Leviathan.Phenomena said:
»Phoenix.Gaiarorshack said:
»Leviathan.Phenomena said:
»so I OC'd my evga gtx 670 ftw with +50 core and +700 memory in the precision X OC. seems to be stable. but I had my cpu i5 3570k at 4.5 with temps at 75-80 at max load. but dropped it down to 4.2 which is 60-65. should i increase the OC on my cpu? or just keep it at a nice 4.2
is it 100% stable at 4.5 then keep it there.
im not talking about passing a few simple furmark runs but under real stress
well I ran prime95 heat stress test using all cores at 100% for 30 mins. it never failed. I just didnt like the high temps of 75-80 even though thats acceptable. didnt want to drain the life of cpu that much faster. lol guess ill go back in and redo it. im sure even a 4.7 would work... come to think of it i think 4.7 was the 75-80 temps. ill have to redo it. also volts on 4.7 were like 1.345 and i thought that was a bit much.
30 mins is not a stress test thats quick test
unless you only play for like 30 mins of a time
a proper done stress testing will tell you much more about the cpu and its life time than temperatures
Ive been running intel cpu's so hot the internal temp diode could not feedback the temperature (tjunktion delta went below 0)
the system ran stable for 16 hours stress prime
8 hours of intel burn test
8 hours of 7-zip compression
6 hours of memtest86+
you have to realise if you are on a intel cpu. you dont maasure the temperature by software.
the actual feedback from the cpu diode is a delta value up to max recommend temperature for the cpu
e.g. if you cpu is 80 and the max is 90 the cpu temp diode reports a value of 10
the software then subtract this value from what it believes the max is to be (the cpu dont tell) so the temperature you see is very depending on the software guessing the right cpu max temp.
this one of the reason you should trust in proper testing rather than cpu temperauter.
i realised a little stresspackage a couple of years ago because to many ppl used futuremark as a stable test wich it is pretty much NOT. Yhe heat output and power usage was alot bigger (sometimes more than 50% more power draw from the wall) compared to futurmark.
i will see if i can find it again.