GTX285 various mods.. which may work for GTX260/280/295 also

[Demo]

Is it possible to remove the heatsink shroud w/o removing the heatsink it self?


edit:

Decided to undo the vmod as i was only getting 1.20vGPU ( from a stock of 1.14 ) and it did nothing for my shader clocks still stuck at 1566, didnt bother checking if the core can go past 720Mhz, but even if it does make 750 performance increase would be minimal.

[Blkout]

Is it possible to remove the heatsink shroud w/o removing the heatsink it self?


edit:

Decided to undo the vmod as i was only getting 1.20vGPU ( from a stock of 1.14 ) and it did nothing for my shader clocks still stuck at 1566, didnt bother checking if the core can go past 720Mhz, but even if it does make 750 performance increase would be minimal.

Maybe minimal but you'll gain more from core clock speed than shader clock speed.

[Blkout]

the superstitious 3Dm*rk06 is not so reliable on gt200 series.
i've got 17000+ on my old 8800GTS g92@vmod 890/2200/1080
but also bout 17000+ on gtx285@vmod 780/1700/1460
but real time game play 285 get much smoother than gts.

to NapalmV5
i actually slam a 500vr few days ago ...but dun really push the vGPU...bro, do you have any idea how much this card can take? not so serious ...24/7 capable ...
on water now load temp at 1.25vGPU(solder bridged 1/2/3/4/11/12) = 50C

Actually if you just change the resolution in 3DMARK06 from the default 1280x1024 to 1920x1200 and/or add 4xAA, you'll see the 285 easily beating the Hell out of the 8800GTS. The benchmark is reliable, but it's outdated if you if you use the default resolution and settings. This generation of cards is geared towards 1680x1050 and higher resolutions.

[Demo]

Actually if you just change the resolution in 3DMARK06 from the default 1280x1024 to 1920x1200 and/or add 4xAA, you'll see the 285 easily beating the Hell out of the 8800GTS. The benchmark is reliable, but it's outdated if you if you use the default resolution and settings. This generation of cards is geared towards 1680x1050 and higher resolutions.

That is what i noticed when running 1680x1050 with 8xAA, the 8800GT would crawl along and i didnt even bother to finish the slide show.

Maybe minimal but you'll gain more from core clock speed than shader clock speed.

At 1.20v i recon i cant get over 780Mhz core.. so not really worth it.


EDIT:

Can anyone check if the fan speed on the 285 is 25% or 40% minimum by default/auto.

[hehee]

Actually if you just change the resolution in 3DMARK06 from the default 1280x1024 to 1920x1200 and/or add 4xAA, you'll see the 285 easily beating the Hell out of the 8800GTS. The benchmark is reliable, but it's outdated if you if you use the default resolution and settings. This generation of cards is geared towards 1680x1050 and higher resolutions.

i see..
cause my old records are defaults....now no chance to try again on old cards.
with try to 19x12 future...

[ReverendMaynard]

the superstitious 3Dm*rk06 is not so reliable on gt200 series.
i've got 17000+ on my old 8800GTS g92@vmod 890/2200/1080
but also bout 17000+ on gtx285@vmod 780/1700/1460
but real time game play 285 get much smoother than gts.

are you high? Get a cpu that doesn't choke the snot out of your card and realize it's actual potential lol.

[Demo]

@ReverendMaynard

Can you do some bench runs for me on that system of yours?

Can you bench it at stock videocard clocks and then at 720/1566/1332, id like to compare how much of a gap the 3.2Ghz Q6600 has on that i7 4.3Ghz.

Thanks in advance.

[CryptiK]

Can anyone check if the fan speed on the 285 is 25% or 40% minimum by default/auto.

On auto it sits at 40% until it heats up and scales up as necessary.

are you high? Get a cpu that doesn't choke the snot out of your card and realize it's actual potential lol.

Absolutely true. Get a wolfdale up to 4.3GHz+ and your ram at ~575 - 600 MHz and you will see a large difference between the cards. A quad or i7 around the same speed with give you even higher 06/vantage scores.

[Demo]

Heh, sucks that my Q6600 only goes to 3.4Ghz at best.

[hehee]

are you high? Get a cpu that doesn't choke the snot out of your card and realize it's actual potential lol.

yo...tats alot for me..

Heh, sucks that my Q6600 only goes to 3.4Ghz at best.

same here....i7 next move?

to NapalmV5
i turn down 500VR to bout 200ohm ....any further than 1.26vGPU protection kicks in ...any clue?max i can get 1.26v sad
mod 1/2/3/4/11/12 bridged / 500VR at the back /no pencil.

[dengyong]

Heh, sucks that my Q6600 only goes to 3.4Ghz at best.

Might be a fsb hole, have you tried going strait to 3.7or 3.8Ghz ?

[Demo]

Might be a fsb hole, have you tried going strait to 3.7or 3.8Ghz ?

I can do 3.6Ghz at 400x9 but the voltage required is just too high to keep it stable, at 1.5xx range, and temps break 85c

[dengyong]

to NapalmV5
i turn down 500VR to bout 200ohm ....any further than 1.26vGPU protection kicks in ...any clue?max i can get 1.26v sad
mod 1/2/3/4/11/12 bridged / 500VR at the back /no pencil.

Does your screen turn a solid color ?

[NapalmV5]

@ hehee/anyone else avoiding step/s

thats why its a 3 step guide

the higher the step 1 the higher the step 2 the higher the step 3

[NapalmV5]

@ Demo/everyone else removing mod/s

boy you guys give up fast

temps went up! quick! remove the mod its a virus!

be very afraid of my mods.. be very afraid

no offense JK


look guys mods or not focus mainly on increasing core clock

[Demo]

@ Demo/everyone else removing mod/s

boy you guys give up fast

temps went up! quick! remove the mod its a virus!

be very afraid of my mods.. be very afraid

no offense JK


look guys mods or not focus mainly on increasing core clock

Temps did rise, but its not hitting 90c or 100% fan speed when gaming.

The mod gave me 1.20v/2.11v , my defaults are at 1.14/2.10 and i didnt get any better gains with the added volts so i just undid the mod, a bit far from the 1.26v/2.15v expected.

Any idea's why my card comes with lower volts?

And what do you use to test for absolute core or videocard OC stability , as i found out that at SSC shader clocks im fully fine with Furmark,Warhead and ATi Tool.. the only game that showed a glitch was dawn of war 2, green flashing textures till i drop the shader to 1566.


EDIT:

my temp max as of now is 735/1566/1404

edit2:

had to reduce the core to 730, got a single beep in ati tool after 10mins.

[Blkout]

Temps did rise, but its not hitting 90c or 100% fan speed when gaming.

The mod gave me 1.20v/2.11v , my defaults are at 1.14/2.10 and i didnt get any better gains with the added volts so i just undid the mod, a bit far from the 1.26v/2.15v expected.

Any idea's why my card comes with lower volts?

And what do you use to test for absolute core or videocard OC stability , as i found out that at SSC shader clocks im fully fine with Furmark,Warhead and ATi Tool.. the only game that showed a glitch was dawn of war 2, green flashing textures till i drop the shader to 1566.

I found by measuring the voltage with a DMM that the voltage varies depending on the load. For instance if you check the voltage while running a less stressful game it will be lower than it would if you were running Crysis. I believe when you checked your voltage you weren't loading the card to it's fullest which is why your voltages look slightly lower, which happened to me too before I realized what was going on.


This is a slight problem because you may have your clock speeds set to run stable in Crysis, but then load a game that's less stressful and fail. You may wonder how could this be when it passed something more stressful like Crysis, but the answer is that the voltage increases with load so even though you're trying to run the same clocks that passes stability testing with Crysis, it could fail in some other game.

I've already experienced this first hand with my clock speeds, even with no volt mods.

My card passes all day long in Crysis or any other heavy load application at 799c/1656s/1440m, however I found out this morning while playing Guild Wars that it fails within about 10-15 minutes consistently. Reducing the clock speeds to 792c/1620s/1440m fixed the problem and now it runs Guild Wars just fine. This is simply because Guild Wars isn't as stressful so the voltage regulators aren't applying as much voltage, however the clock speeds that passed for Crysis failed in a less stressful game.

[Demo]

My voltages seems to fixed at 1.13-1.14 when my clocks kick into 3d, this is regardless of the application running may it be Ati tool/furmark/3dmark06/vantage/warhead/DoW2/warcraft 3. Its also been stated that defaults are 1.16v and not 1.14.

With the mod i still get the same fixed voltage 1.19-1.20 ( rarely drops to 1.19 ) and 2.11v , does not change with the said applications/benches above.

EDIT:

Im open to redoing the mod again though, I noticed that i get upwards or downward spikes from both ambient temp reader and fan speed.

[nickless]

Ok, I'll go for a hard vmod (my first), but I've found these two 150ohm resistors.


Two small resistors are 150ohm 0.60W, the big one is 150ohm 2W. Which one I have to use for the soldering mod, step 3?
Thank's.

[Blkout]

My voltages seems to fixed at 1.13-1.14 when my clocks kick into 3d, this is regardless of the application running may it be Ati tool/furmark/3dmark06/vantage/warhead/DoW2/warcraft 3. Its also been stated that defaults are 1.16v and not 1.14.

With the mod i still get the same fixed voltage 1.19-1.20 ( rarely drops to 1.19 ) and 2.11v , does not change with the said applications/benches above.

EDIT:

Im open to redoing the mod again though, I noticed that i get upwards or downward spikes from both ambient temp reader and fan speed.

Which exact mod are you doing? My default voltage under load is 1.167(1.7) and my voltage with only number 2 bridged is 1.227(1.23). I haven't even bothered bridging more than number 2 as that's what the SSC uses so that's all I wanted.

On a positive note, since bridging number 2, I can now run Guild Wars at the same clock speeds as Crysis so that's a plus. I haven't gained any more clock speed though beyond what I had before the mod, 799c/1656s/1440m. Maybe my card is just at it's limit. Since I'm watercooling, temperatures aren't an issue.

[Demo]

1/2/3/4/11/12 as stated by nalpamv5 that gives 1.26v/2.15v , my def voltage under load is 1.13-1.14 in all tested games/bench. bridging 2 bumped me up to 1.18-1.19 , bridging 4+12 gave me 1.20 and 2.11

Vanilla 285, Palit brand.

Read through page 1 to 9 and it seems that im the only user that has a lower default VID and im also getting a lower volts from this mod.



EDIT:
Tried 1/2/3/4/5/11 now and i get the same 1.20vGPU and no increase on the vMEM (as expected ) no real gains on the core speed, any last tips before i remove this mod ?

[NapalmV5]

Ok, I'll go for a hard vmod (my first), but I've found these two 150ohm resistors.


Two small resistors are 150ohm 0.60W, the big one is 150ohm 2W. Which one I have to use for the soldering mod, step 3?
Thank's.

if you got no other choice than go with the .60w

.25w is widely recommended

.25w resistor provides performance not just clock increase

[NapalmV5]

1/2/3/4/11/12 as stated by nalpamv5 that gives 1.26v/2.15v , my def voltage under load is 1.13-1.14 in all tested games/bench. bridging 2 bumped me up to 1.18-1.19 , bridging 4+12 gave me 1.20 and 2.11

Vanilla 285, Palit brand.

Read through page 1 to 9 and it seems that im the only user that has a lower default VID and im also getting a lower volts from this mod.



EDIT:
Tried 1/2/3/4/5/11 now and i get the same 1.20vGPU and no increase on the vMEM (as expected ) no real gains on the core speed, any last tips before i remove this mod ?

yeh 1.13-14 should at least be the bootup voltage.. my amp 285 had the same 1.13-14 @ bootup before any mods

try diff mobo/psu

hard to believe 1.20 offered nothing over 1.13-14.. but the vid bridging should and it does raise vgpu to 1.26-28.. it did on my amp

[NapalmV5]

Well, to get to that resistor I do have to take the cooler off the card, and that means putting on TIM again after separating the core from the heatsink. Just saying. Well I might try a regular #2 pencil later on. Thanks again for your help.

edit: also how did you confirm that the FTW bios has a 1.29 load voltage? WHen viewing the bios in nibitor it has the same values as a vanilla bios. Or is there something else going on?

i hope this explains better how i can confirm 1.29vgpu load on the ftw

I did the pencil mod using an 8b pencil and the lowest I could get the resistance was around 330. voltage at load went up to 1.30-1.31 from 1.28. However this got me virtually no increase in clocks. My max core went up only 10 mhz and then cod4 started freezing until I removed all overclocks even though it was stable at my previous max clocks the day before. Weird.

10mhz outta .02 aint bad at all

like ive said before above 1.3 it gets to the point where stock/air cooling cant handle the load

[nickless]

if you got no other choice than go with the .60w

.25w is widely recommended

.25w resistor provides performance not just clock increase

So, I've done the step 3 with .60w resistor, reduce vdiff to 242ohm and load vcore was increase from 1.23v to 1.28v. But there is something strange. Now I can't run 3d06 on my previous settings 792/1584/1490 with 1.23-1.24 vcore GPU is under water, RAM sinks and one 120mm fan under the card.
I notice when I run 3d06 on stock setting vcore is 1.24-1.25v, but when I increase shaders, gpu and ram vcore become 1.27-1.28v.
I'll try with .25w resistor later, hope this help.

EDIT: I remove the resistor, back to 1.23-1.24v and I can run 3d06 @792/1584/1490 again. Maybe the problem is 0.60w resistor.