• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

X800gt w/R480 core unlock facts?

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
I'm sorry to ask another question about this topic but after reading as much as i could about it, I am actually more confused now than ever about what needs to be done or if it can even be done..

I'm a nood when it comes to this, but EE is my profession so I feel that we can come to a solid conclusion if this is possible or not.


R480 core on Powercolor PCB.
215rbqaga11f
GF5034.1
0445AA

I've noticed a few differences between the card pictures i have seen and wonder if anyone else has wondered about these also..


looking at the VPU right side up(able to read the etching..) to the right is an array of 10k resistors that are obviously jumper setting of some sort, one side pulling down to ground, the other pulling high.

on my board x800gt/R480, r347(pulling high) is not populated but it's low side counter part R348 is NOT . on another person's card, x800pro/R430? R347 IS populated and r348 is NOT.

on my card, RP213,214,215 are populated with resistor networks but on the pro card they are not.


some have said that the BGAs are locked via cut traces on the backside. Has anyone explored the possibility that they may be more likely to do so on the PCB instead? Here are the arguments for my case..

Almost all of the PCBs I have seen are IDENTICAL to the ATI reference and to each other. I don't see why mass manufacturing would bother with time consuming things such as laser cutting traces when they are buying *better* cores to pass off as lesser parts. we are talking thousands and thousands of boards. each one of these boards has stickers that are placed on the back telling you what it is making me think that it's possible that each one of these boards could be ANY of the related products if configured properly.

I feel that all these boards are the same and are likely locked to a certain configuration via software and hardware combinations. the software will read the hardware codes and configure itself appropriately. However, with any digital hardware these days you usually have more than just the bios code to tell the drivers what the card is. Usually you see some kind of PROM on the PCB that is unrelated to the BIOS IC that supplies device IDs and other information. We need to start looking for those types of situations here.

Can people with the R480 core cards post pictures of the resistor settings in the area mentioned above? i'd like to compare.
 

MD

New Member
Joined
Oct 4, 2005
Messages
23 (0.00/day)
Hi,I am an ee student and do see your argument,The nvidia 6800 series uses a resistor block of 6 resistors to configure the different combinations of card outcome.some are left open while others are connected via a resistor. the 6800 le,and standard (vanilla) use 6 resistor combinations,while the gt and ultra use a block of five resistor combinations,while there are alot of big changes in the engineering aspect of the pcb's aswell as software ect..The basic config can be changed to make almost any one of those cards.However a modded 6800 plain card will never be an ultra due to memory speed and type pcb thickness,two molex connectors,ect ect. But with the ati cards I am currently looking at the same things you bring up for discussion but all I can find that is relivant in some modles of the pro to x850 mod are specific numbers on the chip which tell if they can or cannot be modded,and the laser cut between the contacts on the chip are so far the only things I have found to be the difference in the different configureations so far.
 

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
Hello, The R480 BGA does not have the pads on the top right corner as the other BGAs do. Someone said they are on the back side of the package. I don't doubt that there are some kind of links back there but I'm not so sure they are strictly for determining the type of VPU. I've been in manufacturing and sometimes companies take the time to do things like that but most of the time Time is money and it's must less costly to program the stuffing machine to leave out a resistor or place it elsewhere than to setup a machine to handle processors and cut them at strategic places. I doubt they handle the VPUs much more than absolutely necessary since it is obviously the most expensive part of the assembly by far. Next would be the PCB then RAM and then the other parts. They would also likely setup a number of different product lines to produce the different cards.

the way I am seeing this is that most of the PCBs are identical if not almost completely based on the reference design, they all have various jumper settings for unknown functions, there are a few different programmable ICs(some PROMs, a MCU) among other things like using a R480 core for the GT model. This leads me to think that they are mass producing one card and programming (via hardware jumpers and software) them later to fit demand numbers.

A lot of people also theorize that the R480 in use as a x800GT is somehow inferior and can only be used as a X800 and not a x850. I doubt this 100%. In manufacturing it's a whole lot cheaper and easier to buy larger amounts of one type of part and get the large discount than it is to buy the cheaper parts and have to setup your machines to handle different parts and placements and get much less of a discount.

I wish someone could get a hold of some datasheets/whitepapers on the processor and/or reference design. I worked on a project that used Intel reference design motherboards and I was able to mod the PCB itself to overclock CPUs on a board that would otherwise not overclock. I feel that we should be able to do the same here.
 

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
UPDATE:

I tried ATItool before but it does not run for some reason. everything is 0. tried it again today.. still 0.

I moved the resistor mentioned in the first post and nothing has changed. Rivatuner sees this as a R423 even though the etching on the silicon says R480 so this is an obvious BIOS software setting which i think we all know already.

Any one have a bios that has been converted from BIN to something intelligable? i might have to invest in another card as this is getting way too intriguing..
 

MD

New Member
Joined
Oct 4, 2005
Messages
23 (0.00/day)
Yes I see your point,Like the exapmle I used with the 6800's I too belive that it is very possible the parts must be mass produced in a cost ecinomical way for the developer.I dont see why there still wouldnt be some sort of resistor configuration on the pcb itself like you mention it is simply cheaper,However chips comming off the die vary alot sometimes some that may have one bundle of pipelines damaged say 12 instead of the normal 16 or whatever.The research I have done has showed me amazing differences in one chip made right after the perfect chip.So like you are saying it could be much more cost efective to flip a couple resistors high,and one to ground or whatever the case may be,then asign a bios accordingly to make into high end,mid,or low.If you end up getting a bit more info on that let me know I will start tweaking a few bios's and see if I cant adapt one for an upgrade. Sometomes there's more than one way to skin a cat,let's see if we can find a new one lol :p. Mike.
 
Joined
Mar 16, 2005
Messages
2,058 (0.30/day)
Guys, it is nice that somebody re-opens that issue in general. Originally I used to have exactly that idea in mind but digging in to the problem with quads unlocking (on ATI based cards) I become not so confident in that idea. For sure resistors configuration is one of the methods for locking / unlocking and so on but ATI did something different at least on R420, R423, R430 cores but I am almost sure that on R480 is the same. If that issue is interesting for you, please see that http://forums.techpowerup.com/showthread.php?t=4971&page=1&pp=10 and probably you can contribute with something :) .
About laser cuts on the back side of the core - yes it is true and you can see it also at the provided link.
 

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
Yes I read through that pretty quickly but will re-read when i get a chance to sit down! I didn't see a definate piece of evidence that the cuts on the rear of the BGA were links for locking the IC but I may have overseen that. I *might* be able to get a hold of the guys that stuff PCBs for me and see if they can do BGA stuff. maybe I could get that BGA assembly off and actually try to link the traces and get it back on there... until then I am going to try to figure more out. Back to the initial requests though, I would like to gather as many different hi res pictures of the front and backs of the ATI reference designs like Powercolor and others use so that I can start to look for clues in the resistor settings. this could tell us even more.

I have to beat a dead horse here but I personally have read a lot of speculation and trial and error and not very much fact based progress. I don't mean to be offensive and rude by that, that is just an observation.

Unless ATI supplied the R480 BGA with the quads already disabled I still have a doubt that any manufacturer would dare to modify the core on their own.. this could potentially cost them money through failures and handling. I was looking at my card again last night and noticed that their is also a PIC MCU on the board but it seems more related to the DVI stage than the core stage.

I was working a project once where we were supplied massproduced motherboards made by a well known company.. failure rate was damn near 30% of NEW motherboards out of the cartons! We were assured that the motherboards were tested to their fullest extent but we could not get out from under the 30% failures. We setup a complete datalogging system and found the hangup to be a small EEPROM similar to the 25lv512 on the vid cards not giving the correct information. turns out the code was corrupted on certain PCBs made under certain batch numbers. we logged the batch numbers and simply removed the EEPROM on all those boards. success rate was then 100%. This is the sort of thing that we need to make positively sure of. I haven't seen where someone has actually downloaded the code off of all the programable ICs on a reference design to actually see what each are doing at the firmware level. I would bet that during BIOS boot all the ICs are polled for their unique Device ID/ Serial number.. this is part of the reason flashing bioses give warnings..

And I did flash my card to a R480 based X850Pro non vivo card and it runs fine at Pro stock speeds. Next to tackle are those pesky pipes. I think their is going to be a relation of any of those links to pins on the BGA. I read that someone pulled their BGA off of a board.. maybe we should start to check for those relations.

Again we need to convert that BIN bios to something more intelligable like ASM to see what is really going on.. And compile a DB of pictures of the PCBs and their differences so we can start to nail this down.

what do you think?

EDIT: I see now upon reading further that Dolf has a similar way of thinking to me on the relations of pins and pads..I think we will crack this if we just keep at it and let the facts guide us..
 

MD

New Member
Joined
Oct 4, 2005
Messages
23 (0.00/day)
Hmmm well at the moment all I can provide for pictures is an all in wonder x-800-xt with the r-420 core and as you may know the pcb is alot more complex due to the onboard tv tuner ect ect and might not be very realavant here.But if you decide for future refrence you want lots of pictures of it I will do it for you.I just rma'd an r-480 that wasnt working like an r-480 should.
 

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
please post the pic, i am really only interested in the supporting parts near the VPU itself and on the rear near the VPU. I found some more pics of identical PCBs and see some interesting differences...
 

MD

New Member
Joined
Oct 4, 2005
Messages
23 (0.00/day)
ok great, I will get it for you in the next day or two I am up to my ears in rma's take care. Mike.
 

Svart

New Member
Joined
Oct 5, 2005
Messages
14 (0.00/day)
No problem! Anyone else that has pics or can take them please post them here or provide links.

One of the things I notice are the absent resistor networks on cards that are higher than GT whereas the GT cards have the resistor networks in place near the VPU. I quickly checked on my card and these are some kind of pull up/terminating resistor packs. Along with some other resistors changed on the PCB i think this might start to help determine various support structures for VPUs that are in different configurations.

If i can find a cheapie card that someone *broke* I'll get it and do some (destructive?)testing to determine what these do exactly.
 

Neocal

New Member
Joined
Sep 27, 2005
Messages
6 (0.00/day)
Photos about resistors and core

Here my powercolor x800GT 256MB in detail.
Images are in medium jpeg quality, if you need better pics, just contact me ill send you high definition photos.

Neocal.
 

Attachments

  • GPU core and resistors.JPG
    GPU core and resistors.JPG
    186.6 KB · Views: 533
  • Powercolor X800GT 8pipes.jpg
    Powercolor X800GT 8pipes.jpg
    31.3 KB · Views: 538
  • Rear pcb pic.JPG
    Rear pcb pic.JPG
    154.6 KB · Views: 472
Last edited:
Top