7800GTX overclocks with Volt Mods??

[ViperJohn]

What the hell - is this mod working? And where are the cutted traces and remoced smd parts? oO

The mod should work fine. The mod is just a little different (and simpler) that I just sent off to Malves for a look see to double check my findings. Adding the trimmer creates a feedback voltage divider using the existing 10 ohm SMR as the high side resistance of the divider and the trimmer as low side. I have it on the 7800 I have now and will be firing up shortly to test.

Viper

[Chr0n1c]

Looking forward to ur results.

[ViperJohn]

Looking forward to ur results.

The Vcore mod works just as it should. I used a Bourns 25 turn, 200 ohm trimmer
set at 200 ohms measured (pins 2 and 3) That kicked the Vcore up .050 volts.

I was at 1.440 3D Vcore before the mod an 1.490 Vcore after at 200 ohms.

Will shoot and post pictures later as there are solder points to use the are much easier
than tacking onto the end of a 402 sized SMR.

Viper

[symbolized?]

Thanks a lot guys . You're amazing... .

//Edit: Now, while the mod is available i'm interessted in the difficult mod - cutting traces and so on...

Anyone of you out there having some pics?

[ViperJohn]

Thanks a lot guys . You're amazing... .

//Edit: Now, while the mod is available i'm interessted in the difficult mod - cutting traces and so on...

Anyone of you out there having some pics?

There is no need to cut traces. A 3 pin mod could be applied after removing the 0 ohm SMR in the PSENP circuit trace but there is no need to do that.

Viper

[Malves]

Hey John,

I'd appreciate if you can take a pic for the vGPU check point for me, plz. So I can use it along with the pic you sent me before of the ISL6592 area to update my site. Shoot me an email. Thanx, bro.

[ViperJohn]

Hey John,

I'd appreciate if you can take a pic for the vGPU check point for me, plz. So I can use it along with the pic you sent me before of the ISL6592 area to update my site. Shoot me an email. Thanx, bro.

No problem. I tried to send some pictures of a possible Vcore mod method for
your opinion and to double check me (before the mod showed up at the XS).
The email back to me for the following:

reason: 550 5.1.1 unknown or illegal alias: marceloalves@verizon.net

If Verizon is having a problem you may not be getting a lot of mail.

Viper

[Malves]

marceloalves@comcast.net verizon is dead for me.

[ViperJohn]

marceloalves@comcast.net verizon is dead for me.

Ahw maybe that is the problem lol. I must have to email addy's for you in my address book.

Now onto another issue. This 7800 has lost all hardware acceleration and is only rendering in software mode for both DirectX and OGL.

Had a real soft Vcore bump up from 1.440 loaded stock to 1.500 loaded for a whopping 4%. Was do some benchmarking and finding the clean OC's, logged them and shut the system down to make and eat dinner.

Fired up an hour later to do some more testing and bingo...no hardward acceleration at all. Just pure software rendering mode.

Thought my system might have gone buggy but a PCIe Radeon runs fine. The typical fixes liek driver re-installs, cleaning, registry strips/reinstalls, etc, etc have produced nothing.

The display is immaculate but dropping to 2 FPS in software rendering mode during bechmarks doesn't cut it!

Ya got any idea's??? Really quite clueless here at the moment.

Viper

[Vapor]

What if you undo the mod?

[harleybro]

The display is immaculate but dropping to 2 FPS in software rendering mode during bechmarks doesn't cut it!

Mine was doing that on and off also. What I had found was that I was running 31C which occasionally worked and sometimes didn't. I hadn't flashed the bios. Maybe if you didn't flash and are borderline on temps. that could be the problem.

[ViperJohn]

What if you undo the mod?

Already done but this is not mod related. This is something else. My first though was I disabled something accidently but can't find anything and the computer was shutdown at the.

I thought maybe my system dorked itself but a Radeon still runs like a cat with it's ass on fire. All accelerations setting are still enable but the card is just stuck in software rendering mode.

Viper

[ViperJohn]

Mine was doing that on and off also. What I had found was that I was running 31C which occasionally worked and sometimes didn't. I hadn't flashed the bios. Maybe if you didn't flash and are borderline on temps. that could be the problem.

Temp are fine. Heck they would be SF rendering as there is virtually no load on the GPU. The card has been flashed several times.

Viper

[Vapor]

So very very weird...what about OCing it? Does that help (you know, force 3D clocks might force 3D...)?

[ViperJohn]

So very very weird...what about OCing it? Does that help (you know, force 3D clocks might force 3D...)?

Oh the card is clocking up to 3D clocks and voltage are ramping up to spec as well when entering 3D mode. It just doesn't use it's hardware to do the work and runs software rendering mode instead. It has a beautiful display but is just slower than frozen molassis running uphill in the acrtic during winter!

Just fire up 3DM01se and set it to use Software Rendering instead of Pure Hardware Rendering and you will see what I mean. The difference here is the card won't hardware render when told to.

It is very odd to say the least.

Viper

[ViperJohn]

Here are the as modded Vcore pictures. The trimmer is a 200 ohm, 25 turn and gives a .050
Vcore bump at full resistance. Adjustment is extremely smooth with very high resolution.
.001 volt changes are easy to dial in.

The Vcore metering point is the clip post soldered at C792 in the 3rd picture.

Viper

[VoRtAn_MaDgE]

Here are the as modded Vcore pictures. The trimmer is a 200 ohm, 25 turn and gives a .050
Vcore bump at full resistance. Adjustment is extremely smooth with very high resolution.
.001 volt changes are easy to dial in.

The Vcore metering point is the clip post soldered at C792 in the 3rd picture.

Viper

Very Nice ViperJohn, any idea if there's a ovp kicking us when giving too much volts to the core?

[ViperJohn]

Very Nice ViperJohn, any idea if there's a ovp kicking us when giving too much volts to the core?

I am sure there will be a point where that would happen but it would probably be a lot
higher than any sane (read that actually improves your core OCVcore setting you dial in.

If you run a fanless GPU cooling solution like (like WC for example) you do need the get a
fan to cool the on-card power supply area. It will run insanely hot without the air flow
from an air colled GPU cooler. Hell some parts in the PS area run insanely hot stone
stock in 3D mode!

Viper

[VoRtAn_MaDgE]

I am sure there will be a point where that would happen but it would probably be a lot
higher than any sane (read that actually improves your core OCVcore setting you dial in.

If you run a fanless GPU cooling solution like (like WC for example) you do need the get a
fan to cool the on-card power supply area. It will run insanely hot without the air flow
from an air colled GPU cooler. Hell some parts in the PS area run insanely hot stone
stock in 3D mode!

Viper

I had the same idea about the ovp, not like the ati cards that we really need "killing" him to raise the gpu oc!

About the heat of the pcb of the card, i realize that when i was reading the values of the multiphase controller of the card, really hot indeed!

Guess it would be a nice solution if we spot the "hottest" parts of the vga components and put some good ramsinks on them's...like mainboards...airflow surely will improve stability in oc...waterblock cools down the gpu core and the airflow will cooldown the pcb and its components!

[ViperJohn]

I had the same idea about the ovp, not like the ati cards that we really need "killing" him to raise the gpu oc!

About the heat of the pcb of the card, i realize that when i was reading the values of the multiphase controller of the card, really hot indeed!

Guess it would be a nice solution if we spot the "hottest" parts of the vga components and put some good ramsinks on them's...like mainboards...airflow surely will improve stability in oc...waterblock cools down the gpu core and the airflow will cooldown the pcb and its components!

With FairChild FAN5240 Vcore regulator on the Radeons OVP is fixed intermally at 2.35 volts so nothing needed there. You are probably refering to the Over Current Shutdown Timer Setpoint mod at R1596 on the x800 AGP cards. It is and was never needed either.

Without modding those card would going into Dynamic Vcore Current Limiting by clamping the loaded Vcore at less the 1.600 volts if the regulator was running within spec. You deal with that at R1597 and R1603. Those mods are covered by an NDA and are not published anywhere.

Viper

[ViperJohn]

As many here know my father has been the hospital under intensive care since Friday of
last week.

I just wanted to take the time to say Thank You to everyone who emailed with well wishes
and support and to inform them that at 10:15 PM on 7/8/05 my father slipped the surly bonds
of earth to join hands with god and take his final journey to heaven.

John Hillig
aka
ViperJohn

[Jimbo Mahoney]

Oh the card is clocking up to 3D clocks and voltage are ramping up to spec as well when entering 3D mode. It just doesn't use it's hardware to do the work and runs software rendering mode instead. It has a beautiful display but is just slower than frozen molassis running uphill in the acrtic during winter!

Just fire up 3DM01se and set it to use Software Rendering instead of Pure Hardware Rendering and you will see what I mean. The difference here is the card won't hardware render when told to.

It is very odd to say the least.

Viper

It's a bug in the temperature protection system which works when the card is too hot AND too cold (30 - 35'C):

http://www.tomshardware.com/hardnews...07_151857.html

<shudders as he posts a link to TH>

I think a BIOS flash fixes it on some XFX cards.


PS - I am truly sympathetic about your father.

[ViperJohn]

It's a bug in the temperature protection system which works when the card is too hot AND too cold (30 - 35'C):

http://www.tomshardware.com/hardnews...07_151857.html

<shudders as he posts a link to TH>

I think a BIOS flash fixes it on some XFX cards.


PS - I am truly sympathetic about your father.

Thanks about my dad.

The card is running around 50C. The bug will cause a car to throttle back to the LP 3D clock. This is different. The card is running full HP 3D clocks. It is just not rendering using it's hardware.

Viper

[happyhero]

Thanks about my dad.

The card is running around 50C. The bug will cause a car to throttle back to the LP 3D clock. This is different. The card is running full HP 3D clocks. It is just not rendering using it's hardware.

Viper

was thinking of doing a vmod to my new 7800 when it gets here, but i might hold off untill this issue is fixed. sorry to hear about your dad. and the best of wishes.

[Victorshen]

Here are the as modded Vcore pictures. The trimmer is a 200 ohm, 25 turn and gives a .050
Vcore bump at full resistance. Adjustment is extremely smooth with very high resolution.
.001 volt changes are easy to dial in.

The Vcore metering point is the clip post soldered at C792 in the 3rd picture.

Viper

nice mod there, does lower resistance give ya higher vGPU?

victor