OC - shader manually or auto?

im trying to overclock right now and i was wondering if its better to adjust shaders seperately rather than have it move with the core? why wouldn't u do it seperately in the first place?

 

well i've been messing with it a lil more and noticed that when i adjust my shaders, i get the biggest hit on performance... why wouldn't you adjust shaders manually is the question...

stock = 4800 3dmarks
OC right now = 6300 3dmarks

i can now play cod4 all max settings flawless w00h00

 

Just out of curiousity....can you post the GPU and clock speeds you have it set at, also the settings you ran 3dmark at. I hope you aren't frying your card with that overclock.

 

i hope im not frying my card considering i have not even passed 65C

i have an 8700mGT and i have it at 760/1800/1000 (i think im stopping here) and i have ran crysis for about 2 hours now and NO crashing. NO artifacts. i just can't believe the boost i've gotten in 3dmarks either. AND MY TEMPERATURE AT THESE CLOCKS DONT GO ABOVE 58c YET!

 

ohw wow, so this will give a big performance hit?

then im definetly trying that out, u just have to make sure the 'link clocks' box is unchecked in rivatuner right?

 

right ;p
d

 

Hmmm, I never considered this. I'll have to give it a go. Since I can't get my 8600m GT GPU much above 610, perhaps my shader can go to 1300 or 1400. That'd be cool.

 

Just ran shader at 1400 with GPU @ 600, Mem @ 500 and gained about 220 3DMark06 points. Worth it? Dunno. If it translates into a few more fps in shader intensive games, then I'll run it. Otherwise, I won't bother.

 

in games like crysis and cod4 i found good fps increases by adjusting the shader (good meaning about 7-10 fps).

 

According to this performance scaling test done by AnandTech on 8800 GTX shader does improve performance but only a marginal gain, which to laptops can be considered signification I suppose. The gain range from 5-8 FPS gain in both Oblivion and Fear.

I have also tried running FEAR with/without shader overclocked separately from the core, and I found out that it doesn't really matter due to the biggest bottle neck of the 8600m GT DDR2 Edition is its Memory Bandwidth. I ran the internal bench test twice on each shader clock 600/1200/470 vs. 600 1300/470, not even 1 percent gain in performance. It might do more good for the GDDR3 owners who does not need to worry about bandwidth bottlenecking the card as much.

 

What are you using to OC is it the BIOS or Coolbit or sum thing like that, i need the bust

 

Just an unlocked RIvatuner, and ATITool beta to test for artifacts.

 

ATI? i don't use ATI i use Nvidia

 

Grrr....
Google Google, that's why search engine exists in the first place......

ATITool was originally designed to work with ATIcards, only but now in the newest version, the developer had implemented support for Nvidia card as well. I didn't use it to apply driver setting, I mentioned that I only used it to test artifact, but OC using rivatuner even though ATITool can OC nvidia card as well, just without the ability to set the shader clock seperately.

 

SORRY but that is why we have the forums for asking questions.

 

Man, I was tired the other day, sorry for being so aggressive, got an exam coming up soon.

Also, I found out OCing the shader alot higher than the 1:2 ratio with the core clock can really bring your SM 2 and SM3 score closer. For example, instead of close to 200 points difference between SM2 and SM3 on the 3dmark06 test, I now got 1872 for SM2, and 1803 for SM3.