my 1.6A seems to have hit the wall...but why?

[SmackdownHotel]

Let me start by saying that I've managed a stable o/c of 2133mhz (133fsb) at 1.6v on an Asus P4B266-C. Whenever you can get a 533mhz overclock, I'd call that a pretty damn good day, but here's what's strange. Anything over 133fsb, and it becomes totally unstable. I've increased both the CPU voltage and the RAM voltage as high as they'll go (I get weird beeps at 1.7v), and yet no dice. And by unstable I mean it will fail a Prime95 torture test within minutes. Have I merely hit the ceiling with my chip, or is there something else I'm missing?

My first guess would be that the RAM is crapping out over 133fsb, but would that manifest itself in Prime95?

 

[DarkOrca]

It may be the case that the CPU has hit its limit, just remember not all CPUs have the same "potential" to get insanely o/c-ed.

1.6A@2.1Ghz is already a 33% o/c! not bad at all!

if u still wanna push the limit, maybe u can try upping the voltage more (u have to set the jumper for "extra" high voltage first, see manual). But I wouldn't recommend to do so as 1.7V+ seems a bit high to me.

 

[Jwyatt]

What brand ram is it? Im not familiar with your board, but you might try to run the memory at 200fsb if its available. Then up the fsb to see if it still works right. That way the ram is running at a much lower speed and you can see if its the cpu or the ram.

 

[wildting88]

i am very new to this whole overclocking thing. can you please help me. i think i have the same parts as you. here's what i have and here's what i've done.

Asus P4B266-C Motherboard (bought at newegg.com)
Intel P4 1.6A Chip (bought at newegg.com)
1 stick of 512 DDR SDRAM (bought at Frys)
1 Barracuda II 20 Gig HD
1 Maxtor 45 Gig HD
1 Promise ATA100 Card (both HDs are connected to the promise card)
Geforce 3 Ti200 Video
no sound right now
ATI TV Wonder PCI Television Card
Dazzle Firewire Card
A really fun casing that has a window and blue light.

OK... my friend who knows some stuff about overclocking overclocked it to 133FSB. Everything else was left default. I ran Prime95 and it only lasted for like 10 minutes. I figured that might be because I was also multi-tasking. He told me to not use my computer and then test it. So after that... i got this error after 11 minutes: "FATAL ERROR rounding was .4999482769, expected less than .4. Hardware failure detected. Consult stress.txt."

What does that mean?

Anyway. I am really asking what to do now. It seems that I cannot even overclock my computer to 133 (which is supposed to be standard for this board I have read). Do I increase the voltage and or decrease those numbers that everyone says 2-2-2-2-5. Can someone please PM me or just reply back to this thread letting me know what settings they have if they have this same board and chip? OR... do you think my RAM could be bad or something?

Thanks in advance.

Jason

ALSO, if you have another testing program, that would be great.

i posted this in another place too.

 

[TranceNation]

could be your ram you bought at frys. get pc2700

 

[wildting88]

yeah... it's currently one stick of 512 PC2100 Frys RAM. DDR.

so you think getting the PC2700 will be better? do you have a specific recommendation?

Thanks

 

[Nate420]

SmackDown.....

I realy doubt that your RAM is the problem, unless it's super cheap K-Mart RAM. 133FSB would only run the RAM at 266Mhz, which most any DDR should handle. One easy thing to do would be to set the Memory Ratio to 3:4 and run the FSB at 132 and see what happens.(132*4/3=176)(176*2=352Mhz) If it runs fine then you just eliminated your ram from being the problem. Also, set the Turbo1 option in the BIOS to keep your PCI/AGP bus from going way out of spec. I wouldn't recommend going past 1.65V(just my opinion of course) I have the P4B266(not the C revision)Crucial PC2100 256M@312Mhz 2.6V and P41.6A@2.5, 1.6V.



Wildting.....


I'd try upping the Vcore to 1.55V and try again. If it still fails, up the Ram to 2.6V and try again. With your current setting, does your rig run ok other than Prime95 failing? Can you game without problems? I recommend that everyone set the Turbo1 option in BIOS.

 

[wildting88]

yeah... with my current settings... all the programs run fine and everything. it's just that Prime95 crashes after about 11 minutes (done this 4 times now). I will try your recommendations and let you know what happens. do you know if maybe i need new RAM? i got mine at Fry's.

Thanks!
Jason

 

[Nate420]

Well, if you realy wanna put your RAM to the test, download MemTest86 and let it run. You'll need to copy the app to a floppy and run the app from DOS.

 

[wildting88]

i did increase my VCore to 1.55. my computer booted to WinXP Pro perfectly fine but when i run Prime95, i still fail after about 10 minutes. all my programs seem to be running perfectly. i might go to fry today and return the RAM and buy a PC2700 one ... do you think that would help solve my problem? or does sometimes Prime95 not work on computers?


jason

 

[Boogak]

In my experience, Prime95's stress test sufficiently stresses your cpu and memory to the point where it finds errors when every other app you run seems fine. Try someone else's suggestion and download MemTest86, at least that way you can narrow it down to whether it's your memory or not.

 

[Platinum321]

in my experience fry's ram is cheap and not overclockable at all.. i thought i posted a message earlier.. ahh well.. good luck...

 

[o1die]

Have you increased the memory voltage? My elixyr pc2100 runs with the 4:5 ratio (pc2700) when I increase the voltage to +.2. My old micron won't even post with the 4:5 ratio set. Just the luck of the draw I guess.

 

[SmackdownHotel]

Well I tried the 3:4 option and the memory seems fine at 176mhz, so I guess my chip just doesn't like being pushed that much.

Does anybody know if a future BIOS version will allow the 3:4 option for FSBs over 132? And also, is it me or does Asus seem to abandon the -C versions of their boards?

 

[WarCon]

Yes, I have the -C version and that is exactly how it seems...........

I think if your processor starts to push 50C at around 133fsb, then your not going to go much higher. If it stays cooler, due to really good case cooling or a better heatsink, then you should be able to push it higher and keep it stable. If it was me, I would probably run it at 132 and 3:4 timing. Mine ran very well at that. But I got a better powersupply and am now at 148fsb @ 2 2 2 7 timing. (I read that the last value gains you little or nothing on speed, but can be the source of instability if set too low, so I don't.)

 

[SmackdownHotel]

Yeah, since I can only hit 133 anyways, I'm just gonna run it at 132 with the 3:4 option and get the extra memory bandwith instead.

 

[ElusivePeon]

if it fails within minutes its more likely you are pushing your chip too hard, anything like about 5+ hours is usually heat related.

 

[Def]

More voltage. That's the answer.

If you actually look. the core voltage of Nwoods are actually lower than the set value in the BIOS. At least it is on my EPoX 4BDA(i845).

I've set it up as high as 1.8V, actual was around 1.7-1.74V. Temps got too toasty with the stock heatsink, but the chip should be able to take that easy.

Don't be afraid to give it .025V more if you're failing Prime. Even running at 1.9V for a little while won't permanently damage your chip, and will let you know if yours is a dog or some other component is to blame.