E6600@3.6 on 1.4125 Vcore Fails Orthos in Less than 5 Seconds

mawsoccer

Member
Jul 28, 2006
60
0
0
Here Are My Specs:

Asus P5B-Deluxe
Intel Core Duo e6600 L631A
eVGA 7600GT KO 256MB
2 x 1GB Crucial Ballistix DDR2-1000
Samsung 18X DVDRW Combo
Thermaltake Armor Balck Tower case w/ 25 cm Fan
OCZ GameXtreme 600W PSU

My E6600 is currently running at 3.6 (9 x 400). Temps are in the mid 50s C. Vcore is at 1.4125 in bios. CPUz says its at 1.3 something. Everything seems fine, but ...

IT FAILS ORTHOS EVERY TIME AT 1 SECOND.

It gives a message of:

FATAL ERROR: Rounding was 0.4999998097, expected less than 0.4.

It does the same thing on Prime 95.

WHAT IS WRONG HERE?

http://img137.imageshack.us/my.php?image=pic1yw0.jpg
 

Jumpem

Lifer
Sep 21, 2000
10,757
3
81
More vcore. Try 1.45; if it works back it down from there. Mine needs 1.4375 zat 3.6.

Set Orthos to Small FFT to only stress the CPU for now.
 

Ph1l

Junior Member
Dec 16, 2006
10
0
0
What type of stress are you giving?? Blend (CPU+RAM), only RAM, only CPU. First try RAM if it fails -> you got the answer, then CPU.
 

Jumpem

Lifer
Sep 21, 2000
10,757
3
81
Originally posted by: Ph1l
What type of stress are you giving?? Blend (CPU+RAM), only RAM, only CPU. First try RAM if it fails -> you got the answer, then CPU.

CPU first. 1000Mhz RAM running at 800 with conservative timings likely isn't the issue.
 

Ph1l

Junior Member
Dec 16, 2006
10
0
0
So as far as I understand, the ORTHOS alone CPU is passing, also same RAM test is passing, but when you hit "CPU&RAM" test it fails?? If it does it seems too small voltage on northbridge,
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
its something known as INSTABILITY ;)

try more voltage on board (vFSB, vMCH, vICH) and more voltage on cpu (1.45-1.5 max)
 

Jumpem

Lifer
Sep 21, 2000
10,757
3
81
Originally posted by: Jumpem
More vcore. Try 1.45; if it works back it down from there. Mine needs 1.4375 zat 3.6.

Set Orthos to Small FFT to only stress the CPU for now.

.
 

mawsoccer

Member
Jul 28, 2006
60
0
0
Uh yes, thats why I only have 45 posts. I wont be too new for long though, now that I have a computer that can actually be overclocked. My old one couldn't be overclocked because of the crappy components.
 

CurseTheSky

Diamond Member
Oct 21, 2006
5,401
2
0
Ok, here's a quick guide to overclocking. Just enough to get you in trouble:

DISCLAIMER - overclocking too much (raising voltages too high, leaving temperatures too high, etc.) can DAMAGE and DESTROY your system. Be careful.

Voltage is your friend, up to a certain point. More voltage gives components more headroom for overclocking. Generally, Core 2 Duo processors can stand somewhere around 1.55 volts safely - however, most people (myself included) prefer to keep it below 1.5, or 1.45 if possible. Other voltages (such as memory, FSB, etc.) can allow other components to overclock higher as well, but there is always the risk of burning them out.

Heat is your enemy. Heat causes components to work outside of their rated specs and wear down over time. Always do everything possible to keep your system cool. If it's plugged in next to a radiator, move it. If you have your fans all caked with dust, clean them. You should aim for 30s-40s idle CPU temp, and 40s-50s load (Orthos and similar programs). Higher voltages cause higher heat.

1.4125 Vcore is basically starving your processor. It's like trying to run a V8 Truck for 30 miles on a sixteenth of a tank. Try pushing it up to 1.43-1.44 or so. I bet you'll find Orthos running for quite a bit longer. If that doesn't work, lower your overclock and see where you can get it stable (for at LEAST an hour) at 1.4125 volts.
 

mawsoccer

Member
Jul 28, 2006
60
0
0
I have my bios vcore set at 1.4125 v.

I have Speedfan, CPUz, Asus PRobe II, and another Asus program open, and all 4 read my vcore at 1.37 and during orthos large ffts it droops even more to 1.34.

Is that reading not accurate? I guess there is some decent vDroop on this board. If these volts are true, does that mean I should keep my vcore under 1.45 after vDroop or before?
 

Avalon

Diamond Member
Jul 16, 2001
7,571
178
106
I'd determine first of all whether it's your CPU or RAM failing. Once you get that, I'd either back down your overclocks, or raise the voltage a little. If you're new to overclocking, going to 3.6 without any stability testing isn't a healthy way to start.
 

mawsoccer

Member
Jul 28, 2006
60
0
0
I am stability testing and I'm not a total noob any more. Please answer my post about the vcore reported by several programs.
 

CurseTheSky

Diamond Member
Oct 21, 2006
5,401
2
0
If all of the programs are reporting the voltage around that mark, it's probably accurate. The vdroop on some of the latest LGA 775 boards is pretty insane.

I would start by bringing it up to the 1.43 / 1.44 voltage mark in BIOS. Run Orthos again, and you should see improved stability - if it's failing in 5 seconds now, maybe it'll run for 5 minutes without failing. It'll still fail, but you know it's an improvement. If it's still failing, try 1.45 in BIOS. You should always do a stability test for 24 hours (or more) before you deem it a safe overclock. Even the one in my signature eventually failed (I tested it for 16 hours and thought it was all set). :(
 

n7

Elite Member
Jan 4, 2004
21,281
4
81
Two things are likely the cause of the instability:

1. Not enough vcore
2. Using the 1066 chipset strap (<400 FSB or lower)

The set vcore in the bios isn't necessarily what's being provided, as your programs indictate.
You've mentioned 1.34-1.37V, which is extremely unlikely to be enough for stability @ 3600 MHz

Also, the P5B Deluxe uses the 1066 strap till 401 FSB, which simply put, means there's a major amount of stress on things around the 370-400 FSB mark.
Jumping to 401 as well as increasing vcore may very well be needed for stability @ 3600+ MHz.
 

BoboKatt

Senior member
Nov 18, 2004
529
0
0
Yup all I gotta say is up that Vcore my friend. I was so excited to get my newer week 31A e6600 and my now replaced eVGA 680i board. I have the same RAM as the OP. Anyhow on stock voltage it could easily go past 3 Ghz... but it simply crapped out trying to go to the 3.4, 3.5 and 3.6 speeds.

Currently to be stable at slightly over 3.4 GHz, I have to use 1.54 V on the Vcore, which bites, and a few notches up on the other voltages. To get any higher I have to really push it further voltage wise, which I won?t do.

I have heard of folks doing insanely well with their e6600 in the 1.44V area and hitting 3.6 -- just not me. Hence as someone actually stated bump that Vcore up to say 1.52 or so and try Orthos. I will guarantee you will last way longer if not perfectly stable. Then, again as stated, start reducing your voltages slowly and work your way down if you insist on that current speed.

At idle with 1.54 V my system runs roughly 38c with my Tunic Tower and the fans going at roughly 40-50% using CoreTemp. Under load in Orthos if I let it go for a LONG time I can get it just past 50c but usually it's mid to high 40's which works for me. However, lesson learned here just to get that extra few MHz out of my CPU I really had to crank the voltage up. If I drop to say 3.38 or so I can actually reduce my Vcore tremendously, which in turn reduces my heat, which reduces the speed of which I would normally keep my fans at (which reduces noise), which reduces power consumption, which hopefully stresses less my system. I will be doing just that ? just need to find that sweet spot soon as I can dedicate a few hours.
 

Noubourne

Senior member
Dec 15, 2003
751
0
76
A digital multimeter is the only way to measure exactly what his board is providing. Software isn't usually a good measure in my experience. I just tested my rails using a multimeter on bootup, at idle, and full load, and they were all well within spec and didn't budge based on load. That is not what software was reporting. Not even close.

And I agree that a 1.4125v is not going to be enough to push a 1000Mhz+ overclock. You don't see many stable 3.4Ghz+ overclocks with less than 1.45v.
 

Mike89

Member
Aug 30, 2001
50
0
61
I don't see a MCH Voltage option on the P5B Deluxe. There is a ICH voltage reading.

I currently have my 6600 running at 3.2 gig at 9x356. I tried 8x400 but Windows freezes after about a minute. I have my Voltages set as follows:

Memory Voltage - 1.9 volts (2 gigs Corsair PC6400C4 at SPD)
CPU VCore Voltage - 1.25 volts
FSB Termination Voltage - Auto
NB VCore - Auto
SB VCore - Auto
ICH Chipset Voltage - Auto

I've heard talk about the strap. I assume this is the "Dram Frequency" setting? I have mine set to "Auto". I assume this automatically keeps the Ram speed the same as the FSB speed? CPUZ shows as 1/1.