Overclocking Guide Part 1: Risks, Choices and Benefits

[funnyvlad]

Nevermind

 

[dragosgorjan]

Where do you get this 101C on a DIE Prescott...these are stories for childrens.
I have a 630 Prescott @3600 my idle temp is 39-40 load 53-55(max).I think the majority of the people how owns Prescott core do not know how to mange them to put them right in the soket to cool them right with panel fans etc...My performance in the stock mode @3000 are all most the same with 3500+ Venice(i have the second system with AMD) (all most=little difference) its like 50-50 because in some test Prescott wins and in others the Venice wins but any way with 10-95 points difference its like nothing.

Venice /E6/ is not so good overcloker in comparation with Prescott /NO/ wich is super overcloker.
If we analyse the risks for every core the prescott have the native design to work under hight temp and at high freq wich will put him like(all most)out of the risk,Venice do not have the high thermal design(in my opinion is no need because its working very good in stock mode and if we compare price and performance it is perfect) but if will overclok to much it will be unstable.

The benefits are minimal even if we overclok both at maximum posible it will not be at least in performance like CORE 2 Duo 6300.

 

[Crashman]

"More Stable" is possible due to things like varying operating conditions. If I have a PC that's stable at a room temperature up to 25C, it would be 100% stable in an air-conditioned office but unstable in a hot climate.

The thing is that my office temperature ranges from ~15C to ~45C! I usually leave after a few minutes when it goes above ~42C. Now, if my system was stable 350 days a year but not stable those few hottest days, but I replaced the processor with something that was stable on those hottest days, it would be "more stable".

And it's not just about cooling, as processors get closer to their limits they are more likely to err with tiny voltage changes that are normal in computer system.

 

[caamsa]

Do you run your cpu at that speed 24/7? or is that just for bragging rights and benchmarks? Also how much extra did you have to spend in cooling extra power and better memory to get that over clock? Just curious Lastly what do your benchmarks compare to the 6600 or the 6700 6800?

 

[Crashman]

Great, you come up with a list of the specific information you'd like included.

 

[Crashman]

That's a fairly tall order considering most experienced overclockers don't even know what all those settings are, and only about half the high-end boards have them. Normal settings are tCAS, tRP, tRCD, and tRAS.

 

[caamsa]

Why not explain each one in detail to us one at a time.

 

[caamsa]

I want to know what you think of a computer that auto overclocks and scales back when needed. No need to make adjustments etc. Automatically does it for you. Can it be done? I think it would be great. Remember the old turbo button?

 

[Crashman]

It's been done for quite a while, many boards offer automatic overclocking that kick up CPU speed whever it reaches a certain percent cycle usage. You can even set the amount you want it to overclock by.

 

[enewmen]

Good article on the very basics. I hope to see more articles like this . I also hope someone can explain in more detail about relationship between a BUS speed and RAM speed. Also what the Timing: 5-5-5-15 is and how to adjust it correctly.

 

[Luscious]

Nice article, it's been a while since Toms did an OC piece.

It would be nice to see part 2 look at the various cooling options from some of the higher end suppliers and demonstrate the different technologies such as water, peltier, phase change and heatpipe.

I remember Toms did that 5GHz LN job a few years back on a P4 - how about another extreme OC job this time on two 8800GTX in SLI!

 

[yyrkoon]

Are you serious about the idle temps on Prescotts? I think you are helping propagate an urban legend. Mine has never gone above 50C even at full load on stock HSF and OCed from 2.8 GHz to 3.57 GHz (that's a 27.5% gain) and dare to say it can go beyond 4.3 if it weren't for my cursed Cheapazaki MoBo. I mean no offense but... Are you putting enough thermal goo on your processor die?

About the Numero Uno Oncho: It's supposed to be Spanish but Oncho is wrong, perhaps he meant Ocho so my translation is Number One Eight. By the way it's "enchiladas" not enchaladas .

Saludos

You've got to be kidding me . . . "Honcho" ring a bell ? Come on . . . you know "Head Honcho" . . .

Prescotts, and some D's(90nm) were very hot, figured you guys, who SEEM to be reading this forum, would also read a few articles

The thing is, some people would use their systems for close to 8 years, hell I have an 7-8 year old compaq presario that serves just fine as a Samba file sharing server (granted, 40GB storage isn't a lot, I've got .5TB in my current system . . .). Overclocking, hmm, done it, and I've heard of worse things to do with your time, but what gets me, is people who will spend $400usd or more on on a decent water cooling system to get speeds of a CPU they could have paid an additional $300usd, and trouble free . . .

Also, that little "5% gain for 2 MB more L2" line is BS, trying actually using something that will USE that extra L2, and tell me its only 5%, games aren't everything you know.

Article stank, could have read what was said on wiki, how about some test cases, or is this going to be another orphaned article ? . . .

How about something more like "how-to 802.3ad link aggregation", or how-to inexpensive iSCSI Targets" . . . Seems more and more every day, this site is turning into a hand holding site, for people who know nothing about hardware. I dare say, give us a challenge will ya!?

 

[charge-n-go]

"Electromigration takes place on the actual silicon chip of your CPU, in areas that operate at a very high temperature, and can cause permanent damage to the chip. Before you start to panic, you should first realize a few things. CPUs are designed to run at temperatures between -25 and 80 degrees Celsius. To give you an idea, 80 degrees Celsius is a temperature that nobody is able to touch for longer than 1/10 second. I have never come across a CPU at this temperature. There are plenty of ways to keep the CPU case at less than 50 degrees Celsius, which increases the probability of keeping the chip inside at less than 80 degrees. Also, electromigration does not immediately damage your chip. It is a slow process, which more or less shortens the life span of a CPU running at a very high temperature. A normal CPU is meant to live for about ten years. However, in ten years nobody is going to be using a CPU with today's technology. I won't even use my CPU anymore in 2 months. If you want to be kept free from this electromigration scare, you have to do as much as possible to cool the CPU. Cooling is the Numero Uno Oncho in overclocking! Never ever forget that!"

I dont think the statement above is 100% correct. Temperature is one of the factors in electromigration, but the deterministic one is current density in the interconnections. When voltage is increased, the electric field within the cpu is increased too, causing the electrons to accelerate faster. Thus, even if the temperature is kept low, electromigration effect is still greater than default voltage.

I would say electromigration has 3 contributors:
1. voltage
2. speed
3. temperature

 

[yyrkoon]

Completely missed the point of Hysteresis also. See wikipedia for more info

 

[Crashman]

I'd agree with the article that the matter of importance is:
1.) Voltage
2.) Heat
3.) Speed

The order in the article is from least to greatest, so the list appears reversed but is actually the same order.

 

[charge-n-go]

yea, that's true. Voltage is the biggest enemy in overclocking. Sometimes, voltage set way beyond the spec sheet will cause the CPU to die in a few months even if we are using very good cooling system. It's because the electrons are accelerated so fast that they break the interconnection wall.

 

[Crashman]

Yes, well, voltage is the electrical version of force, it's easier to distort things when you add force. I'm not exactly sure, but I think heat increases platicity, reducing the amount of force required to close a gate.

You're the expert so you tell me

 

[CaptRobertApril]

Where do you get this 101C on a DIE Prescott...these are stories for childrens.

Hmm... let's see... first of all, thanks for calling me a child as at my age any decrease of the number of candles on the cake is welcome. As for where he got it? He got it from this temperature sensor gun that he calibrated within an inch of his life. And it was the lowest temp of his various readings, as his BIOS and Intel Thermal Thingy were always higher.

You've got to be kidding me . . . "Honcho" ring a bell ? Come on . . . you know "Head Honcho" . . .

Hey, Honcho I know. Oncho is just illiterate spelling!

 

[charge-n-go]

well, i am not really an expert. I heard from the production engineers that voltage is always the #1 enemy nowadays, because the CPU already has built in thermal diode at the hottest spot of the CPU. When the temperature exceed the spec, it will automatically do geyserville (or more commonly referred as Speed Step) transition to lower frequency.

These days, when they want to launch certain products with higher speed bin, the headache part is whether to increase the Vcore or not. Well, for overclockers like us, increasing Vcore is darn common. But for them, this method shall be the last resort, as they knew voltage can cut down CPU lifespan quite significantly.

Heat reduce the amount of force to close a gate, but faster switching speed + higher temperature also induced a lot of leakage current or reverse saturation current, which is not desirable in any ICs.

 

[yyrkoon]

Hey, Honcho I know. Oncho is just illiterate spelling!

Yeah, well, what do you expect, his "spanish" grammar was incorrect as well (my spanish is far from perfect, but "Oncho de numero uno ?") *shrug*, but I just Figured the "Oncho" was spelled as if you pronounced the h silent

 

[Crashman]

Doctor Thomas Pabst is a native German speaker who wrote well in English, but add Spanglish and see how easy it is to mess up.

 

[neiroatopelcc]

Are you serious about the idle temps on Prescotts? I think you are helping propagate an urban legend. Mine has never gone above 50C even at full load on stock HSF and OCed from 2.8 GHz to 3.57 GHz (that's a 27.5% gain) and dare to say it can go beyond 4.3 if it weren't for my cursed Cheapazaki MoBo. I mean no offense but... Are you putting enough thermal goo on your processor die?

About the Numero Uno Oncho: It's supposed to be Spanish but Oncho is wrong, perhaps he meant Ocho so my translation is Number One Eight. By the way it's "enchiladas" not enchaladas .

Saludos

I've got no experience with prescotts, but I once had an athlon processor rated at max 110C - and infact it did idle above 80C and peaking at around 105C
Turned out when I trashed the system a year later, that sugar had melted between teh cpu and cooler (I once dropped sugar near the pc) .... probably causing the high temps .. but point is - it IS possible to see high temps in computers, just not if they're kept in a proper working condition.

 

[irishsk8rpatrick]

AH i remeber the Turbo button! i had one when i was in the first grade, i never pushed it..my friend said thats the self destruct button...what did it do?


Ah the 5ghx project i remenber that, with some liguid nitrogen.....ah that stuff was really cool.

 

[Crashman]

The "turbo" button slowed your system down, I believe to 8MHz, for older programs that relied on CPU clock cycles for timing.

I could be wrong about the speed, why not google it?

 

[irishsk8rpatrick]

oh...i never thought of that...lol oops...the computer i had was like...a 99 or 100 mhz pentium pro.