CPU degrade over time?

wuyanxu

wuyanxu

wuyanxu

Associate
Joined
11 Jul 2007
Posts
691
Location
Southampton
this is weird. i got 2 hour prime stable when i first got the computer with 1.395 set in Abit iP35 Pro BIOS.

but after less than 2 weeks, i keep getting a rounding error with the exact same settings. NOTHING has changed.

so i increased Vcore to 1.4050 in the BIOS and it's now priming stable, for 1 hour 30mins now.

is this a sign of normal Overclocking or something more serious?
 
not in 2 weeks it shouldnt... maybe a v droop

v core will be fine at that if its stable and not too hot..
 
i sure hope it's only a Vdroop.

just to be safe, im running at 345x9 now, 3.1Ghz, with 1.365 set in the BIOS, the voltage now ever goes above 1.35 - Intel's reference voltage on the q6600 box.

if i never go above 1.35, does this mean the CPU will work as long as it's intended?
 
Who knows, each chip is diffrent.
Even with the same board, memory, PSU graphics card CPU etc u can get diffrent speeds on overclocks.

So each chip will behave diffrently, to be fair how long are u going to keep ** chip ?
i've never used a Computer for more than 2 year, before selling it to a mate and upgrading !
 
even 1.4 is not enough to degrade the CPU significantly during its normal life span - intel quote upto 1.3625 with absolutely no side effects expected and upto 1.5 with minimal impact.

If its like most of the other P35 boards your going to be drooping by about 0.0725v idle and can increase to as much as 0.1 under heavy orthos or other prime load...

Use everest to check the actual vcore value while its under heavy load - I found I needed 1.4625 (BIOS) to ensure stability at 3.4gig.
 
oh, i thought i ran mine at 1.4 was high.

may be i need a better cooler...... im getting temperatures as high as 70 under Prime, that's with 1.4v in BIOS, 1.35-36 in Everest while under load.

so let me get this straight: when i set 1.4 in BIOS, and im getting 1.37 while ideal, but not much heat, that won't damage the processor. while prime-ing 70c, 1.35v also won't damage the processor?

i've read somewhere that temperature change (deltaT) is what kills most processors. so a good cooler that doesn't make the processor heat up as much is better than worrying about voltage?
 
Within reason, heat is the killer. Personally I dont like mine to go above 65*C under full load. Even though I think the chips can operate fine up to about 75*C, but unfortunately I dont have the time of a CPU's estimated lifespan spare to test this out :p.

Voltage is directly proportional to the temperatures, and on air cooling and I reckon most watercooling, temperature limits will be exceeded before you get chance to put a stupid amount of voltage through it.
 
true, so i will need to try to not reach 70 on my G0 and everything will be happily overclocked.

the problem is, only Prime can push the CPU higher than 65 on my computer, Crysis/Bioshock any thing i use can't even get the CPU past 55c. so can i safely use it without fearing it over heats?
 
This is why Prime is the perfect stability test, as it really implements a worst case scenario. I think most games wont even produce around ~75% of the heat of a load testing in Prime, as all the cores simply aren't 100% utilized all the time. Maybe if you were encoding video you could touch Prime load temps.

As for a safe temperature, Im sure 70 will be ok under full load in prime. At the end of the day, there hasnt been enough time lapsed yet to test the chips lifespans, and any resulting degradation which may occur when constantly strained with high temps and high volts. Even if there was, theres still the scenario that not one chip is really the same, and one could perform better/worse than the next "identical" one.
 
wuyanxu said:
this is weird. i got 2 hour prime stable when i first got the computer with 1.395 set in Abit iP35 Pro BIOS.

but after less than 2 weeks, i keep getting a rounding error with the exact same settings. NOTHING has changed.

so i increased Vcore to 1.4050 in the BIOS and it's now priming stable, for 1 hour 30mins now.

is this a sign of normal Overclocking or something more serious?
Click to expand...


You got lucky with the 2hrs imo, probably would've errored out sometime later, test for 24hrs imo as i've seen prime fail even after 15hrs etc.. 2 hours is not realistically long enough for a proper gauge.
 
a lot of games will only run 2 of yours cores. So you will never reach the temperatures produced by the 4 cores. If you did a stress test on just 2 of the cores you would see the temperature your cpu usualy runs at.

Though i got scorned for trying to point out the fact that a lot of people are stress testing their quad cores on all 4 when they didn't need too. All these people aruging it only uses 2 cores. Then have a go at me for saying why not just prime it on 2 cores lol. Of corse dont make it so it hits 90degress on all 4. But like said above youll never get more then 75% of what prime can throw out anyway.
 
8igdave said:
Then have a go at me for saying why not just prime it on 2 cores lol. Of corse dont make it so it hits 90degress on all 4. But like said above youll never get more then 75% of what prime can throw out anyway.
Click to expand...


Heat is not the only reason for stress testing, you wanna know if your 4 cores are stable as well (vcore, board electronics etc.. 2 cores might pass but all 4 loaded might cause enough to throw off all of them) and one day you might be using all at full bore and it craps out... Theres a whole lot more out there than just games. I regurlarly use all 4 cores when encoding multiple files etc.. just do not see your logic. Testing 2 cores at a time will also not provide a good gauge as to how hot it can get and if your cores will have the same stability at that temp as it had when 2 cores were tested only.
 
Justintime said:
You got lucky with the 2hrs imo, probably would've errored out sometime later, test for 24hrs imo as i've seen prime fail even after 15hrs etc.. 2 hours is not realistically long enough for a proper gauge.
Click to expand...
thanks. okay, i shall re-apply AS5 (when installing, found i put down my cooler in the wrong position, so i had to lift it up, reposition, it might affected the way paste spreaded), and then to for 400x8, 3.2Ghz.

what i don't understand is how can the Quad core database have such low voltage, even on 3.6? they have 1.4v?!? i can't get it 3.2Ghz stable at 1.395v.
 
You must log in or register to reply here.
Back
Top Bottom