It was running at 4,5GHz? Whoaa! I tought it was around 4,2GHz.
No wonder it crashed so often. I would have only clocked an extremely well ventilated gaming machine that high and that only because I shut my machine down for the nights, so it's not at continuos burn. But a server... A bit too extreme.
And over time the dust collected into the CPU and other fans really will have temperature rising to the levels where BSOD will happen. It has happened to me couple of times even with a more modest overclock.
Do you majkp de-dust it at regular intervals?
Well. I'm glad you guys found out. And it's very good that it was caused by an OC because you can control that whereas if it would have been program/script/software problem it might not have ever been figured out.
Don't get confused by the ugly look of the server rack and the case server itself is mounted in, it's a perfectly working cooling machine ..:)
First of all the rack is located in a basement which is not air-conditioned but temperatures rarely reach over 20C. Server case is open (= best for cooling) and the server rack has 2 large holes one sucks cold air at the bottom and the other one let's the warm air out on top so it works like a chimney.
Second - the Zalman CPU cooler is one of best air-cooling solutions on the market. The CPU even @4,5GHz maxed at 55C after an hour of hardcore stress testing.
Third - the CPU is a perfect overclocker (it ran even at 4,7GHz), but I have to agree that there's a major difference between a rig on your table and 24/7 game server. I raised it from 4,2GHz up to 4,5GHz because I finally replaced the temporary memory sticks with the 1150MHz ones and I still thought we can run stable on higher frequency. 2 hour stress testing (which I have done during last name changes) was okay, but nothing can make you sure better than real-time load caused by real applications. So there we are, running stable (I hope) at 4,05 GHz.
But remember, there may still be crashes caused by bugs, cheaters and corrupted files, we can never 100% prevent all the crashes, sad but true.
Nope, wouldn't help, flserver.exe itself is not written for multithreading, which is a shame as it only uses 1 core/cpu leaving the other ones for supporting processes. We use a patch by wodka that forces flhook to run in separate thread leaving the other CPU solely for flserver. Unless source codes are released there is nothing to do about it, and I doubt that even if they were released no one would be able dedicate massive amounts of his/her free time to rewrite it...
a genenral rule of thumb cooler something is longer it runs less cooling more heat more heat faster wear downs in theory a cpu will never fry the theory being it can work in space or a constanyly heat free envroment would have to also be hermeticly sealed so it cant clog unless its liquid cooled also avoid codeing at all costs if you can cuz then you riun into all sorts of problems that you may think you know how to fix but cant(its called hard codeing for a reason and i know the concepct but i would never have the patince to code)..anyway thanks again for fixing the problem
(If you find any mistake in my English, please let me know via a PM)
(Really, I speak terrible English, so please, tell me if I make mistakes. I'd like to improve it a bit )
I think two points were that if it's running cold, and if it is sealed against dust etc. it will run forever and ever without burning out.
As a side note, a CPU can become too cold if you get inventive with your cooling options, and that will slow down/create instablility as things that should be connected by touch get deconnected when the cold make them contract.
John Johnson - Master of Synth.Foods-Convoy|049
Hans Adler - Synth Foods escort wing
You are mistaken.
The colder a CPU runs, the better, as conductivity improves as temperature decreases.
In order for you to damage it out of coldness you should cool it enough so it goes into a state of matter called the Bose-Einstein condensate. And that is measured in mK (mili-Kelvins). Not achievable outside a lab. That, or to be stupid enough to use a compressor, and let it condensate water around the CPU block, which will lead to a short circuit.
(If you find any mistake in my English, please let me know via a PM)
(Really, I speak terrible English, so please, tell me if I make mistakes. I'd like to improve it a bit )