Overclocking AMD Athlon XP 2000

I have got a 2000+ and am using a mobo that handles a 400 FSB. I have so far put the clock to 150, as it wouldnt boot at 160. I have put a Volcano 11+ on it, and put the multipier to 13. So it is running at 1.95Ghz. But I am wondering if it is better to get the clock speed higher, so that i can get better memory bandwidth etc. as i have PC3200 RAM, so at the moment it is being underclocked, well, either that or it is not performing well under the benchmarks.
At the moment I am running the CPU at 32 degrees. This is a brandnew motherboard too, so I am pretty sure I cant have a cracked diode yet.
I am using a thoroughbred core.

here is a screen shot of my temps and CPU settings. (for some reason it now detects as a 2100+ when i changed the multiplier)

Anyway, could someone suggest what I should do?
Does changing the multiplier just put more stress on the CPU?
How about dropping the mutiplier and try for a higher clock speed?
Btw, i am using a albatron KX600S mobo.

 

Temperature is probably the best indicator of wisdom in trying different configurations. I'm really concerned that your "too good to be true because it's so low" 32.5C could be accurate. Heat is the enemy. Whenever you speed things up, you get more heat even without upping the CPU voltage. Think of the bits of data as little buckets inside the CPU. When you make them go faster, they dump their little loads of current through the CPU faster, increasing the current, power consumption, and thus the heat generation. When you speed things up, the bits take as long as before to do the dump, they just do it more frequently. Therefore, the buckets do not somehow get smaller (with less current) when you speed up the clock.

Aside from the heat issue, my research indicates speeding up the FSB is usually more important than speeding up the internal CPU clock, because usually, the CPU must frequently access RAM across the FSB. Only for smaller programs that can fit entirely, or almost entirely within the CPU cache, does access to RAM (and FSB speed) become less important, and then internal CPU speed becomes more important. And if you can up the FSB while letting the CPU coast along at its factory internal clock by just lowering the multiplier, than I figure you get all the cache performance you started with, plus you up the FSB, and maybe you still enjoy stability and longevity.

I have an XP2000+ Tbred on an Albatron KX18D 400FSB with two banks of DDR400 thereby enabling dual-channel (800FSB). To my delight, I discovered I could change the multiplier in BIOS with no wire tricks or painting bridges needed to unlock the multiplier. Which I did to 8x, and then set the FSB clock (aka CPU external clock) to 200MHz. As discussed above, the FSB data rate is CPU external clock x2 for DDR, and clock x4 for dual channel DDR. Of course, this means my internal CPU clock is now 1600MHz (8x 200) instead of its factory 1667. The Albatron is substantially faster than the three year old IWill XP333 mobo I replaced, which would get unstable above 12.5x 100MHz clock with the same CPU (but newer RAM).

This all means that internally, the CPU's less than 5% underclocked and loafing along. But externally, the CPU's FSB line drivers are running at 200MHz instead of the factory 133MHz, so the drivers are overclocked 50%. And so they are generating more heat, but so far everything's okay with CPU idle BIOS temperature about 50C/122F, compared to a second IWill's Pally at 41C/105F, upon which I am writing this. Although 50C is just about the edge. I think the reading is higher than real, but I'm happy to use it as though it's accurate.

So that's as far as I want to take my own system. Now, by definition if I want to see how far I can push it by upping the multiplier higher than 8x, and maybe the FSB too past 200MHz, and probably also the CPU and RAM voltages, then I'm flirting with frying something, because until I actually do fry it, I can't know that the piece has maxxed out. So if I did it, I would do everything very slowly, keeping a keen eye on the temperature. Run it for a while to check operational stability, and then up it a little more.

Checking stability to some folks means stressing the system for a minimum of 24hours using a stress program, many of which are free on the net, like Prime95.

I would guess to first push the FSB to see where the rest of the system is at. Supposedly, this Albatron can get up to a 300 FSB clock (1200 FSB dual channel data rate), and DDR500 is here and DDR600 is right around the corner. Hoo hah! Visions of an XP2000+ CPU on a 1200FSB with a multiplier around 5.5x!

When your FSB becomes unstable, you may think about upping RAM voltage to restore stability. Supposedly, the default RAM voltage of 2.5V may be raised to 2.8V, that seems universally accepted as safe, and higher voltage is, umm, an experiment into the unknown. But above 3.0v frightens most folks. So raise the voltage a little to regain stability, raise the FSB until it's unstable again, raise the voltage again, raise the FSB again. Determine max settings by iteration. Watching the temperature.

After determining the mobo's FSB limit, you lower it a tad to assure stability.

Next, you go to work on the CPU by increasing its multiplier. Note: changing the multiplier maintains the FSB clock, and only changes the internal CPU clock. You again go slowly, and checking each setting for stability. Same process as before: raise the multiplier until unstable, increase voltage until stable again, etc. For the CPU, the highest safe voltage might be 1.85v. But I've read of some folks going up to and over 2.0v. Scary.

Say, what's that acrid smell?

 

I meant, pushing the FSB, only, by itself, while keeping the internal CPU clock the same. This requires that you lower the multiplier every time you speed up the FSB. If you increase the FSB without lowering the multiplier, then the internal CPU clock will increase together with the FSB. But you don't want to increase the internal CPU clock until later, when you increase the mulitplier while not changing the FSB. Also, pushing things a little at a time might mean as little as 5MHz at each step.

 

Well, first, thanks a heap for your reply YankInOZ, that should keep me going for a while, unless i fry my CPU straight away, hehe.

But, just one quick question, I thought the FSB was just the clock speed x2. So when i up the clock speed, i am automatically upping the FSB. So a clock speed of 150 is a FSB of 300. Or have i got something wrong with that.

 

There are two clock speeds, and there is a data transfer speed. There is the faster internal CPU clock speed and the slower external CPU clock speed. These two speeds have a ratio equal to the multiplier. The Front Side Bus (FSB) clock speed is equal to the external CPU clock speed. The FSB data transfer rate can be twice the FSB clock speed, because DDR RAM can transfer data on both the leading and trailing edges of the clock. A clock is a square wave. That is, you could draw a representation of one clock cycle by starting at the left edge of a sheet of paper, drawing a line, say 1/4" level toward the right, then 1/4" up (which is the leading edge), then 1/4" level toward the right, then 1/4" down (which is the trailing edge); and finally you have finished the first cycle, and are ready to start drawing the second cycle. So, DDR transfer is at twice the clock rate, or four times as fast if you have a dual channel mobo (which sounds like in addition to DDR, it is also twice the data width, and thus doubles the data transfer rate all over again).

Here is an earlier bit of information.

What is important here is that the FSB speed number can be ambiguous, if the number is such that it could be in both the clock speed range (100MHz to 300MHz), as well as also within the data speed range (200MHz to 1200MHz). That is, any Venn diagram intersection number from 200MHz to 300MHz is ambiguous, because any number from 200 to 300 might be the DDR number for a 100 to 150 clock, or the clock with a 400 to 600 DDR. Usually you can disambiguate by context, but you could just be precise in the first place by adding either the word clock or data to FSB, such as 133 FSB clock, or 200 FSB data.

I hope this helps because I gotta go to report to a program for the unemployed. I actually have managed a $28 million Civil Engineering project, with an MSME and PhDEE and cannot find a job. Drove me bonkers until I started tuning into http://www.whatreallyhappened.com to overclock my mind.

 

I got my Athlon XP 2000+ rock stable at 166x11. Its at 1.84ghz right now. Vcore's at 1.65 and my timing's are pretty tight, runnin at 5-2-2-2.5, Spectek PC2700 ram, i think its some pretty generic stuff... I just know Corsair uses Spectek ram chips on their value select line... so im GUESSING its not too bad.. i think. But uhm... I've tried pushin it a little farther. Like i put the multi at 11.5, incrase the vcore up to 1.675, and it's a lot more unstable, i think it's unstable all the way up to 1.725. I'm really picky about it being rock stable. I've tinkered around with o/cing the ram. I was able to put it up to 190x10 with really low timings, i think 8-4-4-3. I was able to play UT2k4 for a good 20 mins.. never froze. Nothin. I don't really wanna sacrifice the timings though for an fsb that high. So from any of you really really experienced OCers here.... how much farther can i take my CPU speed up to without increasing the vcore more than 1.75. (Thats my limit). But uh... i wouldnt really mind goin a little higher than that if it means things would be more stable.
Hope this doesn't bother anybody... :S.

 

OH yeah... forgot to mention. The stepping code on my prc is AIUHB. Supposed to be a high end T-bred B right? Thats where it gets me.. its such a shitty overclocker for a tbred b....
So i dunno... how much further do you think i can take it up to without increasing the voltage TOO high. I only got a Volcano 9.. lol. Yeah.. sux i know.

 

Ok so, at 1812.14mhz stable, my 2000xp is running pretty smooth. However; I'm trying to push for >1900mhz . I've managed a relatively stable 1855 at 45-50c but still not happy with that, as the PCI is at 37, i'm not sure if my board will let me lock PCI/AGP. Any clues? (see signature for specs)

Also i'm looking at VCore tolerances, anyone know what the 2000 can be pushed to before obliterating itself?

 

Not sure about your motherboard question, but regards vcore, I have an IWill XP333-R that has a utility program that I run after my Win98SE has finishing booting. This program says that for my current session on an XP2000+ Palomino, the max vcore is 2.19v and the min is 1.79v with a current reading of 1.87v. Very important - I actually run underclocked at 12.5x100MHz (DDR333 running at DDR200), because things get unstable when I go to 133MHz.

Back to your question, I suspect the 2.19v and 1.79v are very brief spikes at poweron, although because the utility is a Windows program, maybe the readings actually did occur further along the timeline, like after booting up Windows. You can get away with spikes that which you cannot running at a constant voltage. I know an electrical engineer who designed a visitor counter for the McCormick Place convention center in Chicago. He used LEDs, and people were counted when they broke the beam. Now LEDs run at a nominal 10mA or so, but the light beam was too weak. So what he did was rapidly pulse them using 10A. That is, he overdrove the LEDs by a factor of 1,000, but since it was only brief pulses, the LEDs kept working.

Therefore, I would recommend you do NOT crank things up to the 2.19v for a constant voltage. From what I have read, each CPU is individual as far as their limit, and things get real scary around 2.0v. The thing is, you can never really know how much it can take until you destroy it.

 

The most it's been at is 1.92v but that wasn't really stable due to the PCI/AGP being too high! (A real bone of contention in my book as i can't get a stable overclock because of PCI/AGP limitations) I know this CPU will go a hell of a lot higher than 1815mhz, as i've seen them runnin 2000mhz+, with a high Vcore to stablise, however i don't know what board it was running, and if it had a PCI/AGP lock. Unless i figure this out soon, I think it may be time for a Barton!
Thanks for input though Yank

 

Thanks for the advice. I had a look through my bios to find

but couldn't find it. Could anyone recommened any freeware windows based overclocking program that i could download that works well? any suggestions would be great. Cheers!

 

am back if you look at first post its me, well i am in need of some help, firstly i have ordered this fan praetor Arctic Cooling Copper Silent 2l (blue Light) (maximum Support XP3400) is ok for overclocking my cpu?
2ndly its a thoroughbred xp2000,
i have a asrock k7vm4
http://www.asrock.com/product/product_k7vm4.htm
as you can see there it supports fsb 333. am gonna see if i can unlock the multipliers by FID Jumpers. whats ur view on that al i got 256mb ram 3200
and 512 3200 is on its way

 

ei, see my specs below? *points to sig* im planning to OC my athlon..ive done it once..and i reached abt 1.8GHz..Windows XP boots, but when i start playing games, d PC stucks up..the temps are around 30C idle..

i also would like to OC my GPU core...but when i do, and start any 3D app, d screen freezes, but i can stl manage to quit d app w/o rebooting..i could OC my Video memory quite well..i jz dont know what's causing the problem when i OC my GPU core...

 

I have a question what should i set my virtual memory if i have 756 megs of ram? and how do i over clock my 1.25 gig processor?

 

At the bottom you see all the specs about my PC. I can take it to 1900 with increased voltage but also in excess of heat (70-75 degrees) and I think that's just not worth the risk.
But I have a stable 1.80 (12x150) system now. Not quite sure whether the difference to the stock is worthy, but... Anyway, my EPOX MB reports the CPU now as 1900+ at 1800... Probably because of the decreased (-0.5) multiplier.
Any further suggestions?

 

Ok, here is a question
I just bought an ASUS A7N8X-E motherboard. supported FSB speeds from 200-400mhz. in my present motherboard <not worth mentioning> I have an Overclocked Athlon XP 2000+. the question is, if i get a good fan any one have a guess if i can run it on this board? only need to do it for 3 months. but would be very happy if i didnt cook the proccessor. Oh yea, the core is a Thoroughbred <cant remeber A or B> if it makes a difference ill redown CPUID