Overclocking Your Graphics Card - What's Your Personal Style and Methology?

[Big.Wayne]

Hello all,

I am interested in different peoples methods of overclocking their graphics card, like which part do you start with (GPU, MEM, SHADERS?) how much do you increment at a time and what simple tests do you do to confirm stability?

I've only really clocked RADEON GPU's in the past and that was easy peasy using ATI Tool or similar (find max GPU etc).

I am currently using RivaTuner to adjust the speeds of an nVidia 8800GT and running 3DMark 06 loop and also the Crysis benchmark, I find that any bad overclocks cause either of these two tests to stall,fail, go up in flames etc, however its pretty slow going. . .

So any tips or methods to get to the max quickly or is it a slow process where you just increase everything a little then go and do some gaming etc

Also I am interested in things like BIOS flashing and more risky 'tweaks'? Have you found in your experience that this helps achieve a better overall clock?

Thanks in advance

 

[no_1_dave]

I check here to see what other people are getting then whack my clocks straight to that. If it works I dont push it further, if it doesnt work I step back the clocks 5 at a time till it does lol
As long as I can get mine to what other people are managing then I dont bother pushing it further

 

[Big.Wayne]

As long as I can get mine to what other people are managing then I dont bother pushing it further

That puts a different perspective on the saying "Keeping up with the Jones"

So when you say 5MHz at a time, is that on everything (GPU/MEM/SHADER) all at the same time?

 

[Scougar]

Nvidia Cards> I think the order you want to do it BW is find the max clocks of the each of those in the following order.

Shaders (unlocked/unlinked). Not sure on limit, but has greatest impact apparently on these newer cards.
GPU Core (Might get to 700 might not, some go high as 800).
Memory (Not very good to overclock on the 8800gt's, might get 50-100mhz from what I've seen)

Ati cards> Check each individually to find max of each (core/mem), then take a slightly lower value on the memory and push the core up. I would say though that CORE has more effect on things with ATI cards though, so really should be trying to get max oc on clock first then memory.

Matthew

 

[Big.Wayne]

Thanks, so what quick tests do you guys do then after increasing the clocks? Is this something you do over the course of a day or do you chip away at it over a week etc?

Basically I am just trying to find out how different people approach the same subject kinda.

[edit] a quick stability test that gives a score would be cool, both 3DMark06 and Crysis Bench are a bit slow, also Crysis hardly shows any additional score, like percents of seconds, at least 3Dmark06 gives you some figures you can work with (shaders etc). Isn't there a cute little app that spits out a number?

 

[bobert50]

I gotta be honest i never do it carefully, i just set it roughly below what most people can achieve on forums with the same card and then play games.

It never crashes and i never push them any further for the extra 2% or whatever.

I just wacked the 8800gt i got to 700 + 1900 and seems 100% stable so that'll do me.

 

[shauncr91]

I do pretty much the same thing as Bobert50. Just see what speeds others are doing and then use the same ones. I then run something intensive such as Crysis or Bioshock for an hour or two to see if it throws up any errors, if it doesnt, I push it until it does

 

[geeza]

Select the highest setting someone else has got on the net then bench. If it works go higher, if it crashes go lower.

Rinse and repeat until it doesnt crash/give white/black flies.

I cant be bothered now because it only gives extra marks in 3dmark. In games you dont notice the difference (Same cant be said for clocking my X2 3800 from 2ghz to 2.8ghz)

 

[Big.Wayne]

In games you dont notice the difference (Same cant be said for clocking my X2 3800 from 2ghz to 2.8ghz)

I dunno about that, true enough if the difference is 135fps vs 170fps but if your really maxed out the difference between 30fps and 48fps would be more meaningful, thats the difference I had years back when playing Farcry, 30fps using a stock P4 and 9800Pro and 48fps when overclocking both CPU/GPU etc, really made all the difference.

That hasn't been the case recently though, these 8800GTs are really great cards right out the box!

Still we are overclockers and its hard to resist lol

[edit] still its interesting the feedback that has come in so far, seems people whack up to the big clocks straight away, I have been scratching my way up from stock!

 

[geeza]

[edit] still its interesting the feedback that has come in so far, seems people whack up to the big clocks straight away, I have been scratching my way up from stock!

Stop messing around. It wont melt, all it will do is restart and be set to stock speeds again

 

[Big.Wayne]

It wont melt, all it will do is restart and be set to stock speeds again

Geeza, have you ever *borked* a graphics card when you overclocked it?

I guess not. . .

 

[Scougar]

I think by the time I get a card, there has been a lot of testing already so the 'normal' limits are already known. I usually search for info first before trying to overclock anything so I will get a rough guide of what to expect.

Normally use ati tool/3mark and just game play to see what's up really.

Matthew

 

[setter]

like a few others have said i look at what sort of clocks people are achieving then i set mine slightly lower and work from there, i usually use 3dmark06 for stability testing and i seem to have hit a wall @785/1095/2000 with the g92 gts.

 

[synixx]

place my computer in a freezer, then i overclock the balls off of it, break the 3dmark world record while blowing my house up. voila.

no, i have only overclocked my cpu 300mhz and my graphics card just to play crysis better i'm not really an overclocker so i don't have my own teqnique.