Radeon X800 Voltmods

[syl]

Thanks Syl ... another note from the specs is it states the maximum clock frequency up to 700mhz

I would also believe it requires very good cooling...

NP Please note the maximum frequency of 700mhz specified is in reference to the GC-14 (14ns) version. Maximum frequency for the GC-16 (16ns) memory used in most VIVO's, XT's and all XTPE's is rated at 600mhz.

 

[Urlyin]

NP Please note the maximum frequency of 700mhz specified is in reference to the GC-14 (14ns) version. Maximum frequency for the GC-16 (16ns) memory used in most VIVO's, XT's and all XTPE's is rated at 600mhz.

I stand corrected thanks again Syl... was trying to brush through it before I ran out the door to vote and off to work ...

 

[ElectriZ]

NP Please note the maximum frequency of 700mhz specified is in reference to the GC-14 (14ns) version. Maximum frequency for the GC-16 (16ns) memory used in most VIVO's, XT's and all XTPE's is rated at 600mhz.

yes i seen that ...
my GC is at 680mhz at the moment

 

[syl]

Not sure why you're getting them... it could need a little more juice.. but I'd be very careful going much higher... Someone correct me if I'm wrong but I was under the belief that the VDD was not suppose to be higher than the VDDQ, but if the max is the same then does it still matter?
you're pushing that card to the extreme

I'm confused as well...
According to Samsung: "Under all conditions, VDDQ must be less than or equal to VDD"

However, W1zzard and other experts have stated that VDDQ should measure higher than VDD on the X800's - and this is indeed the case on my MSI Pro VIVO. This would also mean ATI's reference design has the Samsung memory running out-of-spec

 

[Urlyin]

I'm confused as well...
According to Samsung: "Under all conditions, VDDQ must be less than or equal to VDD"

However, W1zzard and other experts have stated that VDDQ should measure higher than VDD on the X800's - and this is indeed the case on my MSI Pro VIVO. This would also mean ATI's reference design has the Samsung memory running out-of-spec

My conclusion was you wouldn't want you memory running at a higher voltage than what you were trying to push through the board... but by default before vmod my VDD was 2.00v and the VDDQ was 2.08v, so I concluded that what I had read and what the meter said was correct... hmmm

 

[Urlyin]

yes i seen that ...
my GC is at 680mhz at the moment

You know EZ with a core speed of 680mhz you should be able to take the top X800 spot on the Doom3 benchmark here

 

[ElectriZ]

hummmm
i think tomorrow i'll setup doom 3...

edit : 681/681 = 7030 3dmark03
re-edit : 684/681 = 7033
i think that's the max for my little sapphire

 

[Mussels]

Now im wondering.

as urylin has instructed me via PM, i pencil modded my card.
i have the ATI silencer 4, and dont want to solder, so its impossible for me to measure voltages.

VGPU went down by 30 ohms, and gained me from 534/555 to 567/567.
card doesnt go above 60C load. (although my case has good airflow)

VDD and VDDQ i lowered both 24 ohms. ZERO change in OC potential. i did and redid this repeatadly, and never one change.

since the samsung and the ATI specs seem to differ, could someone with the resistor mod please tell us which way works best? higher VDD or higher VDDQ? and at what resistance you are running at?

after adding an OCZ enhanced power lead, i got more OC too... all the way from 567/567 to 571/571... what the hell, im sure for stock cards without volt mods it'd do better.

OHM readings for my card, pre and post mod are following.

VGPU -> reduced 30 ohms (lost the paper with numbers... dammit)
IGPU unchanged.

VDD 937 ohms -> 913 Ohms
VDDQ 760 Ohms -> 736 ohms


oh and for the n00bs to this (as i am) please make sure to keep the black end of your multimeter to the one end... i use the bottom.
if you swap the leads over, you will get a different reading, and mess the mod right up....

 

[Urlyin]

as urylin has instructed me via PM, i pencil modded my card.
i have the ATI silencer 4, and dont want to solder, so its impossible for me to measure voltages.

VGPU went down by 30 ohms, and gained me from 534/555 to 567/567.
card doesnt go above 60C load. (although my case has good airflow)

VDD and VDDQ i lowered both 24 ohms. ZERO change in OC potential. i did and redid this repeatadly, and never one change.

since the samsung and the ATI specs seem to differ, could someone with the resistor mod please tell us which way works best? higher VDD or higher VDDQ? and at what resistance you are running at?

after adding an OCZ enhanced power lead, i got more OC too... all the way from 567/567 to 571/571... what the hell, im sure for stock cards without volt mods it'd do better.

OHM readings for my card, pre and post mod are following.

VGPU -> reduced 30 ohms (lost the paper with numbers... dammit)
IGPU unchanged.

VDD 937 ohms -> 913 Ohms
VDDQ 760 Ohms -> 736 ohms


oh and for the n00bs to this (as i am) please make sure to keep the black end of your multimeter to the one end... i use the bottom.
if you swap the leads over, you will get a different reading, and mess the mod right up....

Mussels if you lowered VDD and VDDQ equally then it should be fine. But not being able to check the voltage change does hamper you a bit. That verifies you are doing the right resistor. I also did those several times to verify that they were correct. I was able to reach 600mhz on mem after the vmod and barely 573 before. I raised my VDD to 2.15 from 2.00 and the VDDQ was 2.08 to 2.22. I don't have my starting ohm numbers in front of me. But before you started you said you only wanted to stablilize it at XT speeds. Well that is what you have an then some... still a decent oc. Enjoy the new speed

 

[Urlyin]

Hey EZ... I see on Nokytech that Nalioutz has a 3.4 running at 4.1 and his X800 is at 609/582 but he is only scoring 5810 on 3dmark05 I'm getting higher than that with a 3.0e at 3.8 w/ x800 550/580 ... I was just curious ...

 

[Goldlocke]

VGPU drop under load

Will the IGPU mod help with the decrease of VGPU under load?
My VGPU set to 1,80V @ idle drops to ~ 1,69V under load.
When VGPU is set to 1,70V it stays there under load!
So I guess, OCP limits VGPU output under load, right?

So solder a 250kOhm VR on each side of R1596 will help?

thx!

 

[ElectriZ]

Hey EZ... I see on Nokytech that Nalioutz has a 3.4 running at 4.1 and his X800 is at 609/582 but he is only scoring 5810 on 3dmark05 I'm getting higher than that with a 3.0e at 3.8 w/ x800 550/580 ... I was just curious ...

Its because its a w800 pro and he dont active the 4 pipelines...

 

[Urlyin]

That would explain it

 

[ElectriZ]

 

[Goldlocke]

Will the OCP mod remove the drop of VGPU under load? see my post above.

If not, why is that drop? From 1,83V to 1,73V under load.
thx!

 

[Urlyin]

Will the OCP mod remove the drop of VGPU under load? see my post above.

If not, why is that drop? From 1,83V to 1,73V under load.
thx!

The OCP is for over current protection which would shut the card down. As for the power drop you may need to turn the VDDQ in the BIOS up or your power supply my not be up to snuff ...

That's an extreme number to push through .... no stock air cooling there

 

[Urlyin]

Will the IGPU mod help with the decrease of VGPU under load?
My VGPU set to 1,80V @ idle drops to ~ 1,69V under load.
When VGPU is set to 1,70V it stays there under load!
So I guess, OCP limits VGPU output under load, right?

So solder a 250kOhm VR on each side of R1596 will help?

thx!

Sorry I see this was your original post... I don't believe that the IGPU (OCP) will aid your droop. Try to make sure nothing else is on the power lead that's plugged into the Video card. If you can take as much load off the PSU as you can to test if tha'ts the issue. The BIOS VDDQ may help a little but, don't think it'll solve it. The VDDQ vmod is for the memory buffers and a little push there may help... I'm assuming you did the VR vmod on the GPU... could it be the type of varible resistor you used?

 

[Goldlocke]

As to the PSU. I use OCZ Powerstream 520 with the dedicated coated Powerline to the vidcard.
It could be the VR type - yes. At the moment I use a cheap one. I'll try a cermet type tomorrow.

What VDDQ do you mean with "bios VDDQ". I did VGPU mod, OCP mod, VDDR and VDDQ mod yet.

 

[Urlyin]

In most MOBO BIOS(CMOS) settings they have AGP VDDQ voltage. I can set mine on my ASUS P4P800-E from 1.50v to 1.80v... used mostly to stablize overclocking of the CPU. It is said that higher voltage on the AGP VDDQ can shorten the life of your video card, but in your case I don't think that'll be an issue... it may help, what MOBO do you have ?

 

[ElectriZ]

The Vagp setting of the mobo is not used with the the card like x800...
the have an other power source ... the molex.
Change the vagp will not help for overclocking on x800.
That's help for smallest card like 9550, 9200, fx5200...

 

[Goldlocke]

I upped the 5V rail of my PSU to 5.5V. guess what?

Now the VGPU stays where it should be under load!
(upping the 12V rail didn't help).

 

[Urlyin]

Nice OCZ PSU ... good info to know ... I would have guessed 3v rail

Goldlocke... what are the core\mem speeds?

Thanks EZ ... your right the Vagp doen't help much on the X800... not realizing the OCZ PSU had dials for votage rails. It was just a suggestion...

 

[Goldlocke]

Goldlocke... what are the core\mem speeds?

620/630 so far. my card is no good overclocker.

VGPU ... 1.83V
VDDR ... 2.20V
VDDQ ... 2.24V

 

[ElectriZ]

what the card do if you go over that frequencies ?
freeze or artifacts ?

 

[Urlyin]

Ez... was looking for you on the Doom3Benchmark ...