Overclocking the 2400+ - Test Plan - Will this work?

[ajskydiver]

Just ordered the following and should arrive Wednesday of next week.

Motherboard:
Epox EP-8RDA+

Processor:
AMD Athlon XP 2400+

HSF:
Thermalright SLK-800A with 80x25mm Variable Speed YS-Tech 3-pin Case Fan 10.2~48.5CFM

Memory:
2x512 MB DDR PC2700 ? CL=2.5 ? Unbuffered ? Non-parity ? 6ns ? 2.5V ? 64Meg x 64
Edit: Crucial memory...which I just read may be a problem with this board at 166...can anyone confirm?

Background: The last processor I overclocked was a P5-166 to 200 MHz through the multiplier...have been using a PIII-800 for the last several years.

Assumptions/Questions:

1. Can I adjust the memory's settings independant of the FSB?
2. Can I adjust the multiplier without needing to unlock the processor?

I've read many, many of the overclocking threads in the forums and many of the techniques seem to have the same flaw - random adjustments until stability is compromised without knowing which part is failing.

With the above assumptions, is this test plan sound? Am I misunderstanding the fundamental process of overclocking the newer processors?

If I can isolate the maximum speed at which the processor will run without any instability through multiplier adjustments only and then isolate the memory's peak speed without stressing any other components, shouldn't I be able to deduce the fastest overclock without having to guess about the maximum FSB/Multiplier/Memory settings?

Test Plan for Overclocking the XP 2400+

Part 1
------
Isolate the processor's maximum, stable speed using multiplier adjustment only.

Part 2
------
Isolate the memory's maximum, stable speed using memory adjustments only (CL/timings).
(Memory should obviously be able to begin testing @ 166 with default settings.)

Part 3
------
Use the maximum FSB possible along with an appropriate multiplier to allow the memory to run asynchronously with the FSB. Knowing the processor's and memory's ceilings, I should be able to get this close the first time.

What do you guys think? Overthinking this or what?

~AJ

 

[pelikan]

I don't think you are overthinking it. A good game plan will drastically speed up the process and give the best results.

What brand of memory are you getting? These board are very picky. If you aren't sure go to AMDMB.com forums to check it out. Don't get generic memory. And if you want high fsb, consider pc3200 or better, unless you get pc2700 that overclocks well. 166MHz fsb will just leave you wanting more. I'm at 200MHz and am starting to consider 220 (will require VDD volt mod).

Make sure the 2400+ is an AIUHB. A 2100+ would probably be the same thing too. Week 0248 and newer seem to be best.

1. Yes you can, but running memory in sync will yield much faster results.
2. Yes you can, but the processor will most likely be unlocked on your board. You may need to flash to new bios.

Lots of good overclocking methods. Here's mine:

Set low multiplier (maybe 10).
Raise fsb to find limit of mb with relaxed memory timings (in 5MHz increments). Vcore up to 1.9V if necessary. Vdimm to 2.8 if necessary.
Tighten memory timings, shooting for Cas2-2-2-6. If you can get 10MHz fsb or more by running Cas 2.5-3-3-6(7) then its worth loose timings. Some of these nforce2 mb's top out at 185 fsb, others 205 fsb - luck of the draw.
Then raise multi to limit. May have to adjust voltages again.
Play some games and run 3DMark2001. If it'll do that run Prime95 overnight. Again may have to adjust voltages to get stable.
I usually compromise on cpu MHz by going as high as I can while not needing so much voltage that my fan gets too loud trying to stay under or around 50C under load.

You've got a great system on the way. It's gonna fly. Have fun!

 

[ajskydiver]

pelikan: Thanks for the great info! Regarding my 2 questions, I was only asking to determine each parts respective maximum tolerance...I will run the FSB and Memory at the same speed once finished testing...

"Part 3
------
Use the maximum FSB possible along with an appropriate multiplier to allow the memory to run synchronously with the FSB. "

Dang! I meant synchronously! So the above is correct...I misstated in the original post! Good catch, pelikan!

~AJ