03-21-2017, 05:36 PM (This post was last modified: 03-21-2017, 05:49 PM by raileanu.)
...now the customer took his computer, he could not wait one more day (after he was waiting for one week ), without me having the chance to complete this modded bios,on the last hundred meters, and now we are all again in the square one....OMG!
This computer is unbelievable... I don't know if anyone had ever the chance and time to finalise this issue.
I was really willing to create the solution and post it here as I search on the entire internet for days and there is no any complete solution, to see it working, that someone had actually done it.
I think I will buy one faulty with the same problem and continue my work...In the end its still a really nice computer.
Success is Persistence - Do not give up
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
lots of replies . From your previous one i didnt understood if you sucesfully implemented the solution into the BIOS and it still didnt worked or not, can you post the moded BIOS anyway? Its pitty that you dodnt have the XPS one anymore, its impossible to solve this without the machine if you are not absolutely sure what are you doing. Anyway, did you chaned the Vendor and Dev ID only? or you made another changes?
Posts: 17
Threads: 0
Joined: Nov 2015
Reputation:
0
03-22-2017, 09:34 AM (This post was last modified: 03-22-2017, 03:01 PM by raileanu.)
(03-21-2017, 10:12 PM)tsalat Wrote: Hi raileanu,
lots of replies . From your previous one i didnt understood if you sucesfully implemented the solution into the BIOS and it still didnt worked or not, can you post the moded BIOS anyway? Its pitty that you dodnt have the XPS one anymore, its impossible to solve this without the machine if you are not absolutely sure what are you doing. Anyway, did you chaned the Vendor and Dev ID only? or you made another changes?
Hello. Yes, lots of replies... I was really into it to fix this problem. And yes I successfully change the vendor id and device id with the correct one DE 10 25 04 from the graphic card vbios, but still didn't show up anything. I believe the subsys values has to be also changed but I didn't have them. I also replace once the Nvidia vbios (posted above), at line 20 value 10000 6553 but still didn't worked. The subsys must be the trick. I might have shortly another same computer and I will continue my work. It will be much easier if someone can help me with these subsys values, which are they and how to spot them on Hexeditor in the right format. Thanks.
Here is the modified bios for Nvidia 8600M GS
Success is Persistence - Do not give up
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
Hi, thx for the info. Its pity that it isnt enough to change the IDs to have it working,,,, iam wondering what is the whole code around, basically the code around the IDs is surrounded by 00 ,, maybe delete the whole block? The questions is what next
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
Hi, i found the location of the subsys i just dont udnerstand one thing... iwas lookin for subsys of the ATI card but found several "reports" from users with the same VEN/DEV id but different subsys...how to know the right one? anyway,,, this is what i was searcching for:
04-02-2017, 12:31 AM (This post was last modified: 04-03-2017, 12:58 AM by tsalat.)
Well, i tried the FX770m which i have without changing the BIOS... strange beeps but the computer booted afterwards... i have teamviewer installed on the machine and connected through to see if the GPU is accepted and yes...the GPU was there..jJust still the blank screen but because it installed the drivers i could see the DEV, VEN and SUBSYS ids
i have my bios ready but didnt flash yet. waiting for something to be able to read the bios from the recovery pins and backup like that... heh...if something will get wrong i will have some backup there
i hope to do so in 2-3weeks...this is the time untill the stuff from china will come
edit: my previous post with the picture....i found the same vlaues next to the DEV/VEN....so i guess thats the right place to edit
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
i modified the bios with the DEV/VEN and SUBSYS id (the location of the SUBSYS id which i modified is not the same as on the picture befoer, i modified the loation after the DEV/VEN, the same string is right next to it) and flashed the bios. The same behaviour and moreover the ATI card was still workin.g..so its obvious that this is not changing anything : have no idea what to do next
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
Guys, iam fighting with this already for almost a month and not going to give up until i will know if its or isnt possible . I found a German forum where one user had the modified bios from here but he told me it wasnt working with it anyway...this doesnt mean that it cant work but its at least some information. Iam already quite sure that the FX770m is not HW compatible with the system, because the beeps i hear and even though the system will boot afterwards with a black screen the beeps are not a good sign I was reading about this a lot but usually found articles about wlan whitelists but at the end ist should be the same. I dont want to remove the whitelist, for me it would be enough to replace it with the ids of the new card. Anyway, after reading a lot i found a nice amisplitter which will split the 1B module of ami bioses. Unfortunately the splitter didnt found the ACPI table but anyway. The 1B module should be the right one to edit in this case... So i downloaded also IDA pro and imported one of the 1B splitted files, the one which included as the only one the DEV/VEN ids of the ATI card... The DEV/VEN is mentioned through the whole bios just once, in this 1B module. So here is the part of the code with the SUBSYS, DEV/VEN....
Ok, here we go with the assembler language and my knowledge ends right now. The subsys is mentioned two times and in between there is the DEV/VEN. Question is if both entries of the SUBSYS are referring to the ATI card or not...well, i really hope that someone could give some hint right now and tell me if to change both SUBSYS entries and also DEV/VEN... or just one entry of the SUBSYS.... Tomas
Posts: 17
Threads: 2
Joined: Jan 2016
Reputation:
1
In the first place, the XPS A2010 which is was modding has an MXM II slot but no entry about the MXM structure inside system BIOS. The reason why its not there is because they modified the video BIOS of the default ATI card to inject this while loading the vBIOS into the memory. Therefore there wasnt anything mentioned about the MXM interface in the DSDT as well, thats logicaly why it should be there in the first place
Hence, i had to modify all these to make it work. One user injected the MXM structure inside the system BIOS according to the MXM specification 2.1. Wasnt so hard i believe but you need to place there appropriated callbacks for the BIOS to load these, this is something which i cant explain more to be honest. The MXM structure consists from information like the interfaces to which the card can push the signal (internal LCD and so on), maximum power, maximum cooling capabilities and so on. Right after the MXM structure, EDID for the LCD can be add if the LCD doesnt have EDID pins to identify himself. Well, this was the part which add absolutely new feature to the system BIOS and the new card started to work with 3 major problems:
The performance was set to throttling
The picture was ok without the drivers but bad, and i mean really like distorted signal, after installing the drivers
The fans were all the time at full speed (system and CPU)
Now, the first two problems are a DSDT issue. Because no MXM structure was present before no was added to the description table. Problem is that the DSDT table has some length. IF you want to inject it back to the SBIOS it has to be the same size or smaller. I was lucky because the XPS has a on-board GPU as well. Well, i deleted it from the DSDT table to make room together with some OS entries. The on-board GPU is in this case never active, even with the default DSDT, and therefore no sense to keep it there. After making room i added the MXM structure according to the MXM specification, you can compare both *.dsl files which are attached to see the changes. The MXM structure has an buffer which is equal to the structure which was injected to the SBIOS. This solved the performance problem. To solve the LCD problem i added the _DDC method to the LCD and created an buffer with the EDID of the LCD. This solved the LCD problem.
Now, the fans were a bit more tricky. The DSDT doesnt have any Thermalzone and neither embedded controllers inside. Therefore the DSDT cant be the problem. The SBIOS could have some white list but no error at all. After some time i realized that the Fintek microchip which should control the fans is not used for that but instead Intel QST is controlling it. Well, dammit The Intel ME is not a part of the SBIOS but, i was lucky here, i found tools which could manipulate with the QST config. I looked on the status and realized that while changing the GPU the address of the fan changed and therefore the QST was in an emergency mod and set all fans to 100%. Well, dumped the settings, removed the GPU sensor and its done