Thought I'd share my first volt mod experience with you guys.

Seeing all the great volt moded 8800 GT OC results on xtremesystems forum, I finally decided to jump into the hard volt mod train. I went to Jalan Ara Saturday, bought an hakko solder iron, DMM, all the hardware needed for the Vmod. I also went to modernopc to buy HR 03-GT to replace the stock cooler.

Everything wasn't smooth but at the end I finally managed to get it working.

Here is my max AT stable OC before the volt mod: 712/1728 @ 1.1V load temp 78*C


Here is my current max AT stable OC after the volt mod and the installation of the HR 03-GT 783/1890 @ 1.2V load temp 51*C


Picture of the moded card: (not very pretty but it does the job )


Although it's not AT stable, I can run 10 loops Crysis GPU benchmark and 2 loops 3dmark06 at 799/1998/975 @ 1.2V idle (still testing)

After testing max stable OC at 1.2V idle, I'll go up to 1.3V idle. So far my best 3dmark06 run is 13025. Not bad for a dual core

The HR-03 GT is pricey, but it is an awesome cooler. I've never reached higher than 54*C at 799/1998 and it was without A/C, after 1 hour of GPU testing almost non stop

Turns out pretty good for a 1st volt mod and considering that I had never soldered anything before.



This post has been edited by r4st4m4n: Feb 25 2008, 10:47 PM

I'm at work right now so I'll post my score tonight. I didn't post because I thought I'll wait once found max OC at 1.3V

Here is my 3dmark06 score at 792/1944/993 @ 1.2V idle



I've actually pass 3dmark06 at 799/1998 @ 1.2V but I saw afterwards that I left the memory at 950MHz so my score was slightly lower than this one.

I'll see tonight if I can continue to increase the clock at the same voltage but i think I'm reaching already the max for 1.2V

I've tried several bioses from different brands before the volt mod and all gave me the same results. Now I'm using my stock bios modified with 1.1V setting and fan increasing with the core temp ... which is now useless since I have done the vmod and change the stock heatsink

This post has been edited by r4st4m4n: Feb 26 2008, 03:09 PM

It is not not AT stable at this speed, but I did could run 10 loops of Crysis GPU benchmark followed by 2 loops of 3dmark06 without crashing/artefacting.

AT is fully stable (20 min) only at 783/1890.

I personally don't really care to be AT stable as long as I am stable in games. So my next step will be to game (mainly Crysis as it stresses the card like nothing else ) and if I can play 1-2 hours without a crash, I call it a day. Later I will also try other games. I think I will probably have to back up a bit from 799/1998 since I generally found that Crysis was stable more or less at the same speed as AT, but I don't really base my OC on AT.

My approach for the CPU is different, There I want to be 100% stable orthos, OCCT, ...

This post has been edited by r4st4m4n: Feb 26 2008, 03:45 PM

My idea was to do stability testing at 1.2, 1.25 then 1.3V (after the OCP will probably kicks in and I'm not sure I want to go higher anyways). It is to know how the card speeds scales with the voltage and if I don't see many benefit from going higher than 1.25V, I'll probably stay at this value for 24/7 use.

Now I'm also in a hurry to know how far it will go with 1.3V.
So maybe I'll skip 1.25V for the moment and I'll try 1.3V tonight and I'll post results here

Probably because of your CPU being clock higher than mine. I want to try later to OC my CPU higher for bench and see what kind of increase in score i can get.

Oh yeah, good find ...

I was starting to wonder how come my score was so low compared to him.

I just tried a run with my CPU at 3.33GHz and GPU at 842/2106/993 to see if it was making a big difference but was still under his score



@ben_panced: I tried 1.3V and seems to be AT stable around 828/1998 ... +.1V doesn't make a huge difference. But I will continue my own stability testing and see how far I can go in game ...

This post has been edited by r4st4m4n: Feb 26 2008, 11:47 PM

Maybe because is using a 19'' wide ? I never tried 3dmark 06 with a wide but if the screen doesn't support 1280*1024 maybe he can't use it.

1440x900 almost equivalent to 1280x1024 in term of numbers of pixels. As you can see your SM2.0/3.0 scores lower than mine now

This post has been edited by r4st4m4n: Feb 27 2008, 09:05 AM

Small update for ben.

Last night I played Crysis. It seems to be stable at 792/1944/950 @ 1.2V idle. At least it was for the whole 1+ hour I played it

More Crysis testing at 1.3V will come soon

This post has been edited by r4st4m4n: Feb 29 2008, 11:06 PM

My memory doesn't overclock very well. For the moment I leave it at 950MHz (1800MHz) in order to make sure that it won't give me instability. Once I found the max on the core and the shaders, I'll OC the memory.

But from the first result I have, it seems the memory is unstable over 993MHz (1986MHz). It is slightly under the result I had with the stock cooler (1000+MHz). It might be due to the 2 memory chips under the HR-03 GT heatpipes. The heatsink provided for them are tiny because the heatpipes are just on top of them, so maybe the cooling capacity is slightly worse than with the stock cooler. It doesn't really matter since the performance increase when OCing the memory by few MHz is barely noticeable.

800 MHz is definitely possible, the hr-03 GT can handle the heat easily. From my first testings, I think I should be able to do around 840MHz core and 2100 MHz shaders at 1.3V and it should be fully stable in games.

This post has been edited by r4st4m4n: Mar 1 2008, 01:54 PM