Do overclocked CPU cores get worse over time?

[apoppin]

Originally posted by: JackBurton

Originally posted by: apoppin
fortunately i did my research and the extreme FSB of a good OC more than makes up for it

Ummm, not so much.

what are you talking about?

let me try again:

the extreme of a good FSB for e4300 is what raises the Mhz of the CPU to make it faster than any stock intel C2D

by itself, the bandwith of the raised FSB is not so important as the ultimate speed your OC'd CPU can attain ... and MANY e4300s can be overclocked way past the stock speed of ANY 'extreme edition' ... the extra cache becoming less meaningful after a +15% or so OC

 

[VERTIGGO]

well I'm actually waiting on a cheaper version than the Prometias or whatever. Hopefully OCZ comes through soon (I know, it was "soon" a year ago). But anyway, I'm going to do what I can with an E6600 and a Thermalright IFX-14 until phase drops in price and the 45nm quads are more reasonable (of course they have to release before that can happen).

 

[Cookie Monster]

IF you can cool the OCed parts very well, then theres no need to worry about OCes being unstable over long period of time. (will require daily maintenance e.g removing dust from HS fins)

 

[apoppin]

Originally posted by: Cookie Monster
IF you can cool the OCed parts very well, then theres no need to worry about OCes being unstable over long period of time. (will require daily maintenance e.g removing dust from HS fins)

yep, i got the Thermalright Ultra-120/Scythe S-FLEX SFF21F combo ... my $115 e4300 runs easily at 3105Mhz ... at *stock* vcore ... as long as my room is cool [up to low80sF], there are no problems with stability or thermal issues whatsoever [unless you consider 62C in TAT as the max temp under load "high"]

and there is a lot of HS to clean out in that Thermalright ... it looks like i could almost use it to replace my car's radiator :Q
--well ... not really ... but it barely fits in my case ... i had to move my case side-fan to the outside of the case

 

[JackBurton]

<div class="FTQUOTE"><begin quote>Originally posted by: apoppin
<div class="FTQUOTE"><begin quote>Originally posted by: JackBurton
<div class="FTQUOTE"><begin quote>Originally posted by: apoppin
fortunately i did my research and the extreme FSB of a good OC more than makes up for it
</end quote></div>
Ummm, not so much.
</end quote></div>
what are you talking about?

let me try again:

the extreme of a good FSB for e4300 is what raises the Mhz of the CPU to make it faster than any stock intel C2D

by itself, the bandwith of the raised FSB is not so important as the ultimate speed your OC'd CPU can attain ... and MANY e4300s can be overclocked way past the stock speed of ANY 'extreme edition' ... the extra cache becoming less meaningful after a +15% or so OC
</end quote></div>

---------------------------------------------------------------------------------------------------------

Ok, you were throwing me off with the "extreme FSB" comment. I took it as you were saying an overclocked e4300 @ 2.93GHz is equal to a X6800 @ 2.93GHz, and the "extreme FSB" makes up for the lack of cache, which is not true. I don't even know why you brought up the FSB. Why not just say overclocking it 15% faster than a stock C2D processor makes up for the lack of cache. e4300s are multiplier locked so obviously you'll need to raise the FSB to overclock it.

And FYI, to even MATCH the X6800 stock performance you'll need to overclock that e4300 to ~3.4GHz (15%). I would have spent the extra money and bought an e6600, but that's just me.

Edit: Am I doing something wrong, or is this quoting jacked up on this new forum software?

 

[apoppin]

I don't even know why you brought up the FSB

so you could clarify my poorly worded comments for all of us
... and it certainly doesn't "hurt" to have the FSB pumped way up ... that IS why i am dumping my XMS PC2-6400 for Ballistix PC2-8500 ... don't tell me performance will be the same with a higher divider and increased FSB

and is my e4300's multi locked? ... there ARE options to change it in my BIOS
... i haven't bothered to even try changing it as i have such excellent results at 345x9 ... if i cool my room ... and up the voltage, who know how far she will go?

you can have your e6600 ... my e4300 is a "placeholder" for Penryn

--and i will still have my thermalright for its OC

 

[StopSign]

Originally posted by: apoppin
and is my e4300's multi locked? ... there ARE options to change it in my BIOS

You can change it between 6x and 9x. That's still considered locked.

 

[JackBurton]

<div class="FTQUOTE"><begin quote>Originally posted by: apoppin
<div class="FTQUOTE"><begin quote>I don't even know why you brought up the FSB</end quote></div>
so you could clarify my poorly worded comments for all of us
... and it certainly doesn't "hurt" to have the FSB pumped way up ... that IS why i am dumping my XMS PC2-6400 for Ballistix PC2-8500 ... don't tell me performance will be the same with a higher divider and increased FSB

and is my e4300's multi locked? ... there ARE options to change it in my BIOS
... i haven't bothered to even try changing it as i have such excellent results at 345x9 ... if i cool my room ... and up the voltage, who know how far she will go?

you can have your e6600 ... my e4300 is a "placeholder" for Penryn

--and i will still have my thermalright for its OC</end quote></div>

------------------------------------------------------------------------------------------------------

What is your 56K NetZero connection a placeholder for? 128K ISDN?

 

[nyker96]

I think you need to clean your heatsink installation overtime. mine gets clogged like in 6months and I completely wash it and clean it then it comes back to original performance.

 

[apoppin]

Originally posted by: JackBurton
<div class="FTQUOTE"><begin quote>Originally posted by: apoppin
<div class="FTQUOTE"><begin quote>I don't even know why you brought up the FSB</end quote></div>
so you could clarify my poorly worded comments for all of us
... and it certainly doesn't "hurt" to have the FSB pumped way up ... that IS why i am dumping my XMS PC2-6400 for Ballistix PC2-8500 ... don't tell me performance will be the same with a higher divider and increased FSB

and is my e4300's multi locked? ... there ARE options to change it in my BIOS
... i haven't bothered to even try changing it as i have such excellent results at 345x9 ... if i cool my room ... and up the voltage, who know how far she will go?

you can have your e6600 ... my e4300 is a "placeholder" for Penryn

--and i will still have my thermalright for its OC</end quote></div>

------------------------------------------------------------------------------------------------------

What is your 56K NetZero connection a placeholder for? 128K ISDN?

inexpensive wireless broadband ... it is finally arriving to my Rural area this Summer

anything else?

 

[CTho9305]

Originally posted by: myocardia
Anyone who knows how to overclock has none of the problems you've just mentioned. I've been overclocking since '91 or '92, and I've never had a single processor go bad, never had to reinstall software or any OS because of errors, and never used anything except air cooling.

Yeah, and I know this one guy who drove home drunk from a bar and nothing bad happened to him either :roll:.

Overclocking is hit-or-miss. It's good that you've had good luck, but not everybody does.

 

[apoppin]

Originally posted by: CTho9305
<div class="FTQUOTE"><begin quote>Originally posted by: myocardia
Anyone who knows how to overclock has none of the problems you've just mentioned. I've been overclocking since '91 or '92, and I've never had a single processor go bad, never had to reinstall software or any OS because of errors, and never used anything except air cooling.</end quote></div>

Yeah, and I know this one guy who drove home drunk from a bar and nothing bad happened to him either :roll:.

Overclocking is hit-or-miss. It's good that you've had good luck, but not everybody does.

total absolute nonsense ... OCing is not "hit or miss" although the final results may vary

perhaps it is "hit or miss" -- for you

everyone who does it right has good luck . ... some better than others

 

[imported_jX]

total absolute nonsense ... OCing is not "hit or miss" although the final results may vary

perhaps it is "hit or miss" -- for you

everyone who does it right has good luck . ... some better than others

I agree. It's like driving. as long as you're a good driver you'll never have an accident.

Oh, wait, I guess that's not how it is at all...

CTho's right. Not all chips can be pushed very far at all, some can go for ages. My A64 managed to take a full 20% push on the FSB speed with nary a hiccup, no need to mess with voltage, watercooling, etc. granted I have a very decent cooler to start with, but I just pushed the FSB from 200 ro 240 and the chip was fine, not a single issue, and barely a degree warmer. However, some chips just can't be pushed far at all.

I grant you there's a right way and wrong way, but since you're taking the chip beyond it's specifications by the very nature of overclocking, there's ZERO guarantee you'll succeed. A good likelihood you'll succeed, but no guarantee. It _is_ hit and miss.

 

[apoppin]

Originally posted by: jX

total absolute nonsense ... OCing is not "hit or miss" although the final results may vary

perhaps it is "hit or miss" -- for you

everyone who does it right has good luck . ... some better than others

I agree. It's like driving. as long as you're a good driver you'll never have an accident.

Oh, wait, I guess that's not how it is at all...

CTho's right. Not all chips can be pushed very far at all, some can go for ages. My A64 managed to take a full 20% push on the FSB speed with nary a hiccup, no need to mess with voltage, watercooling, etc. granted I-have a very decent cooler to start with, but I just pushed the FSB from 200 ro 240 and the chip was fine, not a single issue, and barely a degree warmer. However, some chips just can't be pushed far at all.

I grant you there's a right way and wrong way, but since you're taking the chip beyond it's specifications by the very nature of overclocking, there's ZERO guarantee you'll succeed. A good likelihood you'll succeed, but no guarantee. It _is_ hit and miss.

you guys aren't really overclockers
-most of you - by your replies - don't have a clue about it - just misconceptions

you are wrong ... you will almost ALWAYS get an OC if you do your research BEFORE you buy your CPU and you must know what MB to get as well as the RAM and appropriate cooling

the way *you* do it IS hit-or miss
--the way *i* do it is 100% guaranteed [along with all the experienced guys here]

stick around - keep an open mind ... we can help you OverClock successfully ... you are in the right place
--and welcome to ATF !

 

[StopSign]

apoppin is right. C2Ds are all guaranteed overclockers. Every chip, no matter how "bad", will reach at least 2.5 GHz. Refuting this means you have no idea what you're talking about.

I grant you there's a right way and wrong way, but since you're taking the chip beyond it's specifications by the very nature of overclocking, there's ZERO guarantee you'll succeed. A good likelihood you'll succeed, but no guarantee. It _is_ hit and miss.

I usually hear that from people who have never overclocked before, or don't know how overclocking works. Is it guaranteed that every E6300 will reach E6400 speeds? Yes? Then by the very nature of your argument, you are wrong.

 

[VirtualLarry]

The key factor here is "engineering tolerances". Specifications in the real world vary. Voltages and temps vary. That's why electronic components, specifically CPUs, have a built-in set of engineering tolerances.

Due to manufacturing issues, and how CPUs are made and then sold, some chips have more excess tolerance than others.

Knowing that, and how to extract excess performance by controlling environmental factors (voltages and temps), by pushing the limit of those engineering tolerances, is the key to overclocking.

It is not as hit-or-miss as some think, as ALL CPUs have some level of tolerances built-in. The C2D chips have an abundant excess of tolerance, as Intel intentionally rates them at lower clock speeds than they are capable of. (Primarily due to market competition reasons, not technical reasons.)

 

[VirtualLarry]

Originally posted by: StopSign
You can change it between 6x and 9x. That's still considered locked.

I always thought that it was considered unlocked, downwards. AMD CPUs are the same way.

 

[apoppin]

Originally posted by: VirtualLarry
<div class="FTQUOTE"><begin quote>Originally posted by: StopSign
You can change it between 6x and 9x. That's still considered locked.</end quote></div>
I always thought that it was considered unlocked, downwards. AMD CPUs are the same way.

awesome ... that gives me extra flexibility ... i never had any experience with unlocked intel ES ... so this is really a cool tool .. probably can find the chip's 'sweet spot' easier.

a serious overclocker rarely buys a chip near the top of the speed class unless he uses extreme cooling ... and it is rarely 'luck' ... luck only plays a part in "how far" you can go ...

 

[VERTIGGO]

Unlocked means you can go higher, though back in the day, an FX-55 was unlocked meaning at stock HT bus you could run higher than stock speed. Now (if they are correct in saying 6x to 9x) you can clock up to 3600MHz on a stock FSB. Is this correct, owners?

 

[apoppin]

Originally posted by: VERTIGGO
Unlocked means you can go higher, though back in the day, an FX-55 was unlocked meaning at stock HT bus you could run higher than stock speed. Now (if they are correct in saying 6x to 9x) you can clock up to 3600MHz on a stock FSB. Is this correct, owners?

9x266=2394
for me [e4300] 9x200=1.8Ghz ... dropping to 6 x 200 = 1200

huh?

"unlocked" refers only to the multiplier ... you can "go higher" as i did - so far - only using the FSB

you can clock up to whatever you can using the CPU:RAM dividers
--the multipliers just give you [a lot of] extra flexibility in determining your OC

 

[StopSign]

Originally posted by: VERTIGGO
Unlocked means you can go higher, though back in the day, an FX-55 was unlocked meaning at stock HT bus you could run higher than stock speed. Now (if they are correct in saying 6x to 9x) you can clock up to 3600MHz on a stock FSB. Is this correct, owners?

Do the math yourself. Stock FSB is 266. Highest multiplier the E6600 can use is 9x. How do you get to 3600 MHz with that?

The multiplier thing is still exactly the same as before. All non-Extreme chips have locked multipliers. "Unlocked downward" is still locked.

If you thought the default clock for an E6600 is 400x6, then that's inconsistent with Intel's advertised FSB, which is 1066 (266x4 in reality). Logic, people!

 

[VERTIGGO]

You're right thanks, I did some more reading last night (which added as much confusion as clarification). I was under the impression that a stock FSB was 400MHz (DDR2 800 the same as DDR 400 is 200MHz), now I'm realizing that FSB is totally different than clock speed (HT for AMD). So DDR2 800 (P2 6400) is actually overclocked ram?

 

[StopSign]

I'm not sure what you mean by that. DDR2-800 is 400 MHz, just like how DDR-400 is 200 MHz.

 

[CTho9305]

Originally posted by: apoppin

Originally posted by: jX

total absolute nonsense ... OCing is not "hit or miss" although the final results may vary

perhaps it is "hit or miss" -- for you

everyone who does it right has good luck . ... some better than others

I agree. It's like driving. as long as you're a good driver you'll never have an accident.

Oh, wait, I guess that's not how it is at all...

CTho's right. Not all chips can be pushed very far at all, some can go for ages. My A64 managed to take a full 20% push on the FSB speed with nary a hiccup, no need to mess with voltage, watercooling, etc. granted I-have a very decent cooler to start with, but I just pushed the FSB from 200 ro 240 and the chip was fine, not a single issue, and barely a degree warmer. However, some chips just can't be pushed far at all.

I grant you there's a right way and wrong way, but since you're taking the chip beyond it's specifications by the very nature of overclocking, there's ZERO guarantee you'll succeed. A good likelihood you'll succeed, but no guarantee. It _is_ hit and miss.

you guys aren't really overclockers
-most of you - by your replies - don't have a clue about it - just misconceptions

you are wrong ... you will almost ALWAYS get an OC if you do your research BEFORE you buy your CPU and you must know what MB to get as well as the RAM and appropriate cooling

the way *you* do it IS hit-or miss
--the way *i* do it is 100% guaranteed [along with all the experienced guys here]

stick around - keep an open mind ... we can help you OverClock successfully ... you are in the right place
--and welcome to ATF !

In my book "almost always" and "always" are the difference between "hit and miss" and "guaranteed".

 

[nullpointerus]

Originally posted by: VERTIGGO
You're right thanks, I did some more reading last night (which added as much confusion as clarification). I was under the impression that a stock FSB was 400MHz (DDR2 800 the same as DDR 400 is 200MHz), now I'm realizing that FSB is totally different than clock speed (HT for AMD). So DDR2 800 (P2 6400) is actually overclocked ram?

The Core 2 processors' effective FSB is quad-pumped; DDR2 is double-pumped, meaning: base FSB x 4 = effective FSB speed, and base FSB x 2 = RAM speed. If you are saying that PC2-6400 is for primarily for overclocking because no current Core 2 processor has a 400x4=1600 MHz effective FSB speed, then you are correct. In that sense, it is "overclocked ram."

I think that hi-speed RAM sells because people want to keep the RAM divider at 1:1, which is the normal base FSB x 2 speed. Many processors have a relatively low multiplier, upwards locking, and high base FSB. The only way to overclock these chips is to increase the base FSB, which also increases the memory speed--often past the memory's spec.

I stumbled upon the newer E4400 chip, which is still upwards-locked, but it also has an atypically high multipler and low base FSB, which together should allow me to use very inexpensive PC2-5300 RAM even when the CPU is overclocked to 333x10=3330 MHz, since 333x2=666 MHz which is approximately the speed of PC2-5300 or, as it's also known, DDR2-667.

$64 for 4 GB after mail-in rebate!! Details are here:

http://forums.anandtech.com/me...=2065504&enterthread=y

Let's hope the rebate comes through.


EDIT 1: Clarified a few lines to avoid confusion.
EDIT 2: Darnit, editing posts containing quotes is still screwed up in Opera.
EDIT 3: Here is an article to verify what I'm trying to say.