Discussion in 'HP' started by 2.0, Sep 3, 2010.
So you don't think the Lenovo 3700 M would work better, or work at all?
I cannot guarantee its stable work, but I assume it should be working fine
Anyone knows how to get in touch with .NetRolller 3D?
He could definitely help a lot regarding the graphics card mode as he knows more about it. My skill is not enough to be able to find the existing MXM structure and replace it with the other one. Even though I could easily find FX3700M structures in 8730W (in SSDT) and Lenovo W700 (in DSDT). They can be found by searching for MXM_ signature in the BIOS file.
Everything is described in the MXM specification
I have read the stuff about INT 15h, AX = 5F80h which is described in the MXM specification and found some more details on this page https://forums.laptopvideo2go.com/topic/9098-mxm-beta-testers-wanted/?do=findComment&comment=28872
MOV AX, 0x5F80 (hex B8 80 5F)
MOV BL,01 (hex B3 01)
INT 15 (hex CD 15)
This is a function for calling a return of MXM structure. According to the MXM specification, the response to this code has to be
AX = 005Fh to indicate that the system bios supports this function
ES : DI = Pointer to the MXM structure in real mode memory
The assembly code for AX = 005Fh is MOV AX, 0x5F00 which in hex is B8 5F 00. And then should come "ES : DI = Pointer to the MXM structure in real mode memory". I don't know the hex code for this instruction as I have no idea what the pointer is and what the instruction looks like. But it doesn't matter because the problem is that I couldn't find any instance of the initial code B8 80 5F anywhere in the BIOS at all. Which means that the MXM structure is not called anywhere. Which most likely means that it's absent at all.
The required elements for the MXM support implementation are
1) MXM Structure (appears to be not present in HDX9000)
2) Int15h system methods
2.a. Function 0 (Return Specification Support Level) & Function 1 (Return a Pointer to the MXM Structure) are required methods (not present in the Dragon)
3) ACPI MXMI and MXMS system methods (not present in Dragon's ACPI tables)
I feel like Dragon's 8800M GTS is considered to be an onboard device rather than a normal removable MXM card. Nevermind that the video card doesn't have a BIOS chip on it. You know it's like an Nvidia or ATI chip soldered right on the system board. Same thing here, but wired out with an MXM connector without software implementation.
By the way, have anybody succeeded in changing Nvidia and ATI cards with each other? I discovered that BIOSes for Nvidia and ATI versions of the Dragon are somewhat different other than just vBIOS modules inside it.
If this all is true, then installing FX3700M as a removable MXM device is going to be difficult as hell since there's no trace of the MXM standard in the Dragon's BIOS and so it needs to be hardcoded from the start for an MXM vBIOS to be able to work there. It's not just about copy-pasting MXM related stuff from Lenovo W700's DSDT. It's also about setting up a required callback in POST module.
That’s a really nice research. In that case Lenovo W700 might not be the best MXM example, as far as I know it is just as custom as the Dragon. I would suggest something proven upgradeable, although MXM 2.1 was more hit and miss, where the MXM 3.0’s absolute champion is M15x-R1, only handicapped by lack of eDP (it has, however it is not implemented). Anyway best of luck.
In this case we could go for an MXM to PCIE adapter and install a normal PCIe card
Crazy stuff, so it's just a dream. Their project is dead, but could be a nice solution to our problem
There is an equal product up for sale on taobao, its marketed more for the AW crowd, but it exists, even in x8 and x16 interfaces as well, but its only for MXM3.0..
Also, you say that the card doesn't have an EEPROM for its VBIOS and its instead together wi
the system BIOS, well, Asus did the same thing with the G73 and other laptops that used their specific form factor MXM cards and you could upgrade them easily-ish, all you had to do was to replace the VBIOS in the BIOS with the one of the new card..
So, if you have some other MXM cards compatible physically with your system, and you have an hardware programmer so you can recover either your laptop BIOS or your MXM card, or both, try removing the EEPROM chip from the MXM card, then replace the VBIOS in the BIOS with the VBIOS from the MXM card and try booting it, with some luck it just works, unless someone already tried this and its a dead end..
Yeah, I know. But we could make our own MXM 2.1 adapter. Do you have a link?
Do you have any examples of such cards and some links?.
Done that already. Dragon's BIOS ignores MXM BIOS chip for sure. Removing the vBIOS chip didn't help at all.
The guy in the forum who claimed to successfully install either a 3700M or 2700M likely did so because he never claimed one hundred percent success, but rather a persistent issue. The cards ran, but the screen was dull… Lacking in punch. If he was making it up he wouldn't have made up such a problem. There is no motive for him, no business upside, for him to have claimed incomplete success.
Consistent with that account, he tested the installation and found that the card was only outputting 6 bits, rather than 8, which would explain the dull screen.
This is all unrelated to the fact that he badly treated a forum member, which is unforgivable makeup...
This was further established in my mind as a true account by my long conversation with the technician at Eurocom. He said the likely explanation for both of these issues… Dull screen and low bit rate… was an inverted pair of wires within the LVDS cable.
And of course the thing in the back of my mind was my conversation with HP technical support in Canada way back in 2008 where the technician had a client who had installed a 3700M. No reason for him to make that up.
Rather than say hardwired, I think it's more accurate to say that HP tried to make things proprietary at every step, as the usually do. And as a final kick in the pants, requiring a complete tear down of the machine to replace the CMOS battery, compared with virtually any other laptop, where it can take 5 minutes. I guess they didn't think that people would hang on to their machines for so long.
Well, you remember my photos
I don't wanna be a mythbuster here, but this is how it is.The picture doesn't look dull or anything like that. Again, according to my research, nothing was inverted there and I have posted the proof of that. Moreover, the video BIOS was taken from a 6-bit platform and you can see the output. Looks well with a basic color set. It just didn't let me boot into OS, just did reboot instead.
Those folks didn't provide any explanation of why they thought so. Why did that guy claim the thing about inverted LVDS? It wasn't true, so I assume they all were going a wrong way or some of them were not informed well. Also, I didn't see any proof of the work done. If you have it, please share.
It could be the case that he realized his mistake, but didn't want to admit it on public.
Separate names with a comma.