Need Overclocking Help... Please!

[GML3G0]

thx, i'' try that. o and here's something interesting. i ran the 33 mem divider and put the timing back down to 2-2-2-5. my fsb is 240 and mem is running at 400. it passed super PI!!! It didn't do this before. making progress. will my performance take a hit from running this 6:5 divider? any recommendations?

 

[MTSteel]

Surprise surprise.....
I got home today and Prime95, the small fft(? not sure i typed that right) test is still running after 9 hours. That is a nice surprise, 2.5 hours at 1.475, 9 hours so far at 1.50. Guess it wanted some more juice. I still think I have a poor OC CPU though since it will not prime for more than a minute when I go over 2.25Gig
Ah well, 3500 speeds for the price of a 3000 is still not bad I will play some more.

 

[GML3G0]

wow 260 with the LDT set to 3 and multiplier to 6 and it drove the board crazy. error in 2 seconds. it must be chipset limitation. lol. i lowered it down to 250 and it seems to be running better. question - at 250 FSB, should the LDT be at x3 or x4? x3 would make the HT 750. x4 would make it 1000. 1000 is ok right? RAM is at 3-4-4-10. isolate and consolidate! lol.

EDIT: Well, with it set to 3x, it runs fine at 250MHz, but at 4x, it can barely do 245Mhz (the board, that is). Well I'll keep updating, but it seems like its doing fine so far with vcore at 1.450, RAM at 3-4-4-10, multiplier x7, 245 HTT 4x LDT.
I'm beginning to think you're right. I think it is the board. Grrr, I knew I should've gotten the MSI (too bad I couldn't find any in stock). What about north bridge, south bridge, and hypertransport voltage? should i touch those? i need all the help i can get. i'm open to suggestions. MTSteel, stop jacking my thread! lol j/k

EDIT 2: Crap, spoke too soon. Had to bring it down to 243... let's see how that goes.

EDIT 3: Grrr... I wish I had an Abit Fatal1ty AN8 or DFI Ultra-D right now... I better start looking for a job, lol. Just to throw a quick randow question out there, which of those two is better. The DFI I think clocks higher, but the Abit is more stable. Anyways back to topic.

EDIT 4: Score! Got it 266MHz FSB, lock set to 8:2:1 (266:66:33) as opposed to the agp lock at 66, and ram at 333 (which keeps RAM at 207). Now to increase multiplier and improve ram timings. maybe see if i can run 1:1. wish me luck!. ran super pi stable 100% completion.

EDIT 5: Got it running stable Super Pi with 9x multiplier, so 2403MHz. Now going to try to run RAM at 1:1 with as tight timings as I can. I tried 9.5x, but it crapped out. for now my goal has been reached, so i'll tweak it later to reach 2537MHz. Will run Prime95 torture test and MemTest86 before going any further to ensure stability.

EDIT 6: Will keep the RAM at 6:5 (218MHz - 2.5-2-2-8). Running this divider won't hurt performance right?

 

[MTSteel]

Ah crap, I noticed my RAM for this 11 hour success was at 100 More testing to do..... I have gone back to Auto for all except RAM is 2.5-3-3-7 1T and the Vcore is 1.425, will see how it goes.

 

[GML3G0]

bump, see my previous post for updates.

 

[GML3G0]

bump. crap. prime95 only ran for 2 hours stable. it passes super pi, but idk, should i be using the blend test, because i was using the max heat test.

 

[Regulator07]

Originally posted by: GML3G0
bump. crap. prime95 only ran for 2 hours stable. it passes super pi, but idk, should i be using the blend test, because i was using the max heat test.

is it still overclocked? and if you are failing the max heat test, then it is most likely cpu related as it tests memeory but not much. again go back to stock and run tests. you should be able to run all with no problem, if it does fail then you got a problem. overclock slowly, find out what is stable even if it is only a 100mhz difference, then start going higher, otherwise you are not going to be able to isolate the problem.

 

[GML3G0]

will using a 8:2:1 divider or a locked 66MHz divider make a difference in how high I can go? the fsb that is.

 

[GML3G0]

According to Zebo's OC thread, my max CPU is capable of at 10x is 238MHz (250x0.95).
If I run a lower multiplier, is it possible I can overclock higher? I will try it with multiplier at 9.5x and 9x after I finish running Prime95 to test the long run 10x stability.

 

[GML3G0]

according to my test results, this is the max my stuff can take
MotherBoard - 280+ MHz (didn't bother testing beyond this)
CPU -
10x - 238 MHz FSB // stays the same
9.5x - 247 MHz FSB // stays the same
9x - 261 MHz FSB // stays the same

so the 10x would give me best performance, correct? any other tweaks you recommend I should try?

EDIT: crap, scrap that forgot to set LDT to 3x, that's why it failed. restesting...

EDIT 2: updates posted. results stayed the same, even when LDT was lowered to 3x. strange.

Load temps are 57-58. I should be receiving my XP-90 w/ AS5 any day now. You think that will imprive my roadblock at 2384 MHz?

I see tons of people with 3200+ that reach these insane overclocks... wtf, why can't my CPU do that. It's not the board... I see people with 3000+ that overclock better than mine...
vcore is at 1.5v
locks are set at 66MHz (other options include locked at 74MHz, 6:2:1, and 8:2:1)
RAM voltage is at 2.8v 238MHz 1:1 2.5-2-2-6
LDT is x4

need more overclock... grrr, no way the max i can hit is 238MHz... :sigh: im stubborn.

EDIT: omg! now i find out that the problem was that the av8 has stability problems with the LDT at 3x. it needs to be setr to 2x... grrr.... here I go all over again... new tests!... lol

Another EDIT: Well, seems pretty stable at 267 x9 = 2403MHz. The 66MHz AGP lock sucks. That's what the problem was with the max CPU. On top of that, I was using a x3 LDT. Those have stability problems. The x2 allows me to overclock higher and is more stable.