8800GT voltage mods

[largon]

Niccolò,
Bios mod can not get you more than 1.10V. Hard mod is the only way.


varga,
This will get you to 1.325v.

[varga]

varga,
This will get you to 1.325v.

Thanks! I will try it out soon. If you feel like having too much spare time, you could send me a PM how did you figure that out, I am really looking forward to learn more about vmodding/electronics.

Greets!

[Niccolò]

@LArgon:

Actually it can. I measured with my multimeter and it's 1.25v. Anyway it's not a problem for me to hardmod, but i won't do it as long i'm not sure it will work. Voltage it's enough to get over 700, but it doesn't work, so the problem it's not voltage. Anyway, looking at the picture you posted, voltage control of this card works like 7900 or 9600gt? Is there a code table and a bunch of bit you can short to get what voltage you need? That would be great, i dont' like putting trimmers on my boards..

[largon]

Niccolò

You have a modded bios that goes >1.10V? That's certainly a huge breakthrough as people have been trying that for months and months without success. Btw, what version of Nibitor were you using to mod it?

Please post your bios that allows 1.25v.

And nope, there's no VID bits we could alter. Trimmer is the only way.

[varga]

varga,
Nope, there's no VID bits we could alter. Trimmer is the only way.

Ah.. so it won't work? If you could confirm this I would be graceful, because I don't want to kill my card.

[largon]

varga,
Oops. That quote wasn't intended for you but Niccolò. The pic I posted regarding your card is valid.
The mod works by altering the VID-code bits found in the datasheet (page 12, table 2) - these bits dictate the base of the output voltage (vGPU). At stock your card has VID pin config VID4@1, VID3,2,1@0, VID0@1 = 1 0 0 0 1 = 1.125v. VID-pins are described on page 1. If we were to open (break) VID4 and close (connect) VID3 we get 0 1 0 0 1 = 1.325v.

But not all cards have such bits that can be altered. Such as reference design of 8800GT.

[micheskillz]

Been reading 95% of this thread for the last days, i got 2 questions.

Is a 500 Ohm chiptrimpot going to work for vGPU? (don't know if "chiptrimpot" is any different from a VR pot). It's a 20 turn pot, can change the resistance from 0 Ohms up to 500 Ohms.
The pot i'm using is 5 x 5millimeters tiny btw. Heh.

Is there a +/- on the pot so that you have to solder one of the legs to ground and the other to the FB and not the other way around? I mean, does it work both ways or not?

Thanks!

[largon]

Got a link to the "chip trimpot" you're talking about?

Anyways, it sounds like it would work fine. And these things have no polarity.

[Niccolò]

Niccolò

You have a modded bios that goes >1.10V? That's certainly a huge breakthrough as people have been trying that for months and months without success. Btw, what version of Nibitor were you using to mod it?

Please post your bios that allows 1.25v.

And nope, there's no VID bits we could alter. Trimmer is the only way.

Actually every bios i tryed allowed me to reach at least 1.2v. Version of nibitor is 3.9 (the last i think). Anyway this is my original bios, but as i said i can't get over 700mhz, and dont know if it's a bios issue or something else. I will probably try the hardmod but i don't see much use in it couse volts already are 1.25, and still useless.
http://rapidshare.com/files/105291734/Original.rom.html

Largon can you upload your bios too, so i can try it and see if it solve my problem?

[varga]

varga,
Oops. That quote wasn't intended for you but Niccolò. The pic I posted regarding your card is valid.
The mod works by altering the VID-code bits found in the datasheet (page 12, table 2) - these bits dictate the base of the output voltage (vGPU). At stock your card has VID pin config VID4@1, VID3,2,1@0, VID0@1 = 1 0 0 0 1 = 1.125v. VID-pins are described on page 1. If we were to open (break) VID4 and close (connect) VID3 we get 0 1 0 0 1 = 1.325v.

But not all cards have such bits that can be altered. Such as reference design of 8800GT.

Thanks for the explanation. Just one more question: why dont they simply short the gates instead of soldering an smd resistor(?) there?

[largon]

Niccolò,
How did you mod your bios to 1.25v? I can't see any options >1.10v in Nibitor with the bios you linked.
I don't have a 8800GT (only GTS 512MB) so it wouldn't be of much use for you.

Also, have you measured the voltage with a multimeter that it actually is 1.25v?

varga,
Yup, bridging the pads should work too. Moving the part can be easier to reverse, though. And it's tidier too.

[Niccolò]

Sorry, my fault, thought you wanted my card original bios This is the bios i'm using, it's an EVGA bios and it's set to 1,4v but really it give 1.25v-1.27v (guess this is the maximum for bios mod). Checked with multimeter of course.
http://rapidshare.com/files/105335152/evga14.rom.html

Anyway have you got a clue why a card wouldn't overclock? It seems like additional voltage does nothing at all! At stock it does 700mhz, with mod still 700mhz. If i go further 3d mark will crash, and crysis too. I checked with a damn multimeter and the volts are there, i can't understand this ******* card. I flashed the moon in there too, and still nothing. Could it be an overheating problem? The gpu is on WC, but the rest of the card is totally naked and no fan on it, but i don't think heat (and just on the pcb, couse gpu is fresh) is able to reduce this much you oc, is it?

[jason4207]

Sorry, my fault, thought you wanted my card original bios This is the bios i'm using, it's an EVGA bios and it's set to 1,4v but really it give 1.25v-1.27v (guess this is the maximum for bios mod). Checked with multimeter of course.
http://rapidshare.com/files/105335152/evga14.rom.html

Anyway have you got a clue why a card wouldn't overclock? It seems like additional voltage does nothing at all! At stock it does 700mhz, with mod still 700mhz. If i go further 3d mark will crash, and crysis too. I checked with a damn multimeter and the volts are there, i can't understand this ******* card. I flashed the moon in there too, and still nothing. Could it be an overheating problem? The gpu is on WC, but the rest of the card is totally naked and no fan on it, but i don't think heat (and just on the pcb, couse gpu is fresh) is able to reduce this much you oc, is it?

What is your DMM reading when you use the stock BIOS?

I think your reading the voltage wrong, or your DMM is messed up. Which ground are you using? Try one on the gfx card PCB itself.

I think the fact that your not able to clock any higher means that your not actually increasing the voltage through the BIOS like you think you are.

No one else can change the voltage above 1.1v using the BIOS on a reference design, so I don't know how you would be able to.

...


Alright, I've checked out your BIOS, and you have done the same thing that many people tried a while back. You are changing the VID labels. This does not change the actual voltage. VID03 will always equal 1.1v even if you change the label to 1.4v. The label is for our reference. The gfx card could care less what we want to call it. All it sees is VID03, and it sets the voltage to a predetermined voltage based on that. To get any higher you would have to have a VID04. It is possible to add VID04, & VID05 to the BIOS (there's room in the code), but the card doesn't see anything above VID03 no matter what you do. [It's probably 2-bit, so it can only see 0-3.]

Most reference design cards come w/ VID02 set, and you can change that to VID03 to get an extra 0.05v, but that's it.

[Niccolò]

What is your DMM reading when you use the stock BIOS?

I think your reading the voltage wrong, or your DMM is messed up. Which ground are you using? Try one on the gfx card PCB itself.

I think the fact that your not able to clock any higher means that your not actually increasing the voltage through the BIOS like you think you are.

No one else can change the voltage above 1.1v using the BIOS on a reference design, so I don't know how you would be able to.

...


Alright, I've checked out your BIOS, and you have done the same thing that many people tried a while back. You are changing the VID labels. This does not change the actual voltage. VID03 will always equal 1.1v even if you change the label to 1.4v. The label is for our reference. The gfx card could care less what we want to call it. All it sees is VID03, and it sets the voltage to a predetermined voltage based on that. To get any higher you would have to have a VID04. It is possible to add VID04, & VID05 to the BIOS (there's room in the code), but the card doesn't see anything above VID03 no matter what you do. [It's probably 2-bit, so it can only see 0-3.]

Most reference design cards come w/ VID02 set, and you can change that to VID03 to get an extra 0.05v, but that's it.

Man, my DMM was messed up! Actually, i only had to change batteries to it That really makes me happy couse it means it's only a volts matter. Now it reads correctly and i have 1.03 idle and ONLY 1.06v load
That's the explanation for my poor overclock. I'll be hardmodding as soon as i get a soldering gun, cos mine is more like a soldering cannon

[Niccolò]

I don't understand why a pencil mod is not possible. This mod is made on the FB pin, is it? Can't i reduce the resistance that's already there?

[largon]

Niccolò,
Pencil mod only works when a FB-to-GND resistance is measured in high kilo-ohms, this is because the pencil itself has high resistance. Resulting resistance drop will be miniscule if you add a 30kΩ resistor (= heavy pencil shade) parallel to a resistance of ~15Ω (FB-to-GND on 8800GT).

[varga]

varga,
Yup, bridging the pads should work too. Moving the part can be easier to reverse, though. And it's tidier too.

I will probably move it instead of removing, but I'll have to wait for a new tip for my soldering iron.. Will post pictures later!

[scottc19]

I'm sure that no one runs ~1.5v like I have been doing, but let me tell you that my card appears to degrading now... I haven't done all the testing to see what the lowest volts I can run stock at, but I'm pretty sure that I need more. I also think my cpu also might have issues, even though it passes 3+hours of prime,but I can't be for sure. Just something to keep in mind.

[largon]

scottc19,
You running on watercooling or what?
What clocks @ 1.5v? Temps idle/load?

[Nosfer@tu]

Is your link width still at x16?

No I dident check that.
But 75% Performance droop because of that shouldent happen.
And it was at settings where everything worked out fine erliger!

[jason4207]

No I dident check that.
But 75% Performance droop because of that shouldent happen.
And it was at settings where everything worked out fine erliger!

PCIe at x1 will do just what you described. I've seen it first hand on a P5B-Deluxe.

Some boards are finicky, and will go to x1 or x4 for no apparent reason.

[micheskillz]

I had problem with my 8800GT going into x4 or x1-mode, it happened to be that the 8800GT was malfunctioning and not the motherboard. And i don't know how/why it happens. Was a damn good OC-card Had it @ 748 / 1828 / 2206 without any vGPU/vMEM-mod... Now i can't hardly reach that with 1.25v on my current card.

[asdrubale]

hi guys, my card is a first revision 600/1500/900 with quimonda 1ns so I did the reverse mod but faced a weird problem. First I used Ati Tool Artifact Tester and from the default 1,988 vram (1,972@100K), I reached 1026@1,908 but when I tested the very same frequence with BOP 3dmark03, I had sudden artifacts again and again til I lowered vram to 1,845 to max out @990Mhz with relaxed timings...On the other hand I can pass 3dmark03-06 and Crysis benchmarks at 771/1944 (actual vgpu 1,15) but AtiTool doesn't like this at all beeping as soon as I launch it...So what test should I rely on? Why the results are so different? Thank you for patience and sorry for my errors as english is not my language.

[DINGO]

thanks to all on this thread, hats off to all the people who put the time in to help others. never joined a forum but XS seems the place to be. with that out of the road i have a gainward bliss 8800gt 1024mb golden sample. i used a pic from this thread and did the VGPU mod, stock volts was 1.1 and now has no problem up to 1.5V, i havnt needed to go over 1.5V as my best performance is at 824 core / 1000 mem / 2071 shader at 1.3V, volts jump up to 1.4V during 3D'06 (all average readings as im using a needle meter) temps are in the low to mid 50's using ATI tool test (all stock cooling) i shall post some pics over the next few days. thanks again awsome thread

[Nosfer@tu]

PCIe at x1 will do just what you described. I've seen it first hand on a P5B-Deluxe.

Some boards are finicky, and will go to x1 or x4 for no apparent reason.

HEHE damm, I unsoldered it because of that

I will try once more at some point.