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

[vodka3dg]

amazing work sergio!
looks like the mod worked quiet well

cheers!

[andressergio]

amazing work sergio!
looks like the mod worked quiet well

cheers!

Thanks mate

[NapalmV5]

no mate its not that remember i have the hughe Danger Den block covering ALL the other part, i did this mod as it was only on the upper part, i suppossed it would be the same dont take personal plz

Cheers
Sergio

i guess the smily didnt do it lol i was j/k

look everyone does it however/whichever sees fit.. im not forcing you guys to do it my way/only way.. i dont take it personal.. this is how i take it sergio:

modded evga ftw 285 @ 810/1836/2970 arctic accelero air cooled



10mhz core/72mhz shader better than ssc

under 40C after bench completes according to precision

ill recap C561-4 and see if it sustains 828core or 1872shader on air

[andressergio]

i guess the smily didnt do it lol i was j/k

look everyone does it however/whichever sees fit.. im not forcing you guys to do it my way/only way.. i dont take it personal.. this is how i take it sergio:

modded evga ftw 285 @ 810/1836/2970 arctic accelero air cooled



10mhz core/72mhz shader better than ssc

under 40C after bench completes according to precision

ill recap C561-4 and see if it sustains 828core or 1872shader on air

sorry napalm i didn't got the joke lol, very nice score mate W7x86 keeps scoring a bit better than W7x64

yes i do the same except that i can go a bit more on mem, what vgpu are you using for that bench, will be good to know...

if i go to 821 core OCP cracks even with 10ohms as you suggested...only gain is that now shaders go high but artifacts, i tried till 1,4 yesterday and didnt see much gains...

so far i can sustain 810/1800/3024 to play crysis at 1920x1080 on the settings i showed before wich is an improvement at 1,35vgpu

Cheers mate
Sergio

[NapalmV5]

1.38vgpu.. keep tweaking/modding eventually youll get it to sustain higher clocks


- 1.39vgpu/2.2vmem
- 2.5v 1500uf tantalum @ C561-4
- 828/1836 sustained on air cooling (arctic accelero)

[andressergio]

1.38vgpu.. keep tweaking/modding eventually youll get it to sustain higher clocks


- 1.39vgpu/2.2vmem
- 2.5v 1500uf tantalum @ C561-4
- 828/1836 sustained on air cooling (arctic accelero)

congrats my friend

i didnt do the vmem mod, wish i can find those caps here

Cheers
Sergio

[andressergio]

Napalm today i talked with a mate that's an electronic Eng. he told me to take ground not from the molex but for the proper circuit where the vr's are grounded wich is indeed the voltaje divisor, so if you measure vgu on molex you will get more volts than measuring on the vr's circuit ground. Can you confirm that ?

I got this testing 3DM06
Q9650@ 4608
GTX285@ 810/1836/3024
vgpu LOAD 1,36 / IDLE 1,16 (measured on the circuit)
W7x64






Cheers mate
Sergio

[CryptiK]

Well, of course you should measure from common circuit ground as close to the measure point as possible, however it gives little difference in reality, and you really just want to know the
relative voltage increase from the mod, knowing the exact vGPU is of little importance. For this reason I simply use the molex as ground, and probe the measure point on the card.

[andressergio]

Well, of course you should measure from common circuit ground as close to the measure point as possible, however it gives little difference in reality, and you really just want to know the
relative voltage increase from the mod, knowing the exact vGPU is of little importance. For this reason I simply use the molex as ground, and probe the measure point on the card.

thanks mate

[andressergio]

Testing the card at my stable clocks for Crysis with Q9650@ 4707mhz on XP32



Cheers
Sergio

[NapalmV5]

just as on the ssc i have no success @ high load on the ftw @ crotale vgpu (xtreme cooling required)

light load can be fun though



air cooled at that

[andressergio]

just as on the ssc i have no success @ high load on the ftw @ crotale vgpu (xtreme cooling required)

light load can be fun though



air cooled at that

for sure mate its fun im very happy of the results

why dont you use 190.62 drivers ? perform better than lastest napalam

Cheers
Sergio

[NapalmV5]

yeh 190 > 191 > 195 the diff between all 3 is 0.1 fps @ heaven bench.. havent tested anything else

alright.. i got 1.41vgpu now thru several vmods.. ill test @ 190

well sergio only way to get higher performance per clock and higher clocks is to perform diff vgpu mod/s

could you ask your mate electronic eng.. show him the isl6327 pdf.. show him the front of the card: D14-D21 (diodes) youll see why i tell you to bridge 1/2/3/4/11/12

diff diode configurations yield diff voltages.. would ask him what config does higher voltage.. or if he has/comes up with a better idea

thanks

[andressergio]

yeh 190 > 191 > 195 the diff between all 3 is 0.1 fps @ heaven bench.. havent tested anything else

alright.. i got 1.41vgpu now thru several vmods.. ill test @ 190

well sergio only way to get higher performance per clock and higher clocks is to perform diff vgpu mod/s

could you ask your mate electronic eng.. show him the isl6327 pdf.. show him the front of the card: D14-D21 (diodes) youll see why i tell you to bridge 1/2/3/4/11/12

diff diode configurations yield diff voltages.. would ask him what config does higher voltage.. or if he has/comes up with a better idea

thanks

he wont be able to answer that napalm unless i give him the circuit to study, his is not on the OC he just told me that about his toughs of grounding but soon as i get him i will

he works as an inventor he designs robots and hidraulic machines for factories in a minor scale for our country, i mean the machines that packs prodcuts etc...

i removed 12 bridge as it was no gain on voltaje before i did this mod, do you think that its also important to live that bridged ?

Cheers mate
Sergio

[andressergio]

i added a female connector and a cable with a male and a molex on the end so i can measure volts comfortable, still waiting the tantalums to try

[NapalmV5]

well sergio.. i used a diode from another 285 @ D18.. now i got 1/2/3/4/9/10/11/12 bridged.. + other mods i reached ~1.45 @ load and can push even higher.. same air/arctic accelero cooling and at this voltage the rest of the pwm is holding back further clocks.. the gpu itself runs cool.. ~40 C @ load.. id have to mod the whole freakin pwm.. or use better cooling

[andressergio]

well sergio.. i used a diode from another 285 @ D18.. now i got 1/2/3/4/9/10/11/12 bridged.. + other mods i reached ~1.45 @ load and can push even higher.. same air/arctic accelero cooling and at this voltage the rest of the pwm is holding back further clocks.. the gpu itself runs cool.. ~40 C @ load.. id have to mod the whole freakin pwm.. or use better cooling

congrats napalm very nice !!!

Can you show me the point to solder the VR for the vmem mod i just can find it...

Cheers
Sergio

[NapalmV5]

no need to do vmem mod unless you want to hit 3300+ lol

the vgpu mods alone increase vmem to 2.2 from 2.1

here you go.. 10k ohm = ~2.3v

[andressergio]

no need to do vmem mod unless you want to hit 3300+ lol

the vgpu mods alone increase vmem to 2.2 from 2.1

here you go.. 10k ohm = ~2.3v

thank you mate the higher the better will do it as soon as my 470uf/6v tantalums arrive

should be to solder a 10k resistor to ground right ? or a vr... correct ?

cheers !!!
Sergio

[NapalmV5]

yes to ground

[andressergio]

yes to ground

thanks mate

[8800's@over]

Neighborhood do you think of a Akasa Freedom Force in a GTX 285?

[MaxMax77]

Hi, could someone explain to me which of these mods on the first page would work with an asus GTX 275 non volterra chip VRM card (it has an uP6206 VRM)?
thanks

[andressergio]

no need to do vmem mod unless you want to hit 3300+ lol

the vgpu mods alone increase vmem to 2.2 from 2.1

here you go.. 10k ohm = ~2.3v

Napalm i have a doubt i will use a vr to get those 10k. so it will have to be greater right ? like 20k or more as i dont want mem to be always at 2.3---

what do you recommend ?

cheers
Sergio

[CryptiK]

As always, measure mod point to ground, multiply by 20 and use a VR of that value. Should cause minimal change when set to max resistance and soldered in place.