X800 GTO display problem.. checkerboard pattern after playing awhile.

[UnexplodedCow]

Interesting. You were monitoring your voltages while playing games? I was doing a little of that on my card, and I have a blue GTO AGP as well, and it came with 12 pipes, but is sitting on 16 now. Couldn't find where to check the GPU voltage, but VDD and VDDQ are both 1.99, which is a bit lower than it should be, so I'm thinking that's where my RAM issue is coming from. The core is fine, but I'm guessing the voltage is a touch under there as well. I put some larger heatsinks on the backside ram, and took the little Zalman ones off, put them on the mosfets, and I can do a stable 560 on the ram....probably a slightly out of spec resistor causing too little voltage feedback or something. I'm not an engineer.

 

[Corrupted]

i have asked this before and no one has answered me but isnt the newest ati driver 6.11 only for x1k and up line?? and it does seem that some have had trouble with 6.11 on x8xx cards..what are your clocks it almost sounds like they are too high....but like i have said all night, idk i am an idiot....good luck man...

Only problem I had was with the 3D preview screen in Control Center.It wouldn't work and gives an application error.I had same problems with 6.9 too,currently using 6.5 no probs.

 

[Kammster]

the original R480 bios from sapphire is optimised for the R480, i am guessing?...it will operate at high
clocks with less power tricking you into think it is stable.
the R430 bios removes the "fog of confusion" i got about voltages and the card.

the stock voltage is about 1.08~1.15 (lowest under different loads)
volt mod resistor to 5Kohm 1.18~1.25 (lowest under different loads)


the R480 does 540mhz @1.4v in X850 xt pe
so that about 385mhz per 1v (stock GTO .......1.08v-415mhz;....1.15v-442mhz)

so my max overoclock would be 460~470mhz
dolf said that the gto blue card was designed for a R430 so i suppose that the max voltage i can get.

testing the bios from the R430 in hitman:contracts (really good at crashing the card, nfu:ug2 also)
i was getting checker screens after 10 mins or less, clock backwards from 520 to 470 and was able to
finish the deadly cargo map.


i flashed back to sapphire R480 bios and will continue my tests.

 

[UnexplodedCow]

Sounds like you've been checking stuff. You've got a blue card, right? Where are you measuring your voltages from? Any pictures to show me, 'cause I haven't found anything yet. Mine is 480, and is doing 560 perfectly stable, without volt mod. It's the ram that's sucking. It's not stable at 560 even. I put some coolers on the mosfets, and it seems to have helped some, but I'd like to be 100% on things.

 

[Kammster]

the same capacitors in the x850 volt mod guide, bottom of the first capacitor or the top of the second.

560 on the core in which games?

 

[UnexplodedCow]

the same capacitors in the x850 volt mod guide, bottom of the first capacitor or the top of the second.

560 on the core in which games?

Ahh, thanks. Had no idea about that one. The only 850 AGP I've seen had no rialto, so I figured there'd be differences. Oh, and the core holds 560 stable at all times. It won't hold beyond 560 RAM in any game. It will be artifact free at a higher rate (590 for a half hour, no artifacts) in ATI tool .24. It's fine at 850XTPE levels in 2d, but load any game, (usually TFC, HL2, Jedi Academy, Quake 3, Quake 4, UT04, Rise of Nations, Rise of Legends), it'll checker my screen in anywhere from a few seconds to 30 minutes. It won't do it if I have the ram @ 550 Mhz. The Core has run stable at 570, but I prefer to take it a little easy, and run it at 560 with no problems.

 

[UnexplodedCow]

So I checked the x850 voltmod guide, and it doesn't show my card. In fact, it's the one I'd already looked at. It doesn't look like either the red card in the x800 mod guide, or the blue one in x850 guide. My card looks like this

http://www.newegg.com/Product/ShowI...GTO+256MB+GDDR3+AGP+4X/8X+Video+Card+-+Retail

That's how the back of my card looks, and that's nothing like anything that's been shown that I've seen. Also, not a single component has been numbered on my pcb (same as in the picture) so I'm not sure where to measure the gpu voltage from. I lucked out, and assumed my card had the same VDD and VDDQ measuring points (i.e. near the ram) and it turns out I was right. Any more help guys? I really appreciate it so far.

~Cow

 

[Kammster]

highest stable mem i got is 550

the two capacitors in the center of the board with the same arrangement.

 

[UnexplodedCow]

For testing VGPU voltage?

 

[UnexplodedCow]

Well, I just tested voltage off both caps...pushing 1.96 volts. I know that can't be vgpu from reading the voltmod articles...so I'm guessing this is my VDD and VDDQ? I'm still a little ignorant on this stuff. Anyone know where to test VGPU?

 

[Kammster]

Sapphire X800 GTO blue
P/N:11X-CC661ASA-001
R480 bios
16 pipelines
vcore :1.2v~1.25v ?
vdd : 1.8v


i didn't read a little paragrahp at the top of the x850 volt mod guide, saying some models have little increase or non but pencil modding. so i suspect that my vcore increase was just a little by the bios.

removed the pencil mod on the 5240.
modd the Vdd is bad for me, slightest change in 750ohm resisitor (711ohm - 1.98V) white screen with green lines at game load.
since there is no Vddq i am guessing increasing the vdd increases vcore?


bascially volt modding is useless!


i tried finding my max settings one at a time(starting to think mem is culprit)
apparently stable settings:
520:495
400:540

470:540.....crash (appeared to be stable at volt-mod)
520:540 ....random crash
520:525....testing


used hitman contracts .. deadly crago wait in rain as cut scene comes in if it finishs i am good.

 

[UnexplodedCow]

I've been running mine at 560/550 and haven't found anything that will crash it...no volt modding thus far. Did you try increasing the agp voltage supplied, or does that not work anymore? Just a thought.
Shouldn't your VDD voltage be higher than 1.8?

 

[Kammster]

tested 500:506 had one crsh after couple hours of gaming in two days so i locked it in bios.
seems stable but may be stable enough.

DFI genral boards don't have voltage support.
i'm done!

 

[UnexplodedCow]

Sad to hear it didn't OC so well, and that the volt modding didn't help. So far upping AGP slot voltage has had no effect (didn't think it would). Is it supposed to help? Anyone?

 

[Kammster]

470:500...totally stable

played the entire HL2 at 480:520 never a crash
got one crash in nfs:ms after 3 days so went back to hitman and loaded the cut scene 5 times in a row at different frequency to get the most stable.

 

[Kammster]

the card started cherkerboarding again. so i ripped out the cheap 17A@12v

400w psu and put in a generic 500w 20A@12v.
the gpu voltage booted at a stable 1.3v...not jumping around 1.1~1.2v like

before.
at 520:540 the card handled 4 passes of the cutscene before it

crashed...before it could only do 1 pass.


Testing at 465:499
boot voltage 1.3v
load game : 1.27v @core, 1.83@mem
during the 6th pass: 1.2v:1.73
hopefully its the psu causing this drop.Random crashes???


with a quality psu with 25A on the 12volt rail i "might" never see another

checkerscreen running as an X850XT.


Just to make sure...i dual powered...the 400w at the zalman and x800gto
the 500w at the rest.

Testing @ 520:540
load game : @1.3v core,1.86 mem
during the 8th pass: 1.28v:1.85v

i really am considering dual power but it inconvienent to setup.
Both psu remained cool and never got hot to the touch.
this pcb has to be drawing more power than 75watt PCIe

 

[peach1971]

Just for the record:

I´m running a X800 GTO AGP (R420) with a HEC 300W, only 15A@12V PSU.
6-pin molex power adapter (2x 12V parallel)
GPU clock is OCed to 520 MHz.
Never had any cherkerboarding effects.

 

[Kammster]

the problem is not the core or memory, its about the PCB and how it regulates and supply volatge.the sapphire card roundup at digit-life shows agp card with 4 or more big capacitors on the front vdd and vddq on the back. this GTO model is diffrent from the blue gto and gt in the roundup. it resembles the PCIe cards with 2 capacitors on the front and only a vdd chip.



OK, i believe i've got the final answers to anyone with similar problems.
different games seem to affect the way the card draws power...agp slot (older

games) or 12v molex (new games), based on noise cheap PSUs made.


cooling the gpu below 70c
heatsinks on ram
heatsink on rialto chip

a stable 12v supply will keep the card stable during games like HL2,far cry
but normal agp voltage doesn't seem to keep up?
since i can't change agp voltage and would not reccomend anybody to.
i have to set the clocks based on the lowest voltage.

X850xt
520 @ 1.4v
590 @ 2v (PE)

so little maths and
lowest voltage
1.2V so...445
1.7v so...500


set 440:500 and tested in colin mcrae 2004 and it stable, other clocks

failed.

this sapphire card isn't as much of a freebie as i hoped.

 

[UnexplodedCow]

I've got a 25A 12 rail rated psu (PCP&C Silencer 470) and I had checkerboard issues. I never did find my core voltage on the card, but the RAM voltage is what was a bit lower than I expected. I have a Zalman vf900cu on the core, the small zalman ramsinks around it on the one side, custom ramsinks on the back, and put the other set of zalman ramsinks on the voltage regulator transistors to keep them cool. I also have a custom heatsink on the rialto chip...all this made virtually no difference to how high I can clock my card. The core has been totally stable at 560. The ram is good for 550, and monitoring voltage has showed me that to get any more Mhz from this card, I'd have to voltmod the ram section...I'm totally satisfied with the core (I've had it run at 580 without any hiccups alone, but bumping up ram speed will cause the card to lock, even with minor speed bumps). I haven't seen any sign that upping AGP voltage will increase OCing ability at all. The other way to look at it is this: if you're running an old Athlon XP system (like me) with an x800gto..even at stock speeds, the system is most likely cpu bound.

I actually see the best performance for games if I run them at higher resolutions (1280x1024 and 1600x1200 respectively). As I drop to 1024x768, performance really doesn't improve for any game vs. 1280x1024. As for UT04, it's 1600x1200 all the way, in holy s**t mode and I get my best fps.

 

[Kammster]

my situation is apparently unique for the X800 GTOs 9or i got a bad card) and applies only to
Sapphire X800 GTO blue
P/N:11X-CC661ASA-001
R480 bios

sapphire's red gto ,pcie and r420 blue models don't seem to have the problem.

according to the atitools cube framerate, even at these speeds i only 10~15% slower than X850Xt

i am really done now?

 

[UnexplodedCow]

Hmmm, that is interesting. Mine is a Sapphire blue GTO AGP with R480 and 1.6ns RAM. Oh, yeah, you said you did the volt mod for the ram on the card. Do you have any pics you could post of what goes where? I'm very interested to know where to put the stuff for the ram mod primarily, as the core goes high enough. Oh, what FPS are you pulling in the Ati Tools cube? Mine is averaging 321, although it did run at 280 before I got another 100 Mhz out of my proc. That's with the pipe unlock and card OC'd to 560/550

 

[UnexplodedCow]

Well, I finally got my card to crash at 560/550 in NFS Carbon. Lowered it to 540, and I haven't seen any problems after about a half hour. But jeeze, only a 50 Mhz gain over stock is pretty craptacular for 1.6ns ram. My biggest problem is that I need to know exactly where to vmod my card, but I can't find any guides for the blue agp rialto card. I keep seeing the 800 and 850 agp guides on here, but they won't work on my card. So does anyone have a clue where to do the vdd and vddq vmods on a Sapphire x800 blue agp rialto card? Thanks in advance.

 

[Kammster]

the card seemed stable at 440:500 so i reinstalled some old games see how they look in high dteatil at 1280x960(monitor max)...old card was a 6200. halo, fable, raven shield all well above 60fps.
then i decided to max out hitman contracts (42fps) and BOOM...checkerboard.


finally ended up underclocking to 400:480.
the core reqiured lowering the memory to be stable in the deadly cargo cutscene.
memory temp is always below 52deg.
testing 450:450 (16p) right now.
only a slight increase in frames.

i am totally confused now.
does the rialto chip suck power from the other stuff?

 

[UnexplodedCow]

Well, I finally figured out the vdd Vmod myself. And so far I upped the voltage from the 1.9 to 2.08 (710 ohms), so I'm happy enough with that. Hopefully that'll give me the extra boost to get to xtpe levels. I'll comment back and let everyone know how it goes, or if it even makes a difference. My core voltage is on par at 1.99 volts, so I'm happy there, too, so no v-core mod. Also, Vddq is at 2.08 volts, since I think it's using the same voltage controller chip as the vdd. Oh well.

 

[Nameless]

Just to chime in I also got this issue and for me at least the problem was definitely that the card was running too hot, when I popped open the case and put a fan on it was stable for hours of game play. Closing it back up glitches would begin to occur within 10 or 20 minutes of putting it under load.

The thing that threw me though was that it worked fine for several months before this began happening so I dismissed a hardware issue at first since I hadn’t really changed anything about the machine during that time. So yeah, I'd say if this is happening try to cool off the GPU and see if that fixes the issue.