3.0C @ 3.8 slow?

[daba]

I recently overclocked my 3.0C which was at 3.6 to 3.8 because of some new RAM I picked up. Originally at 3600Mhz I would get a horrendous Super pi score of 41 seconds with most processes disabled and super_pi.exe priority set to realtime. At 3800MHz I could run Super pi again and various other benchmarks and got the score to 39 seconds. This is very slow for a 3.6 and a 3.8. Since the system crashed with 2xPrime95 at 3800MHz and 3750MHz and wouldn't boot with 1.625Vcore, I am currently testing at 3700MHz with 1.6Vcore. Anyone know why the 3.6 is performing so poorly?

Here are Zeitgeist 2004.6's specs:

3.0C @ 3.7 (currently stress testing, usually at 3.6); Vcore @ 1.60 (usually at 1.575)
Asus P4P800 Deluxe
2x512 MB Kingston HyperX PC4000 @ 247FSB running 1:1
ThermalTake 480W Butterfly w/ Active PFC
SP-94 w/ ThermalTake 92mm smart case fan

At 3.6, I originally had some HardcoreCooling.com generic cas 2.0 PC3200 (2x512 MB) running 5:4. I thought this was the root of my performance shortcomings. What could it be?

Thanks in advance,
daba

 

[jpm804]

I'm running the same memory as you and I believe my super Pi score was 39's also..mines current running 250 fsb 1:1 3.75ghz .... I think the loose timings might be causing the those times...

 

[daba]

Well ****, 3700MHz didn't pass 1 hour of 2xP95. I wonder what the problem is?

Back to a stable 3600MHz until I can figure out a better solution. This baby has POST'd at 4050MHz.

jpm804: we have nearly the same computer.

 

[Sjaak]

My 3.0C SL6WU 30 cap ran 3.75Ghz (250fsb) fine for a month, then it started degrading. Could be your problem?

 

[Private Joker]

My 3.0C SL6WU 30 cap ran 3.75Ghz (250fsb) fine for a month, then it started degrading. Could be your problem?

@ what voltage, plz?

 

[Sjaak]

out-of-the-box, 3.75 ran good on 1.58v (bios 1.6), then after awhile it gave me sudden reboots, so i upped to 1.63 (bios 1.65). more reboots after a month, so i settled for 3.6 stock vcore which was stabile AFAIK.

 

[Phrenetical]

daba, my advice is to leave behind the 1:1 ratio, you can probably get your 3.0 stable at 4gig in the 5:4 ratio. I know the 1:1 is good but its hard to OC with that divider with this chip. Give 5:4 a go, if you dont like it switch back.

 

[KOXC2003]

The PI test seems to be better with slower memory at low latencys compared to faster memory with loose timings. Just to show you, My system gets 41 seconds, your system at 3.6ghz has almost 200 mhz higher clock speed and has a slightly lower FSB. But my ram is running at 195 mhz, but with 2-6-3-3.

 

[Sjaak]

daba, my advice is to leave behind the 1:1 ratio, you can probably get your 3.0 stable at 4gig in the 5:4 ratio. I know the 1:1 is good but its hard to OC with that divider with this chip. Give 5:4 a go, if you dont like it switch back.

Well, 4 Ghz on a 3.0C would need ALOT of luck

 

[lilneel12]

i have to put my vcore to 1.625 for it to be dual prime stable for 24h on 240fbs at 3.6, i got it to be stable at 247 at 1.675, but for some reason i did dual prime for 14hours and it was fine, then when i tried it again at the same settings it only went for 5 hours. so i just put it back to 3.6. i'll start ocing again when winter starts.

 

[daba]

The PI test seems to be better with slower memory at low latencys compared to faster memory with loose timings. Just to show you, My system gets 41 seconds, your system at 3.6ghz has almost 200 mhz higher clock speed and has a slightly lower FSB. But my ram is running at 195 mhz, but with 2-6-3-3.

The PC4000 is running at 240FSB with 3-4-4-8, but my old PC3200 ran at 2-3-3-6 at 192FSB at 3,6 GHz. Both get 42 seconds on SuperPi.

 

[daba]

Oh, I should add that RAM did play a crucial role in this situation. Back when I had PC3200 I could not run any program above 3,6 GHz. With this PC4000 not only can I get into XP at 3,8 GHz, I can also run various benchmarking software (but it doesn't survive dual prime).

 

[Atari]

SJAAK, thats pretty scarey man. Do you know why that happened? I mean I have my p4 2.4c @ 3.3 with 1.65 vcore (Love it this way since i get a 1.1ghz FSB). Still though kinda worries me that even at 1.6v your chip degraded :/.
It almost makes me want to build my old AMD system back up and slam in a barton if I see reboots on this box now.

 

[Sjaak]

SJAAK, thats pretty scarey man. Do you know why that happened? I mean I have my p4 2.4c @ 3.3 with 1.65 vcore (Love it this way since i get a 1.1ghz FSB). Still though kinda worries me that even at 1.6v your chip degraded :/.
It almost makes me want to build my old AMD system back up and slam in a barton if I see reboots on this box now.

yep, it wasnt fun Maybe it wasnt the vcore, but the Mhz? i dont know.

I thoroughly tested it at stock speed/voltage (4 days dual prime stabile), so it was safe giving to my dad in exchange for his 2.8. Never want it to happen again , but the flesh is weak

 

[vipergtsr1736]

I have a 1.8a (soon to be a 3.0c) and I used to be able to run it stable at 2.92 with STOCK vcore. Over time, it was less and less stable at lower and lower speeds. Ive changed ram and mobo since then. I can only run it at 2.15 stable. I NEVER changed the vcore and it was ALWAYS at 1.525 even though my chip degraded.

 

[aNTiChRisT]

I guess that physically, the transistors just degraded from such high frequencies.

~t0m