Advanced search

Message boards : Graphics cards (GPUs) : "Thermal cycle"

Author Message
Erich56
Send message
Joined: 1 Jan 15
Posts: 1090
Credit: 6,603,906,926
RAC: 21,893,126
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwat
Message 48909 - Posted: 12 Feb 2018 | 11:49:00 UTC

If I remember correctly, what is called "thermal cycle" are rising and falling GPU temperatures, due to intervals between GPU tasks (or other reasons).
Easy to guess that if this happens too often, it shortens the life of the card.

I have a special situation now with my host containing two GTX980Ti:
Besides GPUGRID, I am running several LHC tasks. Whenever such a task is starting and building up the VM file during it's initialisation phase, the system stalls for about 15-20 seconds, which means there is also a standstill of the GPUGRID tasks (but never ever one failed due to this), causing the GPU temps going down by about 5°C, from 62° or 63°C to 57°C or 58°C. Right after the processes are resumed, the temps climb up the their original values within seconds.
Depending on the duration of a given LHC tasks, and the number of cocurrently running ones, this happens between roughly 5 and 10 time per hour.

Question for the experts: would this behaviour already fall under "detrimental" thermal cycle, or is it (nearly) negligible.

One thing that I noticed is: this happens only on Windows XP, NOT on Windows 10. For what reason ever.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 48910 - Posted: 12 Feb 2018 | 12:30:35 UTC - in response to Message 48909.

If I remember correctly, what is called "thermal cycle" are rising and falling GPU temperatures, due to intervals between GPU tasks (or other reasons).
Easy to guess that if this happens too often, it shortens the life of the card.

I have a special situation now with my host containing two GTX980Ti:
Besides GPUGRID, I am running several LHC tasks. Whenever such a task is starting and building up the VM file during it's initialisation phase, the system stalls for about 15-20 seconds, which means there is also a standstill of the GPUGRID tasks (but never ever one failed due to this), causing the GPU temps going down by about 5°C, from 62° or 63°C to 57°C or 58°C. Right after the processes are resumed, the temps climb up the their original values within seconds.
Depending on the duration of a given LHC tasks, and the number of cocurrently running ones, this happens between roughly 5 and 10 time per hour.

Question for the experts: would this behaviour already fall under "detrimental" thermal cycle, or is it (nearly) negligible.

One thing that I noticed is: this happens only on Windows XP, NOT on Windows 10. For what reason ever.


Believe me when I say your card will be a technological brick before what you describe will damage your GPU.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 7,520
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 48911 - Posted: 12 Feb 2018 | 12:39:56 UTC - in response to Message 48909.

... causing the GPU temps going down by about 5°C, from 62° or 63°C to 57°C or 58°C.
This is negligible.
The "normal" thermal cycle is going from room temperature (~25°C) to ~80°C. That is 55°C difference.
The "extreme" thermal cycle is going from room temperature to -196°C (Liquid nitrogen cooling), that is 221°C difference.

mmonnin
Send message
Joined: 2 Jul 16
Posts: 332
Credit: 3,772,896,065
RAC: 4,765,302
Level
Arg
Scientific publications
watwatwatwatwat
Message 48912 - Posted: 12 Feb 2018 | 14:31:25 UTC

I agree, negligible. Individual areas of a processor probably vary more than 5c all the time as data moves around or cores crunch and idle. Gaming is probably even more variable than crunching, even your situation. Going to a menu, facing a wall, or idling in a game probably allows part of a GPU to idle vs when there is a lot of action.

Post to thread

Message boards : Graphics cards (GPUs) : "Thermal cycle"

//