Q8200 and Gigabyte EP45-UD3L

loafbred

Senior member
May 7, 2000
836
58
91
Just started overclocking today with the following hardware:

Antec mid-tower case with a three-speed 120mm fan in back (set to high speed), and a low speed 80mm in the door, also blowing out.

FSP Blue Storm II 500 watt psu, which is an improvement over the first Blue Storm 500.

Gigabyte EP45-UD3L and Q8200 cpu.

2x1gb Ballistix PC2-6400

Arctic Cooling Freezer Extreme cooler, connected directly to psu lead.

EVGA 8800 GTX.

Audigy 4 Value.

One HDD and one optical drive.

This is my backup PC which I built for the family kids to use, and to serve as backup when I'm working on my main PC. It previously had the same parts with the exception of the cpu and m/b, which were DFI 939 Expert and Opteron 180. I might run F@H on it, though probably not 24/7.
Edit: I forgot to mention that all power saving features are turned off.

Here's what I've tried so far:

7*485 (3.4 GHz) w/1.1625 (default) vcore, RAM 1:1 w/2.2v and 5-5-5-15 timings, Performance Enhance set to "Extreme", PCI-E locked at 100, LLC enabled, all other settings on "auto".
It ran a cpu stress test from a UBCD disk for a few minutes, so I let it boot into Win XP and started two instances of Orthos "blend", with affinity set for two cores each. One instance failed within five seconds.

I rebooted and changed only the cpu frequency and vcore, to 7*457 (3.2 GHz) w/1.188v, which failed after 56 minutes. The hottest core bumped 71*C for a few seconds. I tried Core Temp and RealTemp, and both gave identical readings.

This time I dropped cpu frequency to 7*443 (3.1 GHz), turned off LLC, increased vcore to 1.194, and reduced dimm voltage to 2.1v. The hottest core hit 67*C at times. That core averages 5 to 6 degrees higher than the others. It's still running after three hours.

I've done a couple of forum and web searches and there seems to be a lot of question about whether cpu temp readings are accurate on Q8200's. Any input on this, and your experiences with this motherboard would be greatly appreciated. I hate to spend hours trying dozens of voltage combinations if somebody can tell me what this board likes. BTW, this board has the passive aluminum chipset heatsinks.
 

loafbred

Senior member
May 7, 2000
836
58
91
Here's what I've settled on after much trial and error:

7*433 (3.03 GHz) w/1.1625v (default, set manually) and LLC enabled.

RAM 1:1, 5-5-5-15 w/1.80v.

Performance Enhance: "extreme".

All other settings as in first post.

Hottest core reached 67*C.

Memtest 86+ would fail on first pass at 7*429, and I tried it at 2.0, 2.1 and 2.2v dimm, with all other settings the same. I simply changed frequency from 429 to 433 and it ran two successful passes at 2.0v, so I dropped it to 1.8v and it ran two more passes. I then ran two instances or Orthos blend for four and a half hours, when I stopped it.

I think the cpu could run at least 3.3 GHz with default voltage *if* the temp didn't exceed 65*C.
 
Apr 20, 2008
10,067
990
126
with my q8200 i get it up to 2.9 and call it good. i got it to 3.1 a few times stable on diff settings but it was too hot. i'm at an actual 1.168v at my overclock.

i run my ram and fsb even and i'm set. its easy to overclock it, but i'm so far from mastering it, it isn't even funny.
 

loafbred

Senior member
May 7, 2000
836
58
91
I tried the retail cooler from my E6600 cpu to rule out problems with seating of the giant cooler I'm using, but it showed the same temp patterns on the cores, and tweaking with the two mounting screws on the big one didn't help any. I knew these cpu's had difficulty dissipating heat when overclocked, but I was surprised that default voltage would cause it to go to 70* so quickly. I'm still not tempted to go to watercooling, but I do keep thinking about it. :)
 
Apr 20, 2008
10,067
990
126
I would turn off performance enhance immediately. That could be why you are getting errors/high temps. Isn't that feature the bios' auto overclocker?
 

loafbred

Senior member
May 7, 2000
836
58
91
Reading this thread about the "performance enhance" options gave me a headache, so when I get time I'll try to run a memory benchmark with each setting to see what I get.
 

loafbred

Senior member
May 7, 2000
836
58
91
I ran Sandra memory bandwidth benchmark three times at each setting, waiting a couple of minutes after each reboot to prevent any activity from interfering:

Extreme: 8071, 8060, 8065

Turbo: 8087, 8080, 8072

Standard: 7664, 7658, 7666

I set it to "Turbo" and ran Memtest 86+ for 20 passes without error.