Hello!

I'm using HP HDX9000 with legacy BIOS. Basically there are several things among the others that I want to do: add quad core CPU support and unhide BIOS settings.
Here's a link to all the needed stuff http://www.mediafire.com/file/jazk5pu92u...0_BIOS.zip/file

Here's some basic information about HPUnpack, but I'm not sure if it's useful since I'm already providing the unpacked BIOS.
The tool is called HPUnpack which was originally created for HP 8710W. I've modded it for working with HDX9000.
Unpack BIOS file

HPUnpack bios.rom
 

Replace a certain BIOS module (e.g. POST module)

HPUnpack bios.rom replace 20100 POST 20100_POST_new.bin
 

20100 is the RAM address of the module.

Here's the BIOS structure:

First part ends at 0x1FFF0
The next part consists of compressed modules which start from 0x20000 and end with HPLO module at 0xB3E55 in the BIOS file.
The modules are
POST (0x20000) - (No idea)
POST (0x20100) - POST initialization
SETU (0x31000) - (No idea)
KBC (0x40000) - Keyboard controller
MESS (0x60000) - Text string messages
OSBI (0x9D000) - Startup Intel logo
NVID (0x9F000) - Startup Nvidia logo
VIDE (0xC0000) - Video BIOS
ROME (0xE0000) - Expansion ROMs
F000 (0xF0000) - Start of BIOS execution
MEBX (0x40000) - Intel Management Engine BIOS Extension
DSDT (0x65000) - Differentiated System Description Table for ACPI
SMM2 (0xA0000) - SMM (System Management Mode) setup to perform low-level system management operations while an OS is running using SMI (System Management Interrupts).
MBA (0xD0000) - Boot Agent PXE Base Code. Responsible for network boot
UNDI (0xD1000) - Universal Network Driver Interface. An application programming interface (API) for network interface cards (NIC) used by the Preboot Execution Environment (PXE) protocol. Used for network boot.
P16X (0x10000) - 16-bit PXE (Preboot Execution Environment), PXE Specification v2.1 or earlier. It is a network boot program that is run in DOS mode.
TPM (0x10000) - Trusted Platform Module. A dedicated microcontroller designed to secure hardware through integrated cryptographic keys. Used along with the fingerprint scanner
SEC3 (0x200000) - Responsible for chosing a bootable device
G8E_ (0xC0000) - Video BIOS (useless module, duplicated VIDE module, can be removed)
HPLO (0x90000) - Startup HP Logo (this is the last module)

The padding following after this module can be used for adjusting BIOS rom length after modding the modules.
Then CPU microcodes start at 0xC5BDC and end where the computrace stuff is starting at 0xCEBE0h.
All the other parts are unknown to me except for SLIC and TPM that can be found on the way down to the end.

The BIOS modules are all compressed, so the tool can compress/decompress any module and insert it back producing a new bios file called newBIOS.rom.
After opening the BIOS file in the tool, it will produce a set of files. Those are starting with 9D000 through F0000 are modules where .bin are compressed and .dec.bin are decompressed modules.
The header of compressed modules in the BIOS file consists of 16 bytes. Here's an example of one for the POST module starting at 0x20074h

01 00 14 01 40 E6 00 00 E4 A0 00 00 00 01 02 00
 

It has the following structure:

01 00 - module identifier?
14 - header length
01 - unknown (can be either 00 or 01)
70 00 00 00 - size after decompression 0x00000070
60 00 00 00 - compressed size 0x00000060. The compressed body of a module starts after its name which takes 4 bytes after the header
00 00 02 00 - destination address (0x00020000) in RAM after decompression

This is a basic information.
My question is about the APIC table which is located in the second POST module. You can easily find it in the decompressed module. I want to modify it for quad core CPU support, so I need to add some code to it, but it will result in a bigger size which won't then let the laptop start as the other offsets of the module will change. I can easily swap all the other ACPI tables inside that area, but if it comes to making the section bigger, it bricks the BIOS.
Can anybody find the way how I could increase the size of that section? It seems like relative addresses are used there, so it makes it difficult to do this.
I've tried to remove some useless ACPI tables, but the laptop froze during POST.

If you need more information about anything else, please let me know.

Are you still an active member of this forum?

Did you know you can unlock the HDX's GPU by changing device id?
Were you aware of this?

The site hosting the HDX threads disappeared.

I'm not so much active, but can answer some questions.
I cannot understand the point about changing device id. What are you trying to do and how?

Sorry for the late reply. I didn't think you would answer quickly due to low activity on the thread.

Thank you for your response.

First of all allow me to tell you that it is a real honor to exchange words with you.
I have followed much of the work you have performed on the HDX, and truly admire it.

It's a pity that all of that information is now gone. If you know of forums where other HDX-only threads are hosted, I would love to know them.

In regards to the nvidia GPU, I followed closely what you did with the Quadro FX 3700M mod. I was trying to understand if you were able to successfully change device IDs inside the system BIOS. It seemed you were close to finding a solution (since you had trouble at first), but in the end you apparently didn't need to modify the system BIOS when it came to device IDs only (forgive me if I got this wrong). I know you still had to modify it due to the MXM structure, etc.

What I was referring to is that I came across information from this site (I included a space in the URL because the forum doesn't allow me to add it):

https:// envytools.readthedocs.io/en/latest/hw/io/pstraps.html

If you scroll down to the very bottom of the page, next to the "G80+ families straps sets"; it talks about how in machines where the VBIOS is absent, the device ID is actually set somewhere in the system BIOS instead.

As you know, if we flash the system BIOS with a modded devide ID in the VBIOS, that will simply be ignored during system boot and the GPU will always be recognized as a 8800M GTS.

If I read that article right, it seems there is a way to allow the device ID change by setting it in the system BIOS. The problem is I have not been able to find the entry for it.

I was wondering if you knew if such an entry even exists at all in the HDX's system BIOS.

The reason for all this is because I believe the 8800M in the HDX could have the additional 32 shader units present but disabled. There is of course no way to know this with Rivatuner, because the tool requires a physical VBIOS to scan and determine if additional shader units are available.

As you know, the GTS and the GTX GPUs look exactly the same and are even clocked the same.
I was wondering if you knew anything about possibly unlocking those shader units in the GTS, or if they are at all present in the first place.

I don't think I ever saw you covering anything regarding this, but I could have easily missed it if you did.

There are threads online talking about how nvidia locked the device IDs of any given card (I forget starting with which GeForce series), all you had to do was to find the respective resitors on the board and change their resistance values.

It seems like specific resistance values are responsible for reproducing the desired device ID (though I'm not sure if this applies to the GPU series on the HDX).

I attempted this but without success. Maybe I played with the wrong resistors or as I mentioned, perhaps this is not even doable on the HDX.

I think that HP clearly had in mind to see what kind of mileage they were willing to give the HDX. The proof is in (I think) even having to "reserve" space for the G8E_ module in the system BIOS, which is totally pointless and a dummy copy of the VIDE module. I think they had in mind to upgrade the HDX with the 8800M GTX or in the future even one of the 9800M series cards. That could explain why the G8E_ module is even in the system BIOS, perhaps it was only meant to reserve the space for a real working BIOS for a different card.

Also, this choice of GPU without a VBIOS could discourage modders into finding easy ways to crack their plan, since many variables end up taking place in such an intricated scenario.

In the HDX this could have been very simple for HP, by swapping the resistors on the board or by inserting a different device ID in the system BIOS, the GTS could be turned into a GTX, without impacting production much since the same chip was used.

I would be glad to be proven wrong, so that I can simply move on with this since it's been in the back of my mind for a long time.

Additionally, if the shaders could not be unlocked maybe there is a possibility to unlock openGL Quadro capabilities by recognizing the card as a Quadro variant instead (maybe the FX 3600m?). Some threads on the internet explain how this scenario could take place for cards that cannot be fully unlocked into Quadro variants. Some instructions (but not all) could still be made available to the OS though.

Do you know anything about this that I talk about?

Your input and opinion on it would be invaluable, so thank you very much in advance for any feedback.

Zitat von bolo44 im Beitrag #4

---

If you know of forums where other HDX-only threads are hosted

---

Zitat von remdale im Beitrag #5

---

Zitat von bolo44 im Beitrag #4

---

If you know of forums where other HDX-only threads are hosted

---

I'm not aware of them or don't really remember. The Notebook Review forum was the best one.

---

Zitat von bolo44 im Beitrag #6

---

but the logo images are completely gone during boot

---

Zitat von bolo44 im Beitrag #6

---

found the 8800m schematics for the HDX video card (Orion 1.1)

---

Zitat von bolo44 im Beitrag #6

---

Are you sure you were able to recognize the FX3700M with its respective device ID at least once?

---

Zitat von bolo44 im Beitrag #6

---

In other words I'm positive that this override is preventing your card from being fully utilized.

---

Zitat von bolo44 im Beitrag #6

---

in case you no longer own a 8800M GTS card

---

Zitat von bolo44 im Beitrag #6

---

I know it is not a true windows bitmap and it should be a 4 bit - 16 colours image

---

Zitat von bolo44 im Beitrag #6

---

Is the BIOS in the HDX a AMI BIOS?

---

Zitat von bolo44 im Beitrag #6

---

Is the hardware upgreadability of the HP 8770W limited by its known architecture or, does it support newer generation GPUs?

---

Zitat von bolo44 im Beitrag #6

---

the presence of the G8E_ module

---

Zitat von bolo44 im Beitrag #6

---

If anything all I can help with is to try suggestions

---

Let me begin by saying that even though I am very invested in the HDX (as you can probably tell by the length of my posts), I consider this a hobby so I actually have to make time to put my hands into the laptop. Still, I gladly do so.

With that said, let me thank you not only for providing me with the schematics for the VGA board (this may be useful in comparing GPUs), but for the invaluable material you have provided to individuals like myself around the world on the HDX, including the download on your first post of this thread. Without all this information my HDX would simply be a fancy outdated system and nothing more, but thanks to what you (and others like yourself on the other forum) have done, I consider it so much more than that. Though many have thanked you before me, I now have the opportunity to add myself to the list and thank you for all that you have contributed so far.

I trust your judgement therefore clearly you do not need to test the card into a different system to sort out discrepancies. Like I mentioned in my other post: "I could be wrong" and indeed I am, so I apologize for making a supposition. I know this myself therefore not being able to test hardware whenever you most need to can be crucial to help sort things out. Thank you for providing that bit of information I was not aware of for your situation.

I will respectfully disagree with you on the PCIE revisions issue. First of all you are comparing PCIE with USB standards. That is like comparing apples to oranges. They are simply very different specifications.

More on that, like I mentioned in my other post I have performed the tests myself (because I couldn't believe what I was reading) between different PCIE revisions and lane speeds. The result (like I tried to mention before) is rather disappointing at least for "old systems" (this is not true for today's GPUs which can truly take advantage of large bandwidths).

If you had the time and the hardware, you could perform these tests yourself. This can easily be done with systems that allow you to cap the PCIE revision to 1.1, 2.0 and even up to 3.0 with also the ability to cap lane speeds between x1 up to x16. You could run all kinds of benchmarks to see what I mean for yourself. If you don't have the time, there are numerous (old) articles explaining how subsequent PCIE revisions (at least for "old systems"), didn't really bring the expected theoretical performance gains. Why that is? I personally don't know, but it's documented. In fact like I mentioned in my other post, I believe that manufacturers knew better, which is why we saw PCIE 2.0 GPUs in systems supporting only PCIE 1.1 or PCIE 2.0 systems with speed lanes capped at x8 instead of their full potential, coupled with GPUs that supported x16 speeds. You should be able to negotiate different PCIE speeds in linux (and windows too, but it's easier in linux) with a few simple terminal commands, if the BIOS of your test machine does not allow you to manually set speeds and revisions other than those that are set by default by the manufacturer.

The following is only a guess, so please only consider it that way: if on your HDX with the 8800M at default clock speeds (500/800) you run a benchmark, you will get a base score for those clocks. If you overclock the card to say 640/1000 (ddr speeds), your score will be higher and "not lower". You can of course verify that yourself. Remember that you are clocking the card up while the PCIE revision and lane speeds stay the same. Essentially the bandwidth didn't change yet you still get a performance boost. What does this tell us? That there is still room for performance gains since the PCIE bandwidth hasn't been saturated yet. This is why I only suggested in the other post, that perhaps there is more to be investigated in your case with the FX3700M (which is in a way very good news), since the FX3700M has more shader cores than the 8800M (only 64). Therefore at the very minimum you should get a slight performance boost, but the results you published indicated that the card was even performing slower.

As you know, the owner of an HDX in Italy was able to successfully install (though not without important cosmetic changes) a 280m into the laptop. I am not sure whether this individual got performance gains, but screenshots would suggest that he did. If this is true, the 280m used in the HDX supports 2.0 PCIE revision but was operating at 1.1 in the HDX. Moreover, the GPU has 128 shader cores and the card was overclocked to 630/1550, all running at 1.1 PCIE speeds and not 2.0.

Of course it doesn't help that this individual did not explain everything as meticulously and well documented as you are capable of doing with your mods, yet if a performance boost was indeed achieved, this is significant if you consider that the FX3700M is superior to the 8800M with higher shader cores and clock speeds. Although you are already aware of his mod I am posting it only for easy reference:

https://www.flickr.com/photos/78411477@N...ith/6875351918/

Again, it is unclear whether this individual experienced low performance the way you did with your FX3700M. It is also unclear why this so called (by me only) BIOS override in the HDX, may have only impacted your card and not the 280m. It was my speculation (so please consider it that way only) that this BIOS override in the HDX tends to lock down cards to GeForce variants rather than Quadro GPUs. Given that you remember having seen the card recognized with FX3700M device ID at least once, this entire speculation of having a BIOS override in the HDX goes out the window, which is why I asked you. Because I didn't know. So now at least I, can move on with this absurd idea. So thank you again for providing information I was not aware of.

Last on this, I would certainly agree with you: a USB 2.0 sound card would never work properly in a 1.1 USB port.

I would get a GTX to find out how it works, I was trying to find out if there was a way to have the HDX recognize the GTS as a GTX instead. The plan for the future will have to be this one, so if I do get one be sure to hear my findings on it. I would love to share that info with you.

I will try a 4-bit RLE-compressed bitmap for the logo and see if that works. Thank you.

***To the user who posted about the BIOS logo: thank you for your reply, I am not sure. I think one software definitely had metadata disabled, but there is a chance the other software I used didn't. So I will double check there.***

The reason I am focused on this G8E_ BIOS module, is only because its presence looks fishy...

I tried to explain why and you are welcome to go to the other post and re-read that part, but I think I did not do a good job in explaining it. Let me see if this way works better:

Let's say one morning you wake up and you decide you want to go out on a hike. Perhaps you'll only be out for 4 to 5 hours but you estimate that you could be out for up to 8 hours. If you are a cautious individual, you may want to bring with you more than just lunch. You may want to bring a snack as well. If at the end of the day it turns out that you were outside for 5 hours instead of 8, you only had lunch and did not need to eat the snack you brought along. Everything is good. But in the event your hike took longer than 5 hours, your choice to bring a snack was a wise one if you did. Wouldn't you agree? So, it's perhaps best to have more and not need it than to need more and not have it.

Again, you are welcome to re-read the other posts with this perspective in mind since I likely did a poor job in trying to explain it, so maybe this helps now.

Essentially what I meant is that people at HP (and this is only speculation on my part so please only consider it that way) likely asked themselves: what kind of mileage do we want to give this system? The answer to that was probably: we don't really know. We will tell the public that this system will be a limited production, but in reality we truly don't know so we will be cautious and we will "reserve" space (as in Kilobytes) in the BIOS for another VGA board, but of course we will put a dummy video bios instead of a potential working vBIOS because all we are doing is "reserving" space for something that we might implement. Chances are we won't do anything with this system, but we are being cautious and for now we are only testing the waters. This was all nothing but speculation on my part. This is what I tried to convey in the other post but apologize for having done a poor job.

I can add that having owned other HP laptops, at least in all the HP laptops that I have owned, the system BIOS always had at least more than one vBIOS. These modules were actually for working video cards, for example one for an onboard GPU (whether present or not) and one for a discrete GPU (whether present or not). Another scenario was also one for ATI GPU and one for nvidia GPU. What HP did with the HDX is a little different, but only because the HDX came out with an ATI-GPU-only at first and then with an nvidia one, but that was for a different motherboard because the first one didn't support the nvidia variant. So they put two different system BIOSes in a single download package, but when you examine the ROMs, the one for ATI cards only has one module for the video card whereas the nvidia variant has two vBIOS modules. One is a working vBIOS and the other one is a dummy. This last scenario (the one with 2 VBIOSes instead of just 1) is the one that I have always experienced with HP. This is the reason why I speculated that HP might have had other plans in mind, they simply didn't carry them on in the end. It also suggests that they possibly were going to completely drop ATI and perhaps only move forward with nvidia. As we know this simply never happened.

To your last point: I'm sure it must be frustrating working on this motherboard swap project all by yourself. Like I explained in the beginning of this post, I perform the experiments on my HDX only as a hobby and time allowing it. So believe me, I wish I could help with your project. I would if I could but I can't. I wish you good luck with it and, be sure to know that I am one among many who know who you are, that appreciate what you do.

I'm glad to help and was really obsessed with upgrading the Dragon until I have come up with an idea that I can create my own laptop. Let's see how it goes.

Zitat von bolo44 im Beitrag #9

---

This can easily be done with systems that allow you to cap the PCIE revision to 1.1, 2.0 and even up to 3.0

---

Zitat von bolo44 im Beitrag #9

---

That is like comparing apples to oranges.

---

Zitat von bolo44 im Beitrag #9

---

the 280m used in the HDX supports 2.0 PCIE revision but was operating at 1.1

---

Zitat von bolo44 im Beitrag #9

---

The reason I am focused on this G8E_ BIOS module, is only because its presence looks fishy...

---

Zitat von bolo44 im Beitrag #9

---

but in reality we truly don't know so we will be cautious and we will "reserve" space (as in Kilobytes) in the BIOS for another VGA board

---

Zitat von bolo44 im Beitrag #9

---

I am one among many who know who you are that appreciate what you do.

---

I am speechless.

You make my posts sound utterly pointless. That's why I'm speechless.

Healthy debate is considered good in a democratic world, so consider all that I am about to say as just that.

Who ever said anything about changing PCIE revisions in the HDX9000? It seems I spoke in vain. You should look more into performance between PCIE 1.1, 2.0 and 3.0 on paper and what that translates to in practical real world appplications. It can help you rule out dubious PCIE inconsistencies. (At the risk of sounding like a broken record) As I've said before: manufacturers knew better.

This is not even the most uself article (since there is no documentation on what was done), but certainly a handy one to try to show you what I meant:

https://linustechtips.com/topic/415550-8...mance-benefits/

PCIE 1.1 x16 is essentially 2.0 x8.

I've seen many laptops of that era (2008-2009) configured by manufacturers in strange ways that made no sense. Your FX3700M should have no problems whatsoever running at PCIE 1.1 x16, since it is hard for it to saturate PCIE 2.0 x8 bandwidth (therefore 1.1 x16. Do you now get what I meant?). I have owned an SLI 280m laptop capped (by the manufacturer) at PCIE 2.0 x8 (the cards of course supported x16), yet the bandwidth was still not saturated at x8, even after overclocking the cards. And I still complained at the time thinking I got ripped off by running them at 2.0 x8 only. Today I know that it simply made no difference for old systems.

You can of course find articles supporting this or not (the vast majority do support it), yet the best thing you can do is to try things for yourself and see what your own results align with the most. I did, and my results showed that PCIE revisions in older systems have little to no importance at all when it comes to performance because the GPUs of those times could only do so much anyway. Your experiments with the FX3700M seem to indicate that something else is causing the "lower than 8800M GTS" performace issues in the HDX. Perhaps you got ripped off with the card you tested but you simply didn't know it at the time. Or the card is not fully compatible, therefore not a suitable mod for the HDX even after all that work (sad to say it). Some people have had trouble installing 2.0 GPUs in a 1.1 motherboard. But for them the card simply did not get recognized at all. In your case the card runs, but poorly. It could even be some power throttling mechanism issue that you were not aware of. Your card required more voltage for the SLI connector only (as you stated) and you didn't supply that additional voltage to the board, but maybe some power check mechanism throttles down the card due to the absence of the extra voltage. You've studied the schematics so you've probably identified whether this is an issue or not, I'm assuming. Yet the card still runs poorly.

From your answers I deduce that the fault must be on my end though. I did not make myself understood not once, not twice but three times. That sucks.

My posts are riddled with typos, punctuation and grammar inconsistencies, but I thought the wording would at least be comprehensible. I'll do my best to try and make corrections to see if that helps.

In times when preserving anything Retro becomes more and more invaluable, I cherish my HDX just the way it is and will continue to explore undocumented ways to exploit its original architecture whenever possible. 10 years from today (if we're all still here) this thing is likely to go to a museum for being an original HDX9000, and not because it became something else. It's quite obvious you have long parted from this sentiment with your motherboard swap project (you are totally in your right to feel that way and I admire your diligence), to which again: I renew my best of luck wishes to you with it.

Trying to rule out hardware inconsistencies is never off-topic when BIOS modding, therefore never considered spam since a successful 100% BIOS mod depends on it. Your partial success (as in less than 100%) with the FX3700M is living testimony of this. With electronics you can't just go with a one-off scenario. Instead you are successful with consistency. I'm not implying you should buy hardware you don't want or need just so you can run additional tests, but the protocol to go by would have to be the one I hinted at before and outline here:

1) You should test your FX3700M to find out if it is recognized 100% as an FX3700M in a system other than the HDX. That would disprove or confirm your assumptions on whether the VGA board suffered damage somehow. Either outcome scenario is good so that you can move on to the next two.

2) You should then benchmark the card as much as possible to figure out if you have a real FX3700M in your hands or some chinese BGA chip-swap.

3) Compare its performance to that of other FX3700M cards to rule out whether you have an authentic but faulty GPU, as this too can happen as we know.

I hear you... it's a pain, but nothing compared to the stuff you have demonstrated to be capable of or even to your motherboard swap project.

If you can take my opinion, so far your results seem to indicate that you might have indeed been ripped off with the purchase of that card. Likely whoever sold it to you gave you a bogus 260m disguised as an FX3700M. The net is full of such nightmare stories and it seems you may be just another victim. There is only one way to know, but that option is off the table so you'll never know for sure.

To rule out whether all this is off-topic or not, here is your post on the FX3700M mod (in part, a BIOS mod which ties in perfectly with the same tool you used in the mod and attached to the first post of this thread):

https://web.archive.org/web/202201250423...15212/page-1043

A real breakthrough in HDX9000 history. Truly remarkable.

Like I said before BIOS and hardware modding is only a hobby for me (I am not as hardcore as you are), but one that I have tried to understand as much as I could over the years. Reason why I turn to more experienced users to ask questions, particularly if I have a strong feeling they might know more than me. Many times I've gotten things right and even more times I've gotten things wrong, but in the end it's all a good learning experience.

I have a feeling you'll have something to add but I leave it here because I have too much respect for you, and I don't want to find a huge disappointment by viewing you any differently.

It was actually nice to finally meet you online.
Take care.

Zitat von bolo44 im Beitrag #11

---

you didn't supply that additional voltage to the board, but maybe some power check mechanism throttles down the card due to the absence of the extra voltage

---

Zitat von bolo44 im Beitrag #11

---

Some people have had trouble installing 2.0 GPUs in a 1.1

---

Zitat von bolo44 im Beitrag #11

---

1) You should test your FX3700M to find out if it is recognized 100% as an FX3700M in a system other than the HDX

---

You did plenty by sharing the VGA schematics. I appreciate it.
Boardview would be cherry on top :)
I doubt that one is around though.

Not a problem, here's the link to the whole folder
https://www.mediafire.com/folder/7iaie3rkkmohz/GTS_8800M
Let me know if you need anything else

That is absolutely great.
Thank you so much.

If I find anything worth mentioning, I will keep you informed.