphicsRAM
i5 3570K GA-Z77X-UD3H ASUS GTX 760 DirectCU II OC 2x4GB Samsung DDR3 
Hard DriveHard DriveCoolingOS
128GB Crucial M4 1TB WD Green Thermaltake Water 2.0 Pro Arch Linux 
MonitorKeyboardPowerCase
Dell U2412M 1920x1200 Microsoft Sidewinder x4 Thermaltake Toughpower XT 775W Cooler Master Haf X 
MouseMouse PadAudio
Logitech G700s SteelSeries Experience I-2 Beyerdynamic DT 770 PRO 80 ohms 
  hide details  
Reply
post #1438 of 2146
Quote:
Originally Posted by Cpt.Teacup View Post
So I've been trying all kinds of settings and then testing them in Furmark 1.9.1 and am confused by the results.
I benched the stock clocks (732/1900 0.975v) and got a score of 1216.
Then I benched with 900/2200 1.075v and the score dropped to 888
I did each bench twice to be sure.
Now, on almost all of the clocks I tried the fps varied a lot and caused the animation to stutter constantly (is this normal?) However, when I ran the bench using 870/1900 0.988v the fps stayed rock steady at 19 for nearly the whole test, so there was no stuttering. It performed the same using 1.013v and 1.025v.
This doesn't seem right. The stock clocks, although stuttering, provided an average fps of 20, while every other clocks and voltages I tried gave an average of 19 or less. Except for the second test that scored 888, all of the tests were run with the memory at the stock 1900MHz, I only changed the core speed and the voltage.
I'm going to play with it some more, but I'm not really sure what's going on here. Other benchmarks like Heaven and Kombustor had a clear increase between stock clocks and the OC, but Furmark shows the opposite. Am I doing something wrong, or is there something wrong with my card? If it matters, I'm using a reference card from PNY, I've been using it at stock clocks for almost a year.

OCCT and Furmark cause your GTX570's OCP(Over Current Protection) to kick in. Your score is lower on the OC because the GTX570 is trying to stay within it's TDP, and it does this by modulating the clock speed. More info on this here(GTX580, but applies to GTX570):

http://www.techpowerup.com/134460/Disable-GeForce-GTX-580-Power-Throttling-using-GPU-Z.html?cp=3

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_580/3.html

Charts to illustrate OCP:

gtx580_power_gpuz.jpg

limit2_small.jpg

I highly recommend you leave OCP enabled since you have a reference 570.

 

The stuttering you mentioned could have been cause by the clock modulation.

 

Some OC'ing tips based on my experience:

When your overclocking your GTX570, start by increasing the core clock as high as it will go then test for stability. Once your sure it is completely stable(i run Heaven 12h+), then you can start increasing the memory clock, but i wouldn't go over 2200MHz(2100MHz for everyday use). The benefits of a memory clock higher than 2000 are minimal(I think it was 20 points per 100MHz) based on my 3DMark11 runs, and not worth it IMO(and i read high memory clocks on reference 570's can kill it).

post #1439 of 2146
Quote:
Originally Posted by RainMotorsports View Post

Did you increase voltage on the memory for that 2200? While its possible to run 2200 without its abit on the side of luck. I run 2200 or 2300 around +30 to +50 mv.
But yeah unstable card will get a worse score. 2200 Memory on stock voltage is easily unstable. If you ran OCCT after a few minutes you would register plenty of errors that your not seeing. If an error occurs the graphics card does not necessarily render it. If it realizes and error has occurred it can try to rerender a frame. This will slow the framerate, performance, score down but still provide a satisfactory visual result.
Its odd furmark would be running into memory issues that Unigine is not but still possible.

Yup I use memory 2100mhz from stock 2000mhz for stable daily gaming, sometimes 2200mhz stable but not for long because mycard doesnt has memory volt like TFIII smile.gif so best stay on 940/2100 1.088mhz stable daily BF3, Batman Arkham City etc biggrin.gif
But for benchmark I can use 2200mhz of RAM speed. So far max I get is 970/2100 with 1.1volt.... Also to MrTooShort : I have done 50loop Crysis 1 benchmarking at 940/2100.
SkynetX
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Xeon E5-2623 V3 @3GHz running 3.3GHz MSI X99A SLI Krait Edition Inno3D GTX980Ti Ichill Black Hybrid  Gskill Ripjaws V 4x4G DDR4 2400MHz 
Hard DriveOptical DriveCoolingOS
Seagate Barracuda HDD 1TB SATA3 SAMSUNG Bluray Combo SH-B123L Cooler Master Hyper 212X Turbo Windows 10 Pro 64bit 
MonitorKeyboardPowerCase
DELL U2713HMT Thermaltake Poseidon Z CoolerMaster Silent Pro M2 1000W Silver Acrylic Benchtable 
MouseMouse PadAudio
Razer Death Adder Blue 3.5G Razer Goliathus Extended [Control] Creative Inspire T3030 
  hide details  
Reply
SkynetX
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Xeon E5-2623 V3 @3GHz running 3.3GHz MSI X99A SLI Krait Edition Inno3D GTX980Ti Ichill Black Hybrid  Gskill Ripjaws V 4x4G DDR4 2400MHz 
Hard DriveOptical DriveCoolingOS
Seagate Barracuda HDD 1TB SATA3 SAMSUNG Bluray Combo SH-B123L Cooler Master Hyper 212X Turbo Windows 10 Pro 64bit 
MonitorKeyboardPowerCase
DELL U2713HMT Thermaltake Poseidon Z CoolerMaster Silent Pro M2 1000W Silver Acrylic Benchtable 
MouseMouse PadAudio
Razer Death Adder Blue 3.5G Razer Goliathus Extended [Control] Creative Inspire T3030 
  hide details  
Reply
post #1440 of 2146
Hey gang,
Finding out some things OC'ing my EVGA GTX 570SC reference design GPU. I use EVGA's oc'ing utilities (Precision and OC scanner), OCCT, and 3DMark 11. I have used MSi's Afterburner, but after using EVGA's stuff, I much prefer it over MSi's offerings. What I have found works is to set up a possible over clock, run OCCT for 20 minutes, followed by OC Scanner for 20 minutes. If there are no errors/artifacting/crashing I fire up a benchmark run on extreme setting in 3DMark 11. If it passes that, then I conclude that my OC is stable. I have found that OCCT/OC Scanner can come back w/ no errors, but the same OC will crash the driver running 3DMark 11, or seize the program. I have conluded that that is because the 3DMark 11 benchmark run is truly EXTREME. And, the graphics it forces your GPU to render are DYNAMIC, more like in-game environments, unlike the static graphics that OCCT and all other such programs create.
IMHO you sould NEVER run a GPU stress test w/ your GPU OC'd for more than 30 minutes! If you run it over night, or 12 hours or whatever, you are asking to brick your graphics card. If a stress test, or multiple stress tests, come back error free after 20-30 minutes, I think you can declare your overclock "stable." If it artifacts when running a game after that, no biggy; just lower your clock/raise your voltage/combo of both. You all realize that OC'ing a GPU is NOT like OC'ing a processor. If you have an unstable proc you risk writing errors it reports to your HDD, and if it gets bad enough you'll have to erase your HDD and start ALL over. Graphics OC'ing is not that way.
Anyway, just some tips from my experience.

EDIT: I have not tried Unigine yet. Look forward to seeing what that is like.
Edited by croSSeduP - 12/5/11 at 1:27pm
Mr. Blue
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 2600K; 4.6GHz @ 1.385V MSI Z68A-GD65 B3 EVGA GeForce GTX 680 FTW MSi GTX 680 
RAMHard DriveHard DriveHard Drive
G. Skill Sniper 4GBx4 DDR3 800MHz Crucial M4 SSD WD 1600AAJS Caviar SE WD 32000AAKS Caviar Blue, 16MB Cache 
CoolingOSMonitorKeyboard
Corsair H50 CPU Water Cooling, 2 120mm fans in ... W10 ACER HG270HU Logitech G910 Orion Spark 
PowerCaseMouseAudio
Antec High Current Gamer 850W Apevia Mid-tower Logitech G500 Creative A220 2.1 Speaker System; Sennheiser HD... 
Other
ASUS Black Blu-ray Drive SATA  
  hide details  
Reply
Mr. Blue
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 2600K; 4.6GHz @ 1.385V MSI Z68A-GD65 B3 EVGA GeForce GTX 680 FTW MSi GTX 680 
RAMHard DriveHard DriveHard Drive
G. Skill Sniper 4GBx4 DDR3 800MHz Crucial M4 SSD WD 1600AAJS Caviar SE WD 32000AAKS Caviar Blue, 16MB Cache 
CoolingOSMonitorKeyboard
Corsair H50 CPU Water Cooling, 2 120mm fans in ... W10 ACER HG270HU Logitech G910 Orion Spark 
PowerCaseMouseAudio
Antec High Current Gamer 850W Apevia Mid-tower Logitech G500 Creative A220 2.1 Speaker System; Sennheiser HD... 
Other
ASUS Black Blu-ray Drive SATA  
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA

Gear mentioned in this thread:

Overclock.net › Forums › Graphics Cards › NVIDIA › GTX 570 overclocking thread