Acer MXM BIOS Mods and discussion (was: x920G vs. 9600M GT)

do you have an exact link to the 9300m bios? cause I can't find it on the techpowerup's website

 

The latest 9300M VBIOS is request download ticket | 9920g_9300mgs.rom - ifile.it

@mushu666: Yes, a cross-flash between GS and GT is possible. But unfortunately, I don't have access to the HP 9600M GS VBIOS.

 

netroller3d:
sorry for the late response,was on a trip...
this is your request for the image with the 9600m gt:
request download ticket | 18112011169.JPG - ifile.it

i also took pictures what happens when i try the 8600m gt with the bios for the 9920g on a normal laptop:
request download ticket | 18112011171.JPG - ifile.it

and when you flash the 8600m gt with a normal bios:
request download ticket | 18112011173.JPG - ifile.it

it deffinetly looks like a problem with the gpu bios and not the sbios at the moment

as you can see from the pictures above any normal screen goes nuts when you try the 8600m gt with the 9920g vbios.
i tried on 3 different laptops and the result is allways the same.

if you can solve this problem i will gladly buy a 220-260m card to try it out

thanks in advance

ps: please don't make me change the cards another 10 times because its driving me nuts.....

ouh yeah,as for the 9600m gs vbios,i was unable to track it down on the internet so i can't confirm anything...

i have tried all the bioses that are out there,9600m gt and all 9700m versions from all the manufacturers and no luck

 

is't possible to flash the 4920g with the 5920g bios?
trying to get it to work with the 9600Mgt from MSI gx620,

 

Nope. The 4920G is a completely different system, made by Wistron. The 5920G is made by Quanta instead.

The 4 720G is possibly flashable to 5920G (they are both Quanta machines, and very closely related ones).

However, I'll try to make a mod for the 4920G.

 

@net : news on fix for 9920g and bad color on internal display ?

 

Someone needs to buy a HP 9600M GS (blue PCB) from eBay (or otherwise obtain one - this is the HP TouchSmart IQ816's card, also used in the IQ515), and dump its VBIOS. With that done, I can make a fixed VBIOS for the 9600M GT.

 

@netrolller3d,
it's kinda wierd that the Ati card works with the Msi, while the nVidia card dosn't work with Acer, Wanted to try flashing 9600mGT vBios to a 9600mGS or even acer 9650mGS, If there's a workaround the Acer bios not booting with nVidia..

 

@ pcsavvy:
all mxm cards you can plug in the msi work,at least for my gx620....
even the mxm II he tipe....even the quad core processor x9000 works...to bad i didn't have a 9920 at the time to test the x9000 with it,but i read one person succesfully integrated a x9000 on a acer 9920...

 

is this similar to the HP 9600M GT vBios?

HP Quadro770m

 

9600m gs\gt,9650m gt,9700m gt,fx770 are all the same card,same board same gpu only different clocks and id.
to save you the effort it doesent work,i tried all the vbioses out there i could find,for ddr3 memory of course...
only with 2 or 3 i got the discoloured image,the others black screen or total freeze on boot...

be carefull when flashing not to flash the vbios for ddr2 on a ddr3 memory card,or you would need to do a blind flash after with a crisys disk.

 

The 770M VBIOS comes from a machine with a 6-bit display. What's special about TouchSmarts is that they contain an 8-bit FPDI display, just like the 9920G. (MXM officially supports 6bit and 8-bit OpenLDI displays only.)

BTW, the X9000 is not quad-core.

 

@blackended87
what are the vBios that gave you discoloured images?
perhaps I could give it a try on the 4920g..

 

you have the link on the page 30-something of this toppic.

 

@pcsavvy: It won't help. The problem with the 4920G is the lack of an MXM-SIS in the BIOS. The 9920G, on the other hand, fails due to a non-standard display panel (unusual LVDS encoding, no EDID).

 

@netroller:
is there anything else we can do?
i allready have a bunch of cards...so i don't need any more....

is there a way to determine how the factory 8600m gt bios does the encoding...
or the mentioned 9300m...

i am a hardware tehnician not a programer lol

and you saw the image you gave me looked normal on the display,as the rest was blurry and discoloured all around.

 

Yes, that confirms that the card is unable to detect the display as FPDI. MXM cards were never meant to be used with FPDI displays; FPDI is designed for the inside of LCD TVs, and should not be used in a notebook.

 

I've got similar problem. My laptop is Aspire 7720G. I bought GF 9600M GT and installed it to my computer. But after I turn the computer on, the screen is black all the time. The computer itself is working - I hear the sound of windows loading etc.
I have no HDMI plug in this computer, only DVI and typical VGA.
Does anyone know how to deal with it? Would BIOS-flashing help?
Greetings
Mark

 

The 7720G supports 9600M GT out-of-the-box. Your card is likely defective.

 

Yeah, at a first time I thought so, beacause I have already had GF9600M GT and it was working quite properly. But the card that doesn't work is the new one and I noticed that it has MXM II - connection slight different than the older (working) card. Don't you think this could be a problem?

 

Unlikely, unless it is because the card is not making contact with the slot.

Can you check (using blind-flash techniques) if nvflash is detecting the card? (If it is, then you may need to flash a different VBIOS. If nvflash also shows no card, then it is dead.)

 

I installed the modded 1.20 bios for my 9920G and I updated the 9300gs card with the modded bios. It's ok until I install the laptopvideo2go drivers but after the reboot the screen is full with vertical colored lines and distorted images, do you have any idea what could go wrong?

 

I confirm that the updated Aspire 5920G BIOS works with the 9600M GT flawlessly. Thanks, .NetRolller 3D, great job, respect.

 

@sowi: Sounds like the driver is deluded into treating the internal display as a 6-bit, single-link panel. Which is rather weird, since I clearly remember seeing and fixing this issue when I made the BIOS.

EDIT: Try this BIOS: http://ifile.it/9i2bv1z/CV120_MXM_V3_BETA.WPH

Apparently your version of the driver fails to read the fallback EDID from ACPI if a display is specified as having its own EDID, but reading that EDID fails. This BIOS marks the internal display as EDID-less, which will hopefully force the driver to read the ACPI fallback EDID.

 

i tried, but unfortunately still the same result
check out the pics please:
http://i42.tinypic.com/10zwazb.jpg
http://i42.tinypic.com/2i7wbco.jpg

 

Try a different driver version.

BTW, is every 2nd vertical line black?

 

nope, strange moving colors on the screen and it's constantly vibrating and moving
I tried a few other drivers too, but nothing changes

 

Identic problem with bad colors. In the bios, necessary activated a option or not ?

 

@sowi: So, the driver thinks you have a 6-bit dual link panel.

@momol51: Do you also have moving, vibrating colors, or just mixed up (but static) ones?

(BTW, new 5720G BIOS <s>coming soon</s> out now. HP HDX9000 to follow after that.)

 

yeah, you are the expert should I try to find a correct driver?
I tried a few from the nvidia page and from the laptopvideo2go.com

 

I think something is overriding your DSDT. Can you post the current DSDT Windows is using, with XP_DSDT?

 

color fixe but vibration and bad colors on 9920g

 

Hi

Trying the latest "Black Edition 4.0" Bios

( http://ifile.it/3wlxm6j/icl50_1.45_black_edition_40.fd)

And looking DSDT ;O)

I found 1 "error" :

940 Warning Result is not used, operator has no effect

Concerning this Line of code :

{
If (LEqual (Local1, Zero))
{
And (CAPB, 0xFFFFFFFC) <--- This one
}
}
Else

I make this Correction to obtain 100% "errors" Free (a warning, yes ;Op) :

And (CAPB, 0xFFFFFFFC, CAPB)

The zip file contain .asl and .aml

Voila

Precision :

I'm using the latest IasL win Bin version, perhaps this is why, i find this latest "error"
This is the BEST DSDT so FAR, i've seen in a bios , B-I-G Congrats to NetRoller and other Bios guru who made it !
Thanks Again and again !

Regards

 

@.NetRolller 3D
thanks for new bios for icl50_1.45_black_edition_40 and efi shell interated I guess
does efi boot work on gpt/usb disk or just have a shell only

regards

 

Hi all,

I have a 9920G with a GPU that has given up the ghost as far as the laptop screen is concerned (The external display to a monitor is still working OK).

To replace it I purchased a Nvidia GeForce 9300M GS having read on this forum that it can be used to replace the original Graphics card in a 9920G.

If I have understood some of the threads on this forum right, the 9300M should work in the 9200G after applying the modified System BIOS (SBIOS) which is in the first post of this thread.

Assuming I have got the above info right so far, I am now faced with 2 Problems:

1) Windows refuses to start without the original Graphics Card in the 9920G. Will this make a difference to the flashing process, or should I just be able to swap the Graphics cards after I flash the BIOS ?

2) I am using WinPhlash, which backs up the existing BIOS OK, but refuses to write the new BIOS and throws up the error:
BIOS update failed! Image verification failed
BIOS Image checksum verification failed.
Error code – 104
Anyone had this before, or know what might be wrong ?

 

@osmoTR: Only the shell can be selected from the boot menu. You have to start the EFI boot loader on the USB drive from the shell.

@jimmyjimmy: 1. Flash with the original graphics card installed, then replace the card.
2. Your BIOS image was downloaded incorrectly; try redownloading. (I'll post MD5s later, thanks for pointing this out.)

 

@.NetRolller 3D
Thanks very much for you help with this and for all your hard work in getting the replacement cards working in the 9920G!

UPDATE:
> I have now flashed the system BIOS (SBIOS) on the 9920G and all went OK.

> I then swapped the old, broken GeForce 8600M GT for the new 9300M GS.
When I started up the computer I got some of the crazy colours described
elsewhere, so I updated the 9300M VBIOS also.
Again, all seemed to go OK.

> Upon restarting the computer, the colours is fixed and the laptop screen is working OK. BUT...
> I Still have a problem. I am using Windows 7 and once the startup goes past the windows flag, [Where it says "Windows Starting"] the internal laptop screen then cuts out.
The external display still works OK though.

Any idea what might be wrong ?

Thanks

 

What happens if you try to boot without an external screen plugged in?

 

The screen just remains blank.

It is completly normal up until it says "Starting Windows" and then it just goes blank.

Is there any chance you could repost the 9300M GS VBIOS so I can try reflashing it?

 

@.NetRolller 3D

thanks for great work of bios hack
I have a question about built-in efi shell
in the original_bios there is a string about /boot/efi/bootia32.efi
it's external boot reference for shell/boot_loader
without builtin shell can be boot a shell or boot_loader via directly from hdd/usb

is it possible to hack external shell booting feature so I ask why to open directly to the boot loader don't messing with builtin shell

Sorry for my english if it is bad

Regards.

 

@jimmyjimmy: Weird.

Can you save your DSDT using XP_DSDT? I suspect that something is overriding it.

@osmoTR: I've tried to activate that one, but unfortunately the only reference to that string is in a function that never gets called.
It is likely a leftover from the original Intel Tiano code on which InsydeH2O is based.

In Tiano, bootia32.efi gets called by the "generic" EFI boot manager code, after the "platform-specific" code finishes its job. However, InsydeH2O's "platform-specific" code never terminates normally, handling the entire boot process itself, going into an infinite loop if it fails, so the "generic" boot manager is never reached.

 

UPDATE : It is working OK when Windows 7 is booted into Safe Mode (Even up to the maximum resolution of 1680 x 1050) but when i try and boot Windows normally, only the external screen will work!

 

@jimmyjimmy: That's a DSDT bug. Please save your DSDT table using XP_DSDT.

 

Hi @.NetRoller 3D - Thanks again for the help.

I'm new to all this, so I'm not sure if I done this right, but output is attached.

Had to upload them as txt files!

Thanks

 

@.NetRoller 3D

it's gonna be long post i guess sorry for my english

I think Insydeh2o UEFIs handles if uefi/if legacy boot process something different way
interesting part is a sony vaio(insyde bios with uefi disabled) shows it can boot via uefi mode but must be without internal hdd/odd connected
there is link : http://forum.notebookreview.com/5448774-post29.html

but perhaps our firmwares are very old comes from year of 2007
there is insydeh20 press release about our bioses(icl50hw) :
Insyde Software's Firmware Shipping on Acer (Aspire 5720) based on Intel® Santa Rosa Customer Platforms

but funny think in the press they writen "Native UEFI boot to the Operating System (legacy boot also enabled to support all Oses)."
but there is no uefi boot support in default bios you know that

and you used efi_shell comes from EDK_v1.05 but in the press they said: "UEFI 2.0 Specification ready"
EDK_I or EDK_II which version supported via icl50hw

and one more question i have
how you can insert efi_shell.efi into to firmware what mod you did for boot support i want to use for my custom firmware i don't wanna use extra_menus+mxm_fix+pei_fix
i want to use more closer stock bios
i have already a firmware custom made which is it's updated vbios+merom-l microcode+new_acer_logo..vs,
but i don't know how a new module inserted and preparing corrections to boot support
can you give instructions of uefi_shell_mod

Regards

 

@jimmyjimmy: That DSDT does contain my changes; but it is full of fatal errors (at least for the decompiled version). Looks like Windows 7 is unable to correctly interpret ACPI v4.0a AML code. Weird, since Vista, which is older than 7, had no problem decoding the same DSDT on my test system.

@osmoTR: The Sony VAIO Insyde BIOSes are all 64-bit, while this one is a 32-bit EFI. The reason why they can't EFI-boot from USB is likely a lack of EFI USB drivers.

As for the EFI shell, I actually took it straight from the EDK package; it was not part of the original BIOS image. However, the version of InsydeH2O in the 5720 is definitely not UEFI 2.0, but Intel EFI 1.10.

I used EzH2O v2.1.0.4 to insert the shell module, but I had to hand-craft an FFS container file for the Shell.efi binary that shipped with EDK before I could insert it. In addition, I also had to insert some USB drivers; which I compiled myself from the EDK sources.

I think it should be easier to just remove the unwanted changes - PEI fix (PLL center spread) can be reverted by simply not flashing the PEI part of the BIOS; the MXM structure is in module 183f9ba1-f9de-446b-8db1-2b025ee8fa50 (but having it around in an Intel GMA system will have no effect), and the extra menus are in SetupUtility (fe3542fe-c1d3-4ef8-7c65-8048606ff670) - note however that SetupUtility also contains a change (a byte from 80h to EEh) that is needed for the EFI shell. Also, removing the menu mod also kills Intel VT and dual-IDA capabilities; plus, one of the originally hidden menus contains the "USB mode" option needed for enabling USB access in the EFI shell.

 

Any idea what else I could try ?

 

@.NetRolller 3D

thanks for extra info
I did some small modifications on firmware my logo,vbios,microcode
but big hacks not required for now I don't have much time to hack the firmware anyway for now
I am happy with your mod_bios
thanks for your great work

but small errors I have DSDT related I guess
on ubuntu 11.10 dmesg shows like those errors maybe helpful

Code:

[    0.297431] ACPI Error: [CAPB] Namespace lookup failure, AE_ALREADY_EXISTS (20110413/dsfield-143)
[    0.297439] ACPI Error: Method parse/execution failed [\_SB_.PCI0._OSC] (Node f5825c78), AE_ALREADY_EXISTS (20110413/psparse-536)
[    0.297449] ACPI: Marking method _OSC as Serialized because of AE_ALREADY_EXISTS error
[    0.297457] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])

[    0.309212] ACPI Error: [CAPB] Namespace lookup failure, AE_ALREADY_EXISTS (20110413/dsfield-143)
[    0.309220] ACPI Error: Method parse/execution failed [\_SB_.PCI0._OSC] (Node f5825c78), AE_ALREADY_EXISTS (20110413/psparse-536)
[    0.309233]  pci0000:00: Requesting ACPI _OSC control (0x1d)
[    0.309305] ACPI Error: [CAPB] Namespace lookup failure, AE_ALREADY_EXISTS (20110413/dsfield-143)
[    0.309312] ACPI Error: Method parse/execution failed [\_SB_.PCI0._OSC] (Node f5825c78), AE_ALREADY_EXISTS (20110413/psparse-536)
[    0.309324]  pci0000:00: ACPI _OSC request failed (AE_ALREADY_EXISTS), returned control mask: 0x1d

Regards

 

@jimmyjimmy: I had the same problem with the original microsoft 9300GS driver

 

@jimmyjimmy, @sowi: I think I found it. Apparently the ACPI _DDC method needs to be exposed through the WMI WMMX interface to get the Nvidia driver to read it - for some reason, it does not call ACPI directly, but only through WMI, even on Linux (the "W" in "WMI" stands for Windows).

@osmoTR: Will be fixed in release 4.1, coming soon. (I noticed some more serious DSDT issues in fact - Windows 7 completely fails to restore from hibernation, apparently because I used an ACPI v4.0a AML comiler, and Windows 7 only supports up to V3.x. For curiosity's sake, I also recompiled the DSDT with the newly-released ACPI v5.0 iASL - that one doesn't even boot on 7.)