Overclocking the Q8200

[Evilsizer]

hmm i have never seen that benchmark before. it really does depend on the app though steven, 10% would be the lowest. the SSE 4.1 instructions on 45nm cpus speeds up other taskes like AV encoding/transcoding. that is if the program is coded to use SSE4.1 instruction set.

 

[SamSaveMax]

Please keep it going.....your progress of overclocking Q8200
I bought my Q8200 + Mobo combo ECS G33 a month ago for dirt cheap $129.
(I had no intention of buying it, but it was a half day System Builder's special price which I could not pass.)
Had it overclocked @ 2.82 stable just by adjusting FSB. Couldn't get any further due to limitations of ECS mobo.
Still trying to shop for a newer/clockable motherboard for Q8200. And your Gigabyte motherboard interested me alot.
I heard that Q8200 could do around 3.2~3.4GHz.

http://i37.tinypic.com/6st99t.jpg

 

[Evilsizer]

your Q8200 is running on stock voltage and close to 3ghz. looks like you might be able to get 3ghz on stock cpu-v stable and possibly 3.2ghz with a small bump.

steve,
have you tried ocing it yet or whats going on with the box?

 

[SamSaveMax]

I wish that could bump the Vcore voltage on ECS mobo. It doesn't allow.
It's a very limited board for overclocking.
Reviews of the GIGABYTE GA-EP45-UD3P are very good.
I'll probably get it soon

 

[karkas]

voltage pad mod?

 

[Steven-1979]

Evilsizer - I'm still putting the case together. I just got the corsair psu today (last component I was waiting on) ... cut out the hexagonal mesh in my case (going to use a filter infront of intake fans):

http://img228.imageshack.us/img228/1346/1000792tk0.jpg

http://img409.imageshack.us/img409/1701/1000809rx5.jpg

Looks ugly as hell, but the front will be covered by the front cover and the back is facing a wall. I think for the airflow it obstructed, it was worth removing that mesh. I should have everything installed by tomorrow and beging OC'ing.

I wanted to transfer a few components from my current main rig to this one, and also switch the 585w psu into that one since I got the new corsair, and opened it up. While open, I decided to do a voltmod and have it OC'ed to 2.9ghz appearing stable right now (will know when I run prime 95 overnight).

SamSaveMax, my E2200 is on an ECS mobo and I've had problems OC'ing with it also since you can't adjust vcore. I just did a voltmod and it looks stable. I'll post a thread once I check everything out and it's stable with detailed instructions how to do it. It's pretty simple. But it's a pain removing the cpu everytime you want to adjust the vcore

 

[SamSaveMax]

kakaa What tool did you use to chew out the hole?
On 45nm, I don't think you would need that much of case fans.

My current cheap build Q8200
I removed the tiny stock fan off the graphic card 7300GT as well.
Still cool and works fine as with assist from the 120mm fan to cool the whole area.


I'm currently in process of changing newer computer case + power supply + motherboard

 

[Steven-1979]

I began doing some baseline tests and I was surprised by the temps. Just general use:

Core temp for 0,2 = 45 C , for 1,3 = 40 C

After about 1h on 100% with prime95: 0,2 = 60 C, 1,3 = 55 C

I'm curious why the normal use temps are a little high ... or is that normal for this cpu? I thought they ran cooler, but my E2200 runs at 30 C during general use. Also the different temps for the cores. It looks like cores 0,2 are running hotter. Could that have to do with the way I oriented my heatsink? I oriented it sideways so the fan is blowing right into the exhaust: http://www.ocforums.com/showthread.php?t=587333

SamSaveMax, Nice looking rig ... I think we have the same heatsink. How did you suspend that side fan? I'm not sure what that tool is called. It's this thing:

 

[SamSaveMax]

I used a 120mm fan which has only 1200rpm max. but with nice output amount of air.
That way it is quiete @ 19dba
Screwed to the frame of the computer case for the main support and screwed the other end to the fan (on the other of the fan which is not seen from the picture)
....then suspended with a piece of rope used from window blind.

I notice most of stock fan from the graphic card can be kind of noisy after awhile, so I removed mine as well. Its worked for me fine with a large fan blowing air across the board to cool just about everything including RAMs. Very quiete system.

 

[Steven-1979]

I'm new to OC'ing, so I'm pretty sure I am not setting the other options correctly when adjusting vcore. Here are my test results so far. I did prime95 for 30 minutes just so I'm not bogged down all day, but get a general sense for stability. I plan to run it for 12h on the last setting that worked while I sleep. Note, as soon as I turn off the torture test, the multiplier goes down to 6. But when I'm running load, the multiplier is 7:

http://img117.imageshack.us/img117/838/x7ji6.gif

So in some of the pics it says 6 because I took them with p95 stopped to show no errors, but on load it's 7.

I manually set the RAM timings and volts according to specs: 5-5-5-15 2.1v in bios


*************************************************
*************************************************

400 x 7 (2.8GHz) / Vcore: auto

Ran 30 minutes on prime95 stable.
Average Temps per core:
0: 63
1: 56
2: 56
3: 60

Ave vcore: 1.26
Ave heatsink fan speed: 1190rpm
Ave RAM fans speed: 6253rpm

http://img242.imageshack.us/img242/1247/400x7hc9.gif

500 x 7 (3.5GHz) / Vcore: auto

didn't even get to bios.

450 x 7 (3.15GHz) / Vcore: 1.3v

Gets past bios but doesn't load windows. Am pretty sure this is because all I changed was vcore (read msg at end of thread).

450 x 7 (3.15GHz) / Vcore: 1.35v

Gets past bios got BSOD while windows loaded. Am pretty sure this is because all I changed was vcore (read msg at end of thread).

420 x 7 (2.94GHz) / Vcore: auto

Ran 30 minutes on prime95 stable.
Average Temps per core:
0: 63
1: 57
2: 56
3: 60

Ave vcore: 1.27
Ave heatsink fan speed: 1156rpm
Ave RAM fans speed: 6252rpm

http://img117.imageshack.us/img117/1975/420x7nd4.gif

450 x 7 (3.15GHz) / Vcore: 1.325v

Gets past bios got BSOD while windows loaded.

430 x 7 (3.01GHz) / Vcore: auto

Ran 30 minutes on prime95 stable.
Average Temps per core:
0: 66
1: 66
2: 63
3: 60

Ave vcore: 1.26
Ave heatsink fan speed: 1212rpm
Ave RAM fans speed: 6257rpm

http://img377.imageshack.us/img377/1705/430x7ev8.gif

440 x 7 (3.08GHz) / Vcore: auto

Ran 30 minutes on prime95 stable.
Average Temps per core:
0: 66
1: 59
2: 64
3: 60

Ave vcore: 1.31
Ave heatsink fan speed: 1232rpm
Ave RAM fans speed: 6258rpm

http://img377.imageshack.us/img377/3230/440x7dk3.gif

450 x 7 (3.15GHz) / Vcore: auto

Gets past bios got BSOD while windows loaded.

*************************************************
*************************************************


I think the problem I'm having is what when I edit the vcore in bios, I just edit the vcore setting but leave everything else on auto since I'm not sure what to put. I think 3.08GHz with the voltages set on auto is as far as it will go. These are the settings I'm not sure what to enter:

http://img117.imageshack.us/img117/3560/biosbz6.jpg

The core voltage shows slightly higher in cpuID while prime95 is off than when it's running it's stress test.

 

[Evilsizer]

what memory ratio or multi are you using?

 

[Steven-1979]

It defaulted to 2.66 and I left it at that:

http://img376.imageshack.us/img376/4889/memlw3.jpg

 

[Steven-1979]

I'm a retard, lol. Changed the memory multiplier from 2.66C to 2.0D and it looks like it's good ... also, vcore dropped significantly.


450 x 7 (3.15GHz) / Vcore: auto

Ran 20 minutes on prime95 stable.
Super Pi stable (do people use super pi for stability of just to test speed as a benchmark?)
Average Temps per core:
0: 63
1: 55
2: 57
3: 60

Ave vcore: 1.17
Ave heatsink fan speed: 1164rpm
Ave RAM fans speed: 6246rpm

http://img117.imageshack.us/img117/1563/450x7ou6.gif
(note- super pi done as prime95 was running, so times are slower)

 

[SamSaveMax]

3.15GHz that's very good.

Steven,
Have you tried leave everything on AUTO except for RAMs settings.....and just push FSB frequency. If needed, just increase voltage of vcore 1.35; 1.36; or 1.37

I notice that you have ram's fans. At 6200rpm is that loud?

 

[Steven-1979]

I'm up to 3.5 ghz Just running some prime95 tests right now before posting next batch of results. I did try the vcores on auto, but it took vcore up to 1.408v, so I started manually reducing. Right now I am running a stability test on 1.375v @ 500x7. Looks stable so far.

The ram fans are actually very quiet much to my surprise. The only 2 fans I can really hear deeply are the rear 120mm fan (since I removed the hexagonal mesh, and it is right at my head level) and the side vent fan. The case is pretty thick compared to my other one, and does a good job of dampening most of the noise from inside.

 

[Steven-1979]

500 * 7 (3.50GHz) / Vcore: auto

When loaded, the vcore was 1.408v. A lot of people said not to go over 1.4v so after the stress test, I began lowering the voltage. Stress test ran at lower voltage due to vdroop so I figured I was safe that close.

30 minutes torture test

core 0: 72
1: 64
2: 66
3: 69

Ave vcore: 1.35

ave fan speed 1356
ave ram fan speed 6247

http://img149.imageshack.us/img149/8387/500x7200dyd4.gif


500 * 7 (3.50GHz) / Vcore: 1.39375v

20 minutes torture test

core 0: 69.8
1: 66
2: 62
3: 62

Ave vcore: 1.31

ave fan speed 1356
ave ram fan speed 6247

http://img149.imageshack.us/img149/7959/500x7200d139gp4.gif

550 * 7 (3.85GHz) / Vcore: 1.39375v

Crashes before bios loads

533 * 7 (3.73GHz) / Vcore: 1.39375v

Crashes before bios loads

525 * 7 (3.68GHz) / Vcore: 1.39375v

Crashes as windows loads

520 * 7 (3.64GHz) / Vcore: 1.39375v

Crashes as windows loads

510 * 7 (3.57GHz) / Vcore: 1.39375v

Everything loads, can use cpu, but screen freezes after about 10 minutes of prime95

500 * 7 (3.50GHz) / Vcore: 1.375v


30 minutes torture test

core 0: 71
1: 63
2: 64
3: 67

Ave vcore: 1.29

ave fan speed 1325
ave ram fan speed 6282

http://img149.imageshack.us/img149/2600/500x7200d1375kk2.gif

500 * 7 (3.50GHz) / Vcore: 1.3375v
*prime95 error in one of the cores < 1 minute. Everything else seems fine. System does not crash.

500 * 7 (3.50GHz) / Vcore: 1.35v
*prime95 error in one of the cores < 1 minute. Everything else seems fine. System does not crash.

500 * 7 (3.50GHz) / Vcore: 1.356v
*prime95 error in one of the cores < 1 minute. Everything else seems fine. System does not crash.

500 * 7 (3.50GHz) / Vcore: 1.362v
*prime95 error in one of the cores < 1 minute. Everything else seems fine. System does not crash.

500 * 7 (3.50GHz) / Vcore: 1.368v
*prime95 error in one of the cores < 1 minute. Everything else seems fine. System does not crash.



*****************************
*****************************

It looks right now like the stable setting is 500 x 7 @ 1.375v. I'm going to run this setting through prime 95 as I sleep and see how it does after 12 hours. I'm pretty sure I can get even higher once I start experimenting with the other settings since right now all I am changing is fsb and vcore and leaving everything else on auto.

 

[SamSaveMax]

Quads cores hardly do 500.
I would suggest you drop it down to around 480~490x7 and probably be more stable.

You make me really want to order this Gigabyte board now.

 

[$SOLID$Necro]

See my post in this thread, it has a link right at the top to a review of that board with some good tips for OC.

http://www.ocforums.com/showthread.php?t=586875

The reason you see the multi drop at idle to 6, is the Intel power saving features that reduce voltage and CPU speed at idle.

You can turn it off in the the bios by disableing EIST (Enhanced Intel Speed Step)

Your temps may be on the high side due to an improperly seated HSF, the one you have is known to be a bit tricky. And there is a prefered method for applying the thermal paste on it too, try searching the forums here for more info..there are a few threads on it.

Are you using AS-5 paste or equivalent? The paste that came with your HSF is probably not real good.

 

[Gillbot]

You shouldn't set the voltages to "Auto"
If you want it to run it's stock setting, manually enter that voltage in the field. Some boards adjust voltages to the level they think your component needs when you use the auto setting.

 

[Steven-1979]

500 * 7 (3.50GHz) / Vcore: 1.375v
Prime 95 gave an error in core 2 after about 4 hours of blend test.

500 * 7 (3.50GHz) / Vcore: 1.39375v

Prim95 stable for about 13 hours.

core 0: 70.7
1: 63
2: 66
3: 64

Ave vcore: 1.31

ave fan speed 1356
ave ram fan speed 6238

Note- all settings were auto except vcore and RAM settings.

See my post in this thread, it has a link right at the top to a review of that board with some good tips for OC.

http://www.ocforums.com/showthread.php?t=586875

The reason you see the multi drop at idle to 6, is the Intel power saving features that reduce voltage and CPU speed at idle.

You can turn it off in the the bios by disableing EIST (Enhanced Intel Speed Step)

Your temps may be on the high side due to an improperly seated HSF, the one you have is known to be a bit tricky. And there is a prefered method for applying the thermal paste on it too, try searching the forums here for more info..there are a few threads on it.

Are you using AS-5 paste or equivalent? The paste that came with your HSF is probably not real good.

Great link! Thanks for the info, man. One thing that scared me reading that article was:

"BUT, against our better judgement we allowed all of the stress tests to run even though the auto settings of the board set VTT and PLL voltages at very dangerous levels, 1.60v and 1.85v respectively. We should add here that it has been commonly shared and concluded in the PC industry that 1.4v is about the maximum safe 24/7 voltage for VTT. However, it has come to our attention that some motherboard manufacturers have changed this limit to closer to 1.9v VTT. We will ask Gigabyte for an official answer and discuss this in the comment thread."

It did the same thing to me. I had been adjusting the vcore manually, but left everything else on auto. I went to check and what do you know, my VTT and PLL are the 1.6 and 1.85 they describe as dangerous. Kind of freaked me out because I never even noticed them.

Something else about that article, they seem to be pushing the vcore pretty high on their 45nm chip? I thought 1.4v was the max people should be using on 45 nm chips ... or is that just for the quads?

In terms of the heatsink, I did search before getting it and followed the method of applying a thin strip to each of the copper tubes. But I think the thermal paste I got may not have been the best kind. I had ordered Artic Silver from yesbuy (the only component I didn't get from NewEgg) and it didn't come with the rest. Infact, I STILL don't have it! So I used this from NewEgg: http://www.newegg.com/Product/Product.aspx?Item=N82E16835129002

Once the Arctic Silver comes in, I'm going to reapply it and also look into the way the heatsink is situated. It was a pain to get on and I may have smudged the thermal paste and left some bubbles. I'll have to check. Until then, I'm going to keep the settings on lower voltages and play around with the lower voltages to see what I can get there. Once the new stuff comes in, I'll go back to pushing the higher limits.

Just out of curiousity, does anyone monitor the temps of their NB? If so how?