• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Does ability to overclock degrade over time?

A

awdark

[H]ard|Gawd
2FA
Joined
Mar 22, 2003
Messages
1,633
I know its the summer, but my computer gets hot much faster than before.

My cpu is an Opteron 146 overclocked from 2ghz to 2.74ghz

In the past the highest I would see the chip go was maybe 150F and recently it would go to 170F and beyond. Not really sure about how high it gets since it usually crashes while I turn the monitor off to encode or when I am playing a game.

I do crank my fans up all the way but it just doesn't seem to be able to cool it off as well anymore.

The only change to my setup was a 4 channel sunbeam fan controller... previously my computer would overheat because I forget to crank up the fans but now even with the fans on max it doesn't manage to maintain the temperature. Yeah, the hsf is clean and I have tried it without the fanbus with same results.

SOO the question is am I just nuts or do processors eventually spit out more heat or something?
 
I have a 1.6A @ 2.68Ghz that is running as fine now as it did the day I first oc'd it (the day I bought it,back in mid 2002)
 
Global warming!!

:p

No but seriously, has anything in your environment changed? If you do have AC of any kind, is it possibly not operating as efficiently? I know you said the setup hasn't changed but did you add more drives or any other source of heat? If the OC was borderline to begin with (having to turn up the fans under load to avoid a crash would indicate so) then even the slightest shift in ambient conditions, or even usage patterns, could be making it a less stable OC.

Generally speaking I'd think a lot of chips would OC slightly better after a bit of time (paste settles, burn-in period, etc), but ehh... Everything's possible I guess.
 
Hrm I guess I have to blame the environment for being warmer then.

In the past the limit was not my CPU, but my motherboard for having a 274HTT cap. Heat was just never an issue in the past and now I almost feel the need to get a scythe ninja or something huge to keep it cool. I guess its a good thing that silicon wont degrade THIS quickly then. Time to visit buy/sell subforum!
 
Haha dust bunnies... I clean out the insides at least once a month... but usually every other week or so. :eek: evil dust off company!
 
And humans never landed on the moon and Elvis is still alive right?
 
Some motherboards have cheaper capacitors, which tend to lose their charge more noticeably over time, especially if pushed.
 
Max clock speeds follow a curve that rises a little from new as the CPU beds in and then drops as the silicon degrades.
How degraded the silicon is, the applied voltage and temperature determine its max speed for a given CPU.

The operating temperature will determine the rate at which the silicon degrades.
For approx every 10C increase in temp the silicon will degrade at twice the rate.
This rule applies up to the max rated temp of the silicon. Higher temps may result in increased degrading and other modes of failure (to those below) are possible too.
Not all silicon circuits are created equal, some will have years of life at normal operating temp, others much less.

If you have let your CPU overheat ever, it could cause enough degrading of the silicon to reduce your maximum overclock.

This is through increased current leakage which causes higher temps and reduces available current for the circuits a little too. As higher current is flowing, transients on the voltage rails will cause larger voltage drops, increasing noise on the voltage rail which can reduce stability and hence lower your overclock.
The extra heat will also increase electrical noise within the silicon which can affect stability.
Too large a leakage can reduce the line voltage and voltage rise times preventing logic signals from being transmitted in time.
This is one mode of chip failure.


Luckily because of thermal protection, a bad heatsink contact wont kill a cpu these days but it could still age a lot faster with a bad contact.
Letting your PC run for extended periods of time with little case airflow will age all the silicon devices faster. Its not good for the PSU or harddrives either!

If you let your CPU run hot or overheat severely for a short time, it can cause permanent degrading of the silicon.
 
I just spotted the temps you are getting!

Those temps are too high.
The silicon is rated to around 70C, you should try to remain quite a bit below.
the old rule of thumb was to run at 55C max to allow for temperature drops from the silicon to where the heat sensor was located.

As the sensors are mostly on the silicon die these days, you probably wont need to be quite as stringent.
(You must make sure the sensor you are reading is an on die one though!)
Do not trust the reported die temperature as being especially accurate, as with all electronics, the results could vary quite a bit.
So to this end I wouldnt go within 10C of the max rated temp to be safe.
This gives a max cpu temp of approx 60C (140F).

However, your 170F temp is 76C, above the rated temp for your CPU regardless of method.
 
awdark: I did some testing recently and the ability of a processor to overclock is directly related to the temperature you run them at.

With an E6400 Core 2 Duo I found that I could run Orthos stable up to a core temperature of 80C and a little beyond. For every additional 100 MHz I overclock I need to reduce the maximum core temperature by approximately 5C to remain Orthos stable. To go up 4 steps from 3200 MHz to 3600 MHz I need to reduce the maximum core temperature 4 steps from 80C to 60C.
80C - ( 4 X 5C ) = 60C

Your mileage will vary but there is a direct relationship when you are overclocking your processor near its maximum and 2.74 GHz for an Opteron is at or close to that point with an air cooler.

My room temperature is up 10C today compared to during the winter. To be Orthos stable today I would have to either turn up my CPU fan and create some noise or turn down my overclock. My solution is to stick to gaming which works fine today at 85F even at my maximum 24/7 overclock of 3600 MHz. Most games are still single threaded and don't heat up my CPU anything like Orthos does.
 
RMClock gives a different reading from Speedfan.
At full load, Speedfan gives more conservative 150s while the one in RMclock give freakishly high temperatures like 70C.

I do know the reported temperatures are relative (compared to other temperature readings) but when I idle im usually at a happy 90F usually and when I give it load, it just starts flying up.

And yes, the heat does bother me a lot. I never seen temperatures go up like that except when I forgot to turn plug in my fan. :rolleyes: yeah im somewhat destructive.

Its just strange to experience heat issues like this now when I didnt have problems before. Its understandable that when you overclock you get higher temps and often new coolers are desired but mine worked well enough before, I use a Zalman 7000cu.

Okay guess today is a bit warmer
start.jpg

Startup, within 3 minutes of window sound

5min.jpg

5 minutes in, the downward spike is likely caused from cool and quiet.

10min.jpg

After 10 minutes

Was going to go for a 15 minute shot but I got a blue screen. Trace something something stop error.
Very strange, I usually don't get the BSOD.

Anyways, I was using a funky FLV file that kept crashing my computer so I use that as my load test lol.

After restart
restart.jpg

(reference picture to show overall temps)

O_O I guess my room IS too hot since everythign is quite warm.
 
I didnt answer your question directly so heres a condensed version of the above that answers the question:
"do processors eventually spit out more heat or something?"

CPUs degrade over time and draw more power.
This causes them to run hotter the older they are.
If you have run the chip very hot, it will degrade faster and is more likely to show a marked increase in temperature.
 
Thanks for the tip!
coretemp.jpg

Now the temps look far worse than before because the psycho rmclock one does seem accurate.

:( Oh well, last resort... time to lower my overclock.
 
What type of heatsink and fan are you using? A fan speed of 1339 rpm doesn't look like enough when overclocking an AMD in the summer with your present heatsink. If it's as hot there as it is here then you're probably going to have to back it down to default MHz to be Orthos stable with your present cooling solution.
 
Zalman 7000cu http://www.silentpcreview.com/article92-page1.html

o_O looking at the review the max should be 2400rpm but my fan controller IS at max. This does explain a lot though. Let me try a different channel then a direct connect.

EDIT
Okay, I had connected my power supply fan to that for the 1336 rpm spin... my fanbus didnt come with a splitter so I opted to connect the one from my PSU to the mobo so it wont give me the no cpu fan error.

I connected my fan to the board directly to get the speed reading and its spinning at 2766rpm.
My fanbus outputs a maximum of 11.49 volts, so it doesn't give it the full 12vdc but I say thats more than enough for the fan.
 
piako said:
Global warming is a hoax if you say human activity causes it. Anyway OP check your thermal paste. You may need to reapply some and dust off your hsf. ;)
Click to expand...

definately a bunch of lefty BS. Manmade global warming is sadly an unbelievable display of the ignorance and guillibility of the population.

Its either a) dust. b) power supply capacitor falloff c) electromigration
 
now now, let's not soapbox this thread :p

there are two things that matter with temps here:
1. colder = higher stable speed at a given voltage
2. colder = longer lifespan... approximately 2x longer for ever 10c dropped afaik.

there is no good reason for your temps to be that high.. even on the stock heatsink :confused:
 
My opteron 170 is no longer stable at 2.8ghz after over a year of use. I had to back down to 2.6ghz. Not sure if it was overheating or what but I've replaced all my system fans and will give 2.8ghz another go. I'm thinking my chipset might have been overheating as well so I replaced it's fan as well. Just really haven't had the time and incentive to test out 2.8ghz again, especially since one of my goals with the fan replacement was to allow silent operation during light tasks like watching movies
 
^That sounds just like me then :( so ill have to back down my OC I guess. Why it shoots that high I have no idea. Short of taking off the heatsink to apply thermal goop, I have made sure the screws are tight and the stuff is spinning.

I guess this is the long term result of the Opterown thread... :eek:
 
I've found more memory modules that degrade over time than processors.

awdark: If you're running your AMD processor with a core temperature up into the high 60C range then you shouldn't expect it to be stable. If it was once stable at that MHz and temp that's great but you shouldn't be too surprised if it's not now. Did you ever do any proper testing on this overclock when you originally did it and get some core temperatures back then while running Orthos so you have something to compare to now?
 
i am running an old ass skt 478 Celeron 2.0A CPU on that same Asus P4GE-V mobo with the same crappy PC2100 ram and running it as 2.8Ghz, 140Mhz FSB, with the same crappy HSF it was running way back when (although i did clean and reapply Ceramique about 3 years ago), but still.....runs 2.8Ghz all day long

did i mention it has been shipped across country twice and been through two owners and was used when i got it.....!!!
 
Gheez the unit conversions is driving me nuts haha

In the past it seriously ran much cooler. Like the highest it would go would be 60c even after several hours of gaming or video encoding.

:confused: I always thought 60C for full load was reasonable and 30s to 40s for idle times.
 
60C is the very high end of tolerable temperatures and is not ok.
I wont run my CPU at that temp.

Being reasonable, 60C isnt the number you should be interested in, the 76C is what did it.
That 76C is the reported temp, it could even be a lot higher than that.
You should always leave a safety margin in case of ambient temp increase, dust accumulation, blocked ventialtion, uncalibrated sensors, faulty sensor reading software, reading the wrong sensor etc.....
Your CPU has run at way over the rated temp so it has likely suffered.
 
As a general rule, the first thing you should do if you find your overclock degrade over time is to take out the heatsink, clean it, clean the cpu, and re-apply whatever thermal paste you use and put it all back together.

All thermal pastes degrade over time (or more accurately speaking; the TIM joint degrades over time). Arctic Silver is definitely the fastest to do so that I've tried personally.


Also, every cpu I've ever owned and seriously overclocked has permanently degraded over time and I've always watercooled these chips with a damn fine setup (see sig). My Athlon XP chips were the worst, but then I thought nothing of running over 2 volts through them. My opteron 146 at first was happy at just over 3.0ghz, but then some months later I had to knock it down to 2.97ghz. Only lost 30-40mhz, but it was annoying 'cos I fell out of the 3ghz club. :p
 
Are you sure your heatsink is in contact with the chip and that the fan is actually spinning? (visual check) It doesn't look like it from your temp charts.

There's something seriously wrong, either way. Good luck fixing it.
 
I've had my memory overclocked from day one at 3.2V and 1T command rate. It's still going strong.

The damn left-wing hippie liberals try to scare you with global warming. Don't believe everything the media tells you. They're bullshit lies. :mad:
 
You must log in or register to reply here.
Back
Top