View Full Version : Overclocking debate
Snowknight26
8th November 2008, 07:19
A Q6600 doesn't seem that bad, especially when you can OC it to 3.6GHz+. Encodes things faster than a stock 920.
Split from other thread, started around here: http://forum.doom9.org/showthread.php?p=1210749#post1210749
Chengbin
8th November 2008, 16:57
I can't OC it because I bought it from Dell. They don't allow overclocking on their computer unless it is an ultra high end computer (720, 730). I only have a $hitty XPS 420. My encodings take forever.
burfadel
8th November 2008, 18:17
I'd be very careful on following suggestions of overclocking the Q6600 to 3.6ghz or higher, you can cause damage to the CPU by doing so. Even if you have good cooling I wouldn't recommend going that high. It might be stable at that speed for now, just be prepared to replace it or the motherboard in 6 months time, especially if you do a lot of encoding or other CPU intensive operations. Also keep in mind the computer case design, excellent cooling of the cpu and overclocking usually means more heat expelled in to the case, and with a crappy case this heat can be transferred to other components (motherboard, graphics card, and especially the HDD) which can significantly increase the possibility of failure.
Keep in mind despite what people will tell you and argue, there is NO such thing as a safe worthwhile overclock. A few mhz won't matter, but anything above 10-15 percent you increase the risk significantly of something going wrong. A Q6600 at 3.6Ghz is a 50 percent overclock, sounds good in numbers but is definitely not safe. If you are in a warmer climate, consider the overclock at the maximum temperature your computer is likely to experience, not the current temperature!
It will be interesting on the overclockability and actual performance and price of the AMD Shanghai's soon to come out, they may be a more cost effective solution for those that don't want to go to the expense of a Core i7 with the expensive X58 motherboard and 3xDDR3 ram modules for triple channel memory...
Sagekilla
8th November 2008, 18:57
Overclocking any processor will inherently shorten it's lifetime. It's all a matter of how well you cool your system and whether you can achieve low voltages. Usually, it's the high voltages that kill the processor. If you can keep the processor running at stock voltage, or even lower, you won't experience significantly shorter lifetimes.
But I digress, I'll take the Q6600 off your hands for you ;)
lexor
8th November 2008, 20:07
Overclocking any processor will inherently shorten it's lifetime.
Well AMD chips must have a hell of a lifetime, because my A64 2800 (the oldest/slowest/cheapest A64 ever made) is still going strong since OC done 1 month after its launch. If you don't touch voltages, I think you are pretty safe.
As per Nahelem itself, those latency and such measure don't really tell us how the new patches will affect Nahelem's performance (because we don't really know x264's instruction usage pattern for full encode). Do we have a x264+uncommitted Nahelem patches+common megui profile (blu-ray or one of unrestricted) measurements?
STaRGaZeR
9th November 2008, 14:51
Well AMD chips must have a hell of a lifetime, because my A64 2800 (the oldest/slowest/cheapest A64 ever made) is still going strong since OC done 1 month after its launch. If you don't touch voltages, I think you are pretty safe.
Every processor has a hell of a lifetime.
saint-francis
9th November 2008, 16:36
Q6600 here. Overclocked to 3.6 in the summer and 3.8 in the winter. I've been going like this for a year now. I encode videos with it all the time (maybe 80% of the time it's encoding). I' have had no problems yet. By the time it gets cooked it will be obsolete. I do have pretty good cooling though.
LoRd_MuldeR
9th November 2008, 16:41
Q6600 here. Overclocked to 3.6 in the summer and 3.8 in the winter. I've been going like this for a year now. I encode videos with it all the time (maybe 80% of the time it's encoding). I' have had no problems yet. By the time it gets cooked it will be obsolete. I do have pretty good cooling though.
As said before: Only because it works fine for you and for various other people, it's not guaranteed to work for everybody. Also if it worked 100% stable for you until today, there still is no guarantee that it won't explode tomorrow! So if anybody decides to overclock, he/she should be very well aware of the fact that all guarantees from the manufacturer will be lost immediately...
Just for info:
A friend of mine was running a brand new Q6600 and he turned on "C.I.A.2" (automated overclocking, up to 7%) in the BIOS, just to give it a try. He certainly didn't mess with voltages and he has got one of these "heavy" ZALMAN coolers. Anyways the machine crashed during boot-up and the CPU plus the board were gone for good. He now has a new Q6600 and a new board, but won't use C.I.A.2 anymore, for obvious reasons. I won't either, after I heard that story ^^
ghostonline
9th November 2008, 17:24
your friend is either extremely unlucky or just got a faulty cpu and/or board. people who overclock without knowing the consequence shouldn't overclock in the first place.
afaik, cpu doesn't explode even you dump above 2.0v into that q6600.
i am not sure how you guys determine the board and cpu were gone for good.
plus, under bios default setting, board will automatically drive up the voltage as you increase the cpu frequency unless you change the setting to manual. there is a slight chance that the bios revision on your friend's board is so buggy to the point that it gave way too much voltage for 7% increase in cpu frequency.
LoRd_MuldeR
9th November 2008, 17:44
people who overclock without knowing the consequence shouldn't overclock in the first place.
How can you know that the consequence of using the board's built-in "C.I.A.2" feature is self-destruction?
i am not sure how you guys determine the board and cpu were gone for good.
It doesn't power on anymore, no matter what you do. So you bring it back to the shop and they also can't get it to work.
Hence it's broken. And since you overclocked, there's no chance to get a new one for free...
plus, under bios default setting, board will automatically drive up the voltage as you increase the cpu frequency unless you change the setting to manual. there is a slight chance that the bios revision on your friend's board is so buggy to the point that it gave way too much voltage for 7% increase in cpu frequency.
I got the exactly same board and CPU running rock stable for more than a year now. I just don't mess with overclocking, unless I really need it :rolleyes:
ghostonline
9th November 2008, 18:34
regardless it's cia2 or manual overclock, at end, they are the same thing - overclock. just because they use different terms, doesn't change its nature. if your friend doesn't understand the consequence of overclock, then he/she shouldn't overclock in the first place. as i said, i suspect it is just a faulty board and/or cpu.
you can say your guys got the same model, not exactly same board. no two boards are created equal, same thing applies to cpu. even you guys bought together at same time, you may still get different revision for the board and bios.
as far as i can see, probably just a defective product..
LoRd_MuldeR
10th November 2008, 01:08
regardless it's cia2 or manual overclock, at end, they are the same thing - overclock.
I never doubted that. However one can assume that C.I.A.2 doesn't overclock the CPU in an "aggressive" way, but in an "adaptive" way.
In fact the range that C.I.A.2 uses for overclocking (maximum 5-7% over default clock speed) was considered "safe" by many people here.
All I wanted to make clear was: There is no "safe" in overclocking.
if your friend doesn't understand the consequence of overclock, then he/she shouldn't overclock in the first place.
He knows very well what overclocking means. He just wanted to see how well that C.I.A.2 feature performs. And he saw it :rolleyes:
Atak_Snajpera
10th November 2008, 01:45
All I wanted to make clear was: There is no "safe" in overclocking.
if you don't raise voltage you shouldn't worry to much... My q6600 works stable on default voltage till 3GHZ. Above that I have to play with vcore.
LoRd_MuldeR
10th November 2008, 01:49
if you don't raise voltage you shouldn't worry to much... My q6600 works stable on default voltage till 3GHZ. Above that I have to play with vcore.
I say it again:
Statements like "you shouldn't worry to much" and "it works stable for me" are still no guarantee that it will work for everybody.
If you overclock your CPU (even slightly) and something goes wrong (including CPU destruction) you can't complain to the manufacturer!
People that consider overclocking must consider the risks too...
Sagekilla
10th November 2008, 01:54
Okay I think we're getting a bit off topic here..
Highlights of last few posts:
1) Overclocking is inherently dangerous
2) In moderation (no voltage bumps) you have reduced risk
3) Don't overclock (automated or manual) unless you know the dangers and what you're doing
And now back on topic: Are all the patches for Nehalem (that have been made) already committed or are there some being tested still?
Audionut
13th November 2008, 02:20
he turned on "C.I.A.2" (automated overclocking, up to 7%) in the BIOS, just to give it a try. He certainly didn't mess with voltages and he has got one of these "heavy" ZALMAN coolers. Anyways the machine crashed during boot-up and the CPU plus the board were gone for good.
Power spike or some other issue unrelated to the overclock.
Sorry, 7% over standard will not kill the board or the cpu. Tolerances are built into hardware.
All I wanted to make clear was: There is no "safe" in overclocking.
Yes there is. I've had q6600 @ 3.6 50% over, e6750 @ 3.85 45% over, e6600 @ 3.6 50% over, D940 @ 4.2 32% over, D920 @ 3.8 36% over, among various others.
The D940 and D920 are still going strong after 3+ years, the e6600 has been running for 2+ years, the e6750 for 8 months and the q6600 for 6 months. A few p4 and such are floating around in a box still working fine.
The only cpu i've killed due to overclocking, was one of the p4 prescotts that got it's core ripped out cause unbeknown to me at the time, it was clued to the heat spreader. :eek:
LoRd_MuldeR
13th November 2008, 04:11
Yes there is. I've had q6600 @ 3.6 50% over, e6750 @ 3.85 45% over, e6600 @ 3.6 50% over, D940 @ 4.2 32% over, D920 @ 3.8 36% over, among various others.
So if there is a "100% safe" overclocking, are you the person that will give me the guarantee for that ???
And will you buy me a new machine, in case it turns out that the "100% safe" overclocking didn't exist after all ???
If you are so sure, why not? :p
RunningSkittle
13th November 2008, 04:14
Dells can overclock (using the pin mod)
Tricks the (otherwise locked) motherboard into running the chip at 333fsb instead of 266.
Lord_mulder: stop spreading FUD... Your 'friend' who fried his system obviously had something else wrong. The Q6600 is well known to be able to easily reach 333fsb without any voltage increase. Your not even stressing a modern motherboard with that FSB, as most are rated for 400fsb (and can reach 480+ on air).
So with some cheap DDRII 667 or 800 (downclocked to 667) and a mildly overclocked FSB... its extremely unlikely that you would ever burn out the chip, motherboard, ram or any other component.
LoRd_MuldeR
13th November 2008, 04:44
Lord_mulder: stop spreading FUD... Your 'friend' who fried his system obviously had something else wrong. The Q6600 is well known to be able to easily reach 333fsb without any voltage increase. Your not even stressing a modern motherboard with that FSB, as most are rated for 400fsb (and can reach 480+ on air).
I just tell facts! Obviously you did not read my previous posts or you are not able to understand the words :rolleyes:
Unless there is somebody who actually gives you a guarantee (and will replace your hardware in case of failure), overclocking will be inherently unsafe! You are running the hardware outside of the manufacturer's specifications, so you loose all your warranty immediately. In the case something goes wrong - no matter how likely or unlikely that is - it's your problem! If you personally think the risk is small enough, fine. It's your decision. But don't tell people there is a "100% safe" overclocking, unless you volunteer to give them guarantees they can rely on. If at all, there's a "in 99 of 100 cases it went good for quite some time" overclocking. But certainly not more. This won't change, even if you tell us another 10 times that it does work for you (so far).
Sagekilla already summed all that up nicely in post #15 (http://forum.doom9.org/showpost.php?p=1211624&postcount=15), there is no need to discuss all that again and again...
Audionut
13th November 2008, 04:55
Motherboard manufactures wouldn't have things like C.I.A.2 if it wasn't safe. Intel now releases cpu's that overclock themselves.
And no, you won't loose you're warranty.
It is safe. If you're an idiot with no idea what you're doing, that doesn't make it any more unsafe. Just means you're an idiot.
It's the same as any other thing. If some idiot puts a bigger cam in his engine and f**ks it, it doesn't mean replacing a cam is dangerous.
If you don't feel comfortable overclocking, then don't do it.
If you don't feel comfortable replacing engine parts, then don't do it.
I have a friend who rebuilt a HK Monaro from the ground up with all the shizz. It is a sweet car too. Perhaps now I should go to online forums and tell people they can do the same cause my friend did it, so it's obviously safe. :cool:
LoRd_MuldeR
13th November 2008, 05:00
It is safe. If you're an idiot with no idea what you're doing, that doesn't make it any more unsafe. Just means you're an idiot.
It's the same as any other thing. If some idiot puts a bigger cam in his engine and f**ks it, it doesn't mean replacing a cam is dangerous.
This clearly is a rule #4 violation. It's not in my power to strike you for that, so I will simply use my "ignore" list and move on to some useful discussion...
squid_80
13th November 2008, 06:00
Lord_mulder: stop spreading FUD... Your 'friend' who fried his system obviously had something else wrong. The Q6600 is well known to be able to easily reach 333fsb without any voltage increase. Your not even stressing a modern motherboard with that FSB, as most are rated for 400fsb (and can reach 480+ on air).
I have a Q6600 that won't reach 3Ghz at all - B0 stepping. Since there seem to be plenty of experts here let's see if anyone can tell me what that implies.
There is always a risk with overclocking. Especially with stock low-end pcs. Sure motherboard manufacturers provide overclocking utilities - customers would harrass them if they didn't, plus if they sell a motherboard that lasts forever they're not going to have much future business. Most cases of overclocking gone wrong that I've seen involve a cheap power supply; more clock = more heat = more power = bad smell as the PS craps out and dumps a high enough voltage into the board to kill it and everything attached to it.
RunningSkittle
13th November 2008, 06:44
^ I find that pretty hard to believe.
what were your other settings? There is more to it than simply raising the FSB...
Audionut
13th November 2008, 13:55
This clearly is a rule #4 violation. It's not in my power to strike you for that, so I will simply use my "ignore" list and move on to some useful discussion...
For what it's worth, my statements regarding idiots weren't directed at you.
Although after reading it again I can clearly see how that can be interpreted. :o
Audionut
13th November 2008, 14:07
I have a Q6600 that won't reach 3Ghz at all - B0 stepping. Since there seem to be plenty of experts here let's see if anyone can tell me what that implies.
So I assume you're getting about 2.9Ghz then. That's still over a 20% overclock. Well in excess of the 5-10% that a motherboard will apply with it's "smart" overclock.
There is always a risk with overclocking.
Don't single out overclocking.
Sitting here happily typing this message, there's a risk that my monitor could blow, a plane could crash on the house, a car could fly off the road through the lounge room wall, a tsunami could wipe me out etc etc.
Mobo manufactures would not implement an overclocking system on their boards if it killed every cpu and/or ram their customer bought.
Video card resellers would not sell overclocked cards if they all broke in 2 months.
Hardware manufactures do not sell their hardware that runs within an inch of its life.
Hardware has tolerance.
setarip_old
13th November 2008, 17:11
@Audionut
Hi!Don't single out overclocking.Perhaps you've overlooked the Title of this thread:
"Overclocking debate"
Blue_MiSfit
14th November 2008, 02:28
My G0 Q6600 will happily hit 3 GHz on silent air cooling, but it runs quite hot. I can actually run passively cooled at stock speeds, but it also gets rather hot.
In short, silence is golden to me - and I would rather be silent and a bit slower than screaming fast and noisy :)
My only video encoding benchmark is whether or not I can complete an encode in the time between going to bed, and getting home from work the next evening :) This realistically means 18 hours per movie, which is quite a lot, but acceptable to me. I can comfortably do this with my Q6600, even for BluRay backups, so I don't need to go much faster.
~MiSfit
MeNTaIPaTleNT
16th November 2008, 10:47
My Q6600 has been running at a steady 3.6 for over about a year now. It was also my first attempt at overclocking.
If you don't have the money to replace a blown part due to overclocking don't do it in the first place. Plain and simple.
Yes, if you overclock a cpu and it dies, it is out of warranty (depending on the manufacturer).
I don't condone the following, but in all seriousness, if you were to send it back to the manufacturer for a warranty replacement, how could they possibly tell it was overclocked unless there were burn marks on it??? It's dead...it's not like they can test it...and if there are burn marks on it that would be a result of raising the vcore too high and not having the proper cooling, hence you didn't fully know what you were doing and weren't completely prepared which means you shouldn't have done it in the first place.
I don't trust any "automatic" overclocking settings in the BIOS. IMO the only way to do it is trial and error to keep voltages to a minimum which ensures a decent lifespan for the component. It seems that most PC enthusiasts upgrade their hardware about once every year or two. By that time the hardware is obsolete and you may as well get the performance gain from overclocking.
plonk420
17th November 2008, 06:54
My Q6600 has been running at a steady 3.6 for over about a year now. It was also my first attempt at overclocking.
do you run yours at 100% 24/7/365, though? not just for your happy little three hour Fallout 3/Supreme Commander session...
MeNTaIPaTleNT
17th November 2008, 07:59
It's always clocked at 3.6 but it's never at 100% load unless I'm stress testing it for stability...and since it's not always at 100% load that should mean that the lifespan won't degrade as fast (obviously every chip is different).
*.mp4 guy
17th November 2008, 20:13
I made an OCed c2d rig a while ago, its rock stable at 3ghz (stock voltage) I got it up to 3.6, but it required a voltage increase, and I was building it as a video editing machine for someone else, so I kept it at 3 (still a free 25% improvement). At load it hits 46-48 celsius, and it is 72 hour encoding + dualprime stable (thats better then quite a few oem machines, I had an old dell that couldn't do that).
I realise we are talking about c2q's here, but they are basically the same, you just need more cooling. though this isnt why its a c2d, thats because when I built it the cheapest quad was nearly $1K, and there weren't any compelling 4 core capable aps.
As long as you stay withing the thermal and voltage specs of a chip, and all of your hardware is good enough quality , a stable overclock is perfectly safe.
Your "friend" shouldn't have trusted the auto overclock feature, they are notoriously bad, often increasing vcore far above what it needs to be at even for tiny overclocks, and obviously, as soon as you take the chip out of voltage/temp spec, there is risk. No automatic overclocking is ever going to be as safe, high performance or reliable as an actual overclocker who knows what they are going.
plonk420
17th November 2008, 20:57
how many of you OTHER OCers run your proc at 100% 24/7/365? i sure wouldn't touch it (unless it was like a low-single-digit-percent OC) as i for one do...
Audionut
18th November 2008, 03:51
how many of you OTHER OCers run your proc at 100% 24/7/365? i sure wouldn't touch it (unless it was like a low-single-digit-percent OC) as i for one do...
How many people run their cpu's at 100% load 24/7/365 regardless.
RunningSkittle
18th November 2008, 04:23
I have no problem leaving mine doing heavy rendering, FAH, or encoding a BR rip for days at a time. I would not leave it on 24/7/365... the power bill would be enormous.
Are you suggesting that an OCd processor with temps no where near the TJmax will somehow burn out because its running a few degrees warmer than at stock?
plonk420
18th November 2008, 09:18
i'm just wondering if any of you "extreme OCers" (or even just mild OCers) would trust your OC job running it at 100% all day every day. i trust my NON overclocked machine to do so.
i run at 100% all winter long as my stupid apartment complex has electric-only heating (and each unit pays for their own elec). so it's either waste electricity NOT doing anything, or waste electricity crunching numbers for Distributed Computing Project X. hell, i do it all year except for the hottest months of the summer.
i might trust a single digit OC if running continuously, but if there's added risk and such a small benefit, why bother?
i also say, "why bother?" if it's not safe enough to run 24/7. (i guess if it's safe enough to do a full day+ long encode every week or two or so (at no detriment to long term stability), i concede that THAT is ok in my book)
RunningSkittle
18th November 2008, 10:02
Of course I trust it to, but my room would turn into a space heater... and the roommates would look at the power bill and be like "wtf"?
of course x264 cant bring a CPU anywhere near the TJmax... so im never worried about my chip.
MeNTaIPaTleNT
18th November 2008, 16:09
Yes, I would trust my OC'd rig to run 100% 24/7/365 because it is properly cooled, but who runs it at 100% 24/7/365????
As a gamer, OC'ing a chip from 2.4 to 3.6 will be a very noticable gain in performance.
AFAIK the only real risk is overheating the cpu. I havn't heard many stories where if a CPU goes, it takes everything else with it unless voltages for the CPU, NB, and RAM are through the roof. Again, if you can't afford to replace the component, don't do it.
If you keep voltages down and cool it properly it is safe IMO.
*.mp4 guy
18th November 2008, 21:34
I would trust the previously mentioned machine to run 24/7/365, It is completely rock stable. I woudn't generally trust a highly over clocked machine to do that, but intels core chips just have such unbelievable headroom, that its silly not to overclock them.
If I wanted to buy a chip that was rated to run at almost the speed I got out of that machine, it would have cost an extra ~$800-$900, and been slower, and it wouldn't have been measurably more stable.
saint-francis
19th November 2008, 17:49
MY Q6600 has been running at 100% load almost all of the time for over a year. I would certainly trust it to run at 100% load 24/7/365. The only reason it doesn't is because occasionally a movie finishes encoding before I have the chance to queue up more.
Oh, and that's at a 50% OC minimum.
burfadel
21st November 2008, 22:33
It all depends on room temperature too, if it barely gets above 22C you're okay. If it varies then you have to keep in mind whats good at 20C may be disastrous at 30C! (or more)
vBulletin® v3.8.5, Copyright ©2000-2012, Jelsoft Enterprises Ltd.