Printable View
vmod works fine, but i wasn't able to increase the performance at the moment. will try that next week with new cooler and ram chip coolers.
when setting vmem under 1,85V, i wasn't able to go over 1000MHz without errors.
I'll probably do some testing next weekend.
btw, my card is a reference one. got it in the end of november. i expect the perfect vmem for my card will be around 1,94V, but maybe this changes when the ram chips are getting cooler ;)
@jason4207
Thanks for the comments! The hot melt glue-gun was definitely a major asset in getting everything to stay where I wanted it on the card. If anyone else tries the same, I might suggest using the strongest hot-melt glue you can get your hands on. I used really cheap stuff, and although it comes off easily if you ever want to undo the mod, I also noticed that it's on the edge of letting go when the card heats up under full load.
As for the OCP, I did a few more tests using FUR, and you guys are definitely right about it being hard on the cards. Although I adjusted both cards to be around 1.275V under load (ATItool), when FUR was running I measured 1.302V! I could imagine if you were already running a really high voltage, then the donut of death could definitely push your card over the egde. I might wind up doing the OCP mod after all.
As for the Chimay, you're absolutely right, the Blue is one of the best beers money can buy, besides maybe a Westvleteren if you happen to live in Belgium! I actually find a little bit of alcohol helps tame the hand shakes if you're not the steadiest person in the world. Don't get trashed or anything, but a little bit of relaxing definitely helps when soldering that one joint near the Primarion!
Cheers!
XSS,
As it happens, Palit 8800GT uses the same Richtek vGPU controller chip as your Gainward meaning the vGPU mod for Palit works for your card too.
The PCBs are almost identical so even the solder points are exactly the same.
Here's a link to the mod guide:
http://forums.overclockers.ru/viewto...617777#4617777
I have deluge's board and I have been trying to fix it. It boots up and the motherboard makes a long beep and nothing shows on the screen and the board is showing no volts... Anyway I managed to have it start up once, after soldering on some resistors and making a trace, with the beep no image however, and it showed 2.1v!!! So I shut it off and went to change it to another resistor, but somehow I messed the rest of the other traces up... Anway the main point is that after the traces were destoryed the rest of the way I used two variable resistors from each of the two black points I circled and set them both close to 15 ohms. I went to a variable resistor and then straight to a ground, as the other traces are no longer hooked up. Do I need them to be? because the board still shows no sign of volts... It seems that the other point I circled in black might have got messed up, but I dont see any traces that go to it, so maybe it was just used to hold the chip on... anyone have any ideas on what to do... I'm pretty sure this board is a gonner, but any ideas would be great. I think this is the data sheet, but Im confused on how to read which points corralate to where... Anyway http://www.primarion.com/PrimarionMa...s/PDB-3544.pdf
http://i242.photobucket.com/albums/f...mylesboard.jpg
That's the PX3544 whitepaper, it's not the full datasheet with full pin-out and functional diagram.
I think this is the problem. Only 1 (the one on far right) of the 4 original resistors ground in a real GND. If we just had the PX3544 datasheet...Quote:
I used two variable resistors from each of the two black points I circled and set them both close to 15 ohms. I went to a variable resistor and then straight to a ground
With mods 845 Core !
http://www.techpowerup.com/gpuz/dw5kv/
Thanks largon, that would make more sense, because when I had the 2.1v the left trace went to the two end traces with a 16ohm reistor. (the ones written with ok) at this time the resistor with the blue circle was intact, but it didnt match the resistence of the other card, so I soldered a 33ohm resistor on top of it, however soon after this blue circled resistor broke... The remaining traces and resistors on the right are still ok... I'm just not sure what to do in place of the resistor with the blue circle... I'm thinking put a vr there and set it to 25ohms... It will probably be easier to just look at the pictures, all of the x marks show resistors that no longer exist, the black lines indicated traces that are ok... The second picture shows my new plan.. :)
http://i242.photobucket.com/albums/f...esboardnew.jpg
This is what I plan to do now...
http://i242.photobucket.com/albums/f...boardnewer.jpg
I think the thing circled in blue is a capacitor.
And I think Largon is saying that black question mark on the far right is the real GND.
That would make a lot more sense! I wonder if that is even necessary then...
The SMD circled in blue is a capacitor, not a resistor. :)
It appears it's not necessary for the card to function (I *know* it's not) but ofcourse, it's there for a reason...
Those two resistors marked as "X" can be overriden with a single resistor @ 14-16Ω (been there done that).
jason4207,
:yepp:
The far right resistor is indeed soldered directly to the main GND plane.
Pencil voltmod 8800gt Xpert Vision(Palit)(pink)
907/2322/2246 with 1,47v core
24/7 with 1,34v core
My is Xpertvision 8800GT 512MB DDR3 PCI-E
Non-Reference | Non-Sonic model (That's the card )
Thanks Largon & Jason, the card now WORKS!!! The only problem now is that the card is plastered with electrical tape and the solder connections are extremly delicate, but it works! Total hours soldering and desoldering about 6ish all in all, but it was worth it and I learned a lot. Thanks again guys. Now my only problem is that I seemed to left my sli-bridge at home...So it will be a few days before I will be able to test them in sli. :) :clap: :up:
:eek: :nuts:Quote:
Originally Posted by XSS
1251MHz GDDR3 (DDR-2502)
Whose GDDR3 chips are those and what are they rated? Samsung? 1.0ns? 0.9ns?
:eek2:
They are Qimonda i have cool it with the stock cooler @ 60
Clockero @ post please Picture of this mod at the card !
1[/URL] , 2, :up:Quote:
Originally Posted by XSS
layer on a lot of really tough glue... a lot. Epoxy would probably be best. Let dry 24 hours.Quote:
Originally Posted by scottc19
Clockero @ You have pencil that
http://img254.imageshack.us/my.php?i...20mark1hn6.jpg ?
So just the points with a pencil , if so, how often?
Can you make a Macro foto of this , i will too make this mod !
I Can't make a Macro foto of this , only this :( .default reistor(yellow) is 521Ω , make it 480Ω for the ~1,35v core. red is for the mem .Sorry for my bad English !:slapass:
@XSS: amazing to get those qimondas up to 1,25GHz. can't believe it :D
what's the maximum mem clock you can bench with and what's your current vmem?
btw, did you use looser timings?
Look how high clockero 3dmark score is... I want my mem to come close to that-his sm2 is about 1k higher than I had mine at 900core... 1kmem :( http://service.futuremark.com/compare?3dm06=4796070
On a side note I just did my rma on my other 8800gt today...since it wouldnt do stock mem clocks without mods... :hm: maybe next one will get closer to clockero's card :shocked:
hi guys, could i have a very high resulution photo of the back GPU pcb of 8800gt ? i nedd it to look if there are some resistences missing in a used 8800gt i have to buy, really thank you!!
Would be easier if you took a picture of your card on the area you think something is missing. Anyways, there are tons of unused SMD solder pads on the backside...
Im not sure how to reconise the OVP mod kicking in?
If it kicks in the Vgpu should be 0 or 0.2-0.3
If the chip crashed because of to high OC it should keep the 1.35 VGPU ?
OVP causes a trip without any load just as soon as a certain voltage level is achieved. OCP would cause tripping at load. Both result in 0.X vGPU.
Crashing due to a too high OC shouldn't cause any change in vGPU.
i have no digit cam right now :( i'll be gratefull to anyone can help me posting an high res photo of the back of the card in oppite site of gpu ( pcb right back gpu) thank youuu :bounce:Quote:
Originally Posted by largon