Pencil mod for 8600gs

[keny]

Ok guys need a little input on this retro card i have had lying around for what seems like an age...... I checked the bot and you can get a few decent boints if you can make the top few spots on this card (G98), i can make 7th place with it at stock but i could do with a boost and need "more powa" so if anyone knows which resistor i should be putting the pencil over i would appreciate you're help as i don't have a clue, i have taken a few pics, mostly of the area i think the resistors are ( i'm assuming there's one or mem and core), i know you will want me to measure resistance from the FB pin etc, just let me know what you need and i will get back with the results asap.

Cheers
Mark

 

[Bobnova]

Here ya go.




Be sure to measure vcore before you start modding!

 

[keny]

Thanks Ed I knew you would chirp in to help

You were right resistor 560 is the one with a resistance of .824 to the FB pin. How would one measure the voltage while running ? I know its probably one if the big blobs of solder on the back and ground but which one lol.

Thanks

 

[Bobnova]

Hot glue a Molex (PSU end) to the card, run one of the wires of it to one of the two big solder points one the right side of the middle. That make any sense?
Or you can just poke it with a multimeter, don't poke two things at once though or BANG~!

 

[Operant]

Ahhhhh, ed beat me to it. Only had time to peek at the datasheet. Surprisingly full of information! You could add some caps to keep things under control if you want

 

[Bobnova]

I love datasheets. The majority of the manufacturers put everything in the datasheets. IR (who bought CHiL, who bought Volterra) are the exception. They give you a pinout and that's about it. The real datasheets are NDA

 

[keny]

Do you mean these two solder points for the vGPU reading , if so, can i just run a wire from each one to the molex, as i assume one is a ground and one is the live, or am i completely wrong

 

[Bobnova]

One is the switched 12 V that the inductor/capacitor pair filters to vcore, the other is vcore.

 

[keny]

Ok ! So which one to use and where's the ground other than the bracket ?

 

[Bobnova]

Just use a PSU molex that's still attached to the PSU.
Both middle pins are GND.

 

[keny]

Cool

 

[keny]

Will this work then lol

 

[keny]

Think I may put this lot on it just for a laugh and see if it gives any sort of improvement, but it runs cold to start with so not too sure

 

[Bobnova]

Will this work then lol

Yup, that should work fine.
If you get a reading that looks out to lunch, try the other pin.

 

[keny]

I'm getting 1.6v , seems high, but maybe these old cards take a lot of volts?

 

[Bobnova]

Try the other pin, 1.6v is excessive.

 

[keny]

Ok, I will get on it after I finish work... Just started night shift

 

[keny]

Finally got a spare hour to mess with this and got the readings right (1.1v stock) ramped it up to 1.3v and managed to make some gains but not massive, how much voltage can these little gpu's take? Even with the stock cooler it sits around 36c constant so I don't know if the little pelt I've got would be that beneficial plus I would have to remove the heat from the hot side and the stock cooler is crappy and I tried to mount a heatpipe cooler to it but the card is just too small and the cooler touches the mobo before the card makes a good seat in the slot is there a resistor for the vRam as I'm capping out at around 600 on the ram and I can't find another programme that will let me clock any higher on the core (910) I've maxxed out riva, afterburner, and evga precision, the latter being the only one to go as high as 910!

 

[Bobnova]

What it can take scales inversely with how much you want it to not die.

1.3 is still stock for that generation. 1.4-1.5v on air it'll probably live for a lot of if it's cool enough.

 

[keny]

Nice I will bang some more in it then, I don't care if it dies or not but I want some boints from it before it does go pop, that's why I'm asking before I ramp it up some more. You're suggestion of that pin for voltage monitoring was correct, it was my blatant stupidity of having the multimeter set on the wrong setting which lead to the 1.6v reading, I flicked it to the next setting and it gave a more realistic value and after penciling the resistor to 10% less resistance it gave the correct value of 0.1v increase on the core happy chappy