Heat vs. Performance

[STEvil]

Anyone ever had heat cause a slowdown without actually influencing stability on their T-Bred or Barton?

I've been having performance issues with my 2500+ for a while.. 195fps drago low/high and 74fps nature for example where I used to push 350/200+ and 124+ respectively at everyday settings.

I played with drivers and such, 1t/2t etc to try and figure it out, but nothing would budge. Finally I cranked on the clock speed and voltage with the crappy arkua cooler (dr. thermal rip-off) i'm using on here and was running 75c @ 2.4ghz 1.8v according to the CPU diode as read by MBM5/a7n8x bios. load.

2.3ghz 1.7v was 61c load.

That was unacceptable ;)

So I pulled the heatsink and there was a peice of solder stuck to the heatsink that had been directly between the heatsink and the core! wtf?? I'm not sure how or when it got there as i'm very particular about cleanliness when doing heatsinks.. but thats rediculous, lol :D

Anyways, after having removed that and re-lapping the heatsink i'm now doing 2.4 @ 1.775v 61c load (down to 47c load now, I think the TIM is setting)... and all my mysteriously missing performance is back too :D

Odd that it was never unstable (if I stayed at around 2300mhz) and only things that needed high bandwidth suffered though, isnt it?


As a side note, the TIM around the solder was burned brown.. and this stuff is rated way past 175c as far as I know.

http://www.wcroller.com/los/linkpics/unitol.jpg :slobber:

[Kanati]

Hmm, I've never seen one lose performance at high temps, but I think you might be lucky that you didn't damage your core. With a piece of solder in between the heatsink and core, the pressure was likely focused on a very small area.

[STEvil]

it covered about 1/8th of the core spread out.. was nice and flat due to the temps I guess. Still had a fairly large lip around it though and you can see where it wasnt touching the core at all.

Tough chips these things.

[bachus_anonym]

that's one weird story, STEvil :stick: i haven't had an issue like that or at least i wasn't that particular and observant as you... but i want to say that you're one lucky guy that nothing actually happened to the chip :)
on a side note, get a better cooler, man... with my Tbred B 2.5GHz, SLK-900A and 120mm QUIET fan i get like max 55C. buy that's at almost 2.1v and it's not that cold here either.
and as we are at buying stuff, get some socks too :p: :D :lol:

[STEvil]

my psu gives out over 2460mhz-ish.. i think. used to have an slk-900u on it and could bench to 2500, but the slk-900u is being refitted to work with socket 754 and 939 boards now. Board also isnt stable over 202fsb so 12.5x202 is absolute max unless I mod it for higher multipliers. Too lazy ;)

2500 1.825v was 48c load or so iirc.

http://www.wcroller.com/los/linkpics/slk10k.jpg

Antec 400w smartpower, btw. Have an allied 450w but its on the a64 until my 600w OCZ Powerstream arrives.

[reject]

i dont get the pic? do esso make your TIM?

[STEvil]

its not a TIM, its a lubricant. Works great though and withstands temps from -50 to +???. 400g is cheap, too.

[Entity_Razer]

whats a TIM?
:s
Sorry to sound noobish but I have NOOOOOO idea
:stick:

[STEvil]

Thermal Interface Material.

Also known as thermal grease/paste

[Entity_Razer]

Ah ok, so like Arctic silver and such.... :)
thx

[spaceman]

Hmm, I wonder how solder evenly applied would work as a TIM? :D Mostly joking, but also have gears grinding in the back of my head, lol. ;)

[Jethro]

Ive had artic silver make what appear to be solder on my cpu's often. I think its something to do with the silver content and the high pressure the heatsink applies to cpu. Concentrated heat probably contributes as well.

Im not such a big fan of the product as a result specially for aircooling anyhow. Im had thought about getting some Nanotherm PCM to try however with my very cold water cooling i dont get the "solder" effect so not to worried anymore myself but it may be an option for you Stevil.