Overclock.net › Forums › Graphics Cards › Graphics Cards - General › Does stable at 60c, but not 80c (even if I can keep it colder) = stable ??
New Posts  All Forums:Forum Nav:

Does stable at 60c, but not 80c (even if I can keep it colder) = stable ??

post #1 of 5
Thread Starter 
Say im using OCCT GPU or Furmark which everyone can really agree will put alot more stress on your graphics card than just an average gaming session....

In OCCT...

100% load @ 65C w/ 50% fan... stable and no errors for 60+ minutes

BUT!

If i drop it to 35% fan... I can get my load temp to sit at 79/80C and i will have 10-20 errors crop up over 10-15 minutes...



Obviously I will never see temps this high when gaming so I will call this stable and make it my 24/7 clock, but! in the long run... is this considered 100% stable?

Should a 100% stable GPU be technically able to run without errors up to its thermal limit? or does cooling capability and max relative temps depending on ambient temps all factor in on what is truly a stable overclock ?

Obviously I can push my overclock higher on water... but am i really just hiding the limitations of my chip by finding better cooling or am i actually overcoming the limitations and kind of "changing the game" in a sense?

I would be surprised to see if every single stock clock and stock cooled GPU (say a gtx 460 or something) would all be able to run OCCT error-less at say 90C, 80C, 70C, and so on...

I would think that many stock cards would show errors PRIOR to their listed mfg thermal limit....

Edited by rmp459 - 1/3/11 at 10:08am
    
CPUMotherboardGraphicsRAM
5960x @ 4.4ghz (1.19v) evga x99 micro2 (2) evga gtx 980 ti hybrid 32gb hyperx 2666mhz 
Hard DriveCoolingOSMonitor
samsung 950 pro 512gb NVMe M.2 ssd ek everything win 10 xl2430t / u2515h (x2) / u2913wm 
KeyboardPowerCaseMouse
cm novatouch w/ pbt caps. evga 1200 p2 lian li pc-9f nixeus revel 
Mouse PadAudioOther
steelseries qck heavy gustard U12/X12 -> H10/Lyr presonus 22vsl > mxl770 (mic) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
5960x @ 4.4ghz (1.19v) evga x99 micro2 (2) evga gtx 980 ti hybrid 32gb hyperx 2666mhz 
Hard DriveCoolingOSMonitor
samsung 950 pro 512gb NVMe M.2 ssd ek everything win 10 xl2430t / u2515h (x2) / u2913wm 
KeyboardPowerCaseMouse
cm novatouch w/ pbt caps. evga 1200 p2 lian li pc-9f nixeus revel 
Mouse PadAudioOther
steelseries qck heavy gustard U12/X12 -> H10/Lyr presonus 22vsl > mxl770 (mic) 
  hide details  
Reply
post #2 of 5
Quote:
Originally Posted by rmp459 View Post
Say im using OCCT GPU or Furmark which everyone can really agree will put alot more stress on your graphics card than just an average gaming session....

In OCCT...

100% load @ 65C w/ 50% fan... stable and no errors for 60+ minutes

BUT!

If i drop it to 35% fan... I can get my load temp to sit at 79/80C and i will have 10-20 errors crop up over 10-15 minutes...



Obviously I will never see temps this high when gaming so I will call this stable and make it my 24/7 clock, but! in the long run... is this considered 100% stable?

Should a 100% stable GPU be technically able to run without errors up to its thermal limit? or does cooling capability and max relative temps depending on ambient temps all factor in on what is truly a stable overclock ?

Obviously I can push my overclock higher on water... but am i really just hiding the limitations of my chip by finding better cooling or am i actually overcoming the limitations and kind of "changing the game" in a sense?

I would be surprised to see if every single stock clock and stock cooled GPU (say a gtx 460 or something) would all be able to run OCCT error-less at say 90C, 80C, 70C, and so on...

I would think that many stock cards would show errors PRIOR to their listed mfg thermal limit....
No, temperature can actually change whether an Overclock is stable or not, that is why on AIR I would just run the fan at max to get my overclock and then turn it back to auto after.

As you said OCCT and Furmark are not going to give you real world temperatures. Unfortunately they also do not stress your GPU in the same way a game does and can often mislead you into thinking your OC is stable, but then you start gaming and your system starts crashing.

What you should use to test the stability of your system is a benchmark.

Personally, I have had a lot of success using the Metro 2033 Benchmark on very high settings. If your system can pass 15 runs @ very high then it will be stable, at least IMO.
Mixed Bag
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 6950X 4600Mhz ASUS Rampage V Edition 10 1080 Ti FE HyperX 12-13-13 @2400Mhz 
Hard DriveCoolingOSMonitor
Samsung 850 EVO Corsair H100i GTX Windows 10 x64 ASUS ROG PG279Q Black 27" 
KeyboardPowerCaseMouse
Corsair K95 RGB EVGA SuperNOVA 1000 P2 Corsair Air 540 Corsair RGB 
Mouse Pad
Some naked Anime with a plump butt 
  hide details  
Reply
Mixed Bag
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 6950X 4600Mhz ASUS Rampage V Edition 10 1080 Ti FE HyperX 12-13-13 @2400Mhz 
Hard DriveCoolingOSMonitor
Samsung 850 EVO Corsair H100i GTX Windows 10 x64 ASUS ROG PG279Q Black 27" 
KeyboardPowerCaseMouse
Corsair K95 RGB EVGA SuperNOVA 1000 P2 Corsair Air 540 Corsair RGB 
Mouse Pad
Some naked Anime with a plump butt 
  hide details  
Reply
post #3 of 5
Thread Starter 
Quote:
Originally Posted by tconroy135 View Post
Personally, I have had a lot of success using the Metro 2033 Benchmark on very high settings. If your system can pass 15 runs @ very high then it will be stable, at least IMO.
By pass you mean not crash and not artifact like crazy in game ?

Im gonna have to develop my own little routine :\\

I've kinda just always gone overboard w/ CPU cooling and never really pushed one to the max.... GPU behavior is a tad different imo and i've never really bothered to OC it myself... sorta playing around... thanks for the help
Edited by rmp459 - 1/3/11 at 11:46am
    
CPUMotherboardGraphicsRAM
5960x @ 4.4ghz (1.19v) evga x99 micro2 (2) evga gtx 980 ti hybrid 32gb hyperx 2666mhz 
Hard DriveCoolingOSMonitor
samsung 950 pro 512gb NVMe M.2 ssd ek everything win 10 xl2430t / u2515h (x2) / u2913wm 
KeyboardPowerCaseMouse
cm novatouch w/ pbt caps. evga 1200 p2 lian li pc-9f nixeus revel 
Mouse PadAudioOther
steelseries qck heavy gustard U12/X12 -> H10/Lyr presonus 22vsl > mxl770 (mic) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
5960x @ 4.4ghz (1.19v) evga x99 micro2 (2) evga gtx 980 ti hybrid 32gb hyperx 2666mhz 
Hard DriveCoolingOSMonitor
samsung 950 pro 512gb NVMe M.2 ssd ek everything win 10 xl2430t / u2515h (x2) / u2913wm 
KeyboardPowerCaseMouse
cm novatouch w/ pbt caps. evga 1200 p2 lian li pc-9f nixeus revel 
Mouse PadAudioOther
steelseries qck heavy gustard U12/X12 -> H10/Lyr presonus 22vsl > mxl770 (mic) 
  hide details  
Reply
post #4 of 5
Quote:
Originally Posted by rmp459 View Post
By pass you mean not crash and not artifact like crazy in game ?

Im gonna have to develop my own little routine :\\

I've kinda just always gone overboard w/ CPU cooling and never really pushed one to the max.... GPU behavior is a tad different imo and i've never really bothered to OC it myself... sorta playing around... thanks for the help
I would expect it to crash if the OC is not stable, but yes also artifacts or anything out of the ordinary.

With my 580s, the Metro 2033 benchmark, caused failure(Driver Crash and system lock up), GPU(not system) shut down, artifacts and some other anomalies I would not necessarily call artifacts until I had stable voltage(1.1v@902Mhz).

Even with SLI setup your GPU will be under 100% load for the majority of the benchmark, but they will also be used more dynamically than in Furmark, which can show you issues furmark doesn't.

Edit: You could go for more than 15 runs ~15minutes, but at that point you will be putting enormous stress on your GPU that you will never see in games.
Mixed Bag
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 6950X 4600Mhz ASUS Rampage V Edition 10 1080 Ti FE HyperX 12-13-13 @2400Mhz 
Hard DriveCoolingOSMonitor
Samsung 850 EVO Corsair H100i GTX Windows 10 x64 ASUS ROG PG279Q Black 27" 
KeyboardPowerCaseMouse
Corsair K95 RGB EVGA SuperNOVA 1000 P2 Corsair Air 540 Corsair RGB 
Mouse Pad
Some naked Anime with a plump butt 
  hide details  
Reply
Mixed Bag
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 6950X 4600Mhz ASUS Rampage V Edition 10 1080 Ti FE HyperX 12-13-13 @2400Mhz 
Hard DriveCoolingOSMonitor
Samsung 850 EVO Corsair H100i GTX Windows 10 x64 ASUS ROG PG279Q Black 27" 
KeyboardPowerCaseMouse
Corsair K95 RGB EVGA SuperNOVA 1000 P2 Corsair Air 540 Corsair RGB 
Mouse Pad
Some naked Anime with a plump butt 
  hide details  
Reply
post #5 of 5
Download Folding@home if you can pass 10-20 WU you are stable. My first OC was stable or so I thought until I got errors folding.
    
CPUMotherboardGraphicsRAM
2500k ASUS P8P67 WS Revolution 2 x Asus GTX470 900/1800/1700 Corsair Vengence 8gb 
Hard DriveOSMonitorPower
120gb ssd, 4 x 1TB F3's Raid 0, 2TB F4 Backup Windows 7 Catleap 27 1440p Enermax Revolution 1050w 
Case
Case Labs M8 with pedestal  
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
2500k ASUS P8P67 WS Revolution 2 x Asus GTX470 900/1800/1700 Corsair Vengence 8gb 
Hard DriveOSMonitorPower
120gb ssd, 4 x 1TB F3's Raid 0, 2TB F4 Backup Windows 7 Catleap 27 1440p Enermax Revolution 1050w 
Case
Case Labs M8 with pedestal  
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Graphics Cards - General
Overclock.net › Forums › Graphics Cards › Graphics Cards - General › Does stable at 60c, but not 80c (even if I can keep it colder) = stable ??