Modifying / Rewriting a BIOS?

[ArchStudent]

Hello,

I was just curious if anyone knew who would be able to rewrite a BIOS or modify it, and if it were even possible to do so on a notebook? I am curious to know if something could be added to the notebook's BIOS (Intel speedstep technology).

I've written about this to death in another forum. I upgraded my notebook that had a micro-pga2 Mobile Intel Celeron 450MHz to a Intel Mobile Micro-PGA2 PIII 800MHz with SpeedStep technology (and it only runs @ 650MHz, which is the lower clock from the speedstep).

Now the notebook was only tested to work with the early Mobile PIIIs that were 450MHz-500MHz and did not have speedstep technology with the split voltage, and Micro-PGA2 Celerons up to 900MHz which never had speedstep.

Now speedstep cpus have a split voltage (ex. my 800MHz PIII runs at 1.6V, and when it is clocked at 650MHz it runs at 1.35V).

Now this notebook was never tested with speedstep, but using common sense I figured that any micro-pga2 cpu should work... and I proved a few people wrong with this by popping in a PIII 800MHz and it worked right away (only at 650MHz though). I am still quite happy with the outcome, and a PIII 650MHz is a MAJORmprovement over the Celeron 450MHz....

Now if anyone could point me in the right direction I would appreciate it. I think that if I could find someone to rewrite or modify the BIOS to recognize speedstep it would be really cool! If this is too difficult a task then I'll stick with what I've got

Cheers

 

[Peter]

Adding stuff to the BIOS requires access to its source code, a license to do so, insight into the mainboard's schematics, and of course the skill set to make sense of all that. Forget it.

Furthermore, Speedstep is not just BIOS software, there's hardware involved too. If it ain't there, it's not going to appear out of thin air.

But with the original Speedstep in P-III being rather primitive and driven by a single pin on the chip, you should pursue tricking the CPU into Fast mode by wiring this particular pin the right way round.

 

[ArchStudent]

Originally posted by: Peter
But with the original Speedstep in P-III being rather primitive and driven by a single pin on the chip, you should pursue tricking the CPU into Fast mode by wiring this particular pin the right way round.

Okay, that seems like a novel idea. I was just curious if it were possible to add or modify the BIOS, but this sounds like an easier alternative.

Where would I find the information regarding this procedure, so that I may attempt it? What tools would I need to accomplish this, and do you have any success stories with this kind of procedure?

Cheers

 

[tinyabs]

Most motherboard has bios update function, so it is technically possible to modify bios and update it. You can hack a bios but it will take deep technical knowledge to do a proper hack; the hardware IO stuffs are undocumented.

Some virus even corrupt your bios in the past.

 

[Peter]

ArchStudent, go to developer.intel.com and fetch the tech docs for the mobile P-III. That'll point you to the speedstep mode sense pin and its properties.

 

[ArchStudent]

Alrighty, I found a few documents that have to do with the mobile PIII, and one that has a diagram of all of the pins on the underside, but does not clearly(to me anyways) explain what each pin is for...

I'll be honest, I am treading into an area I haven't been before, but want to learn and give this a go. What tools would be necessary for this procedure as well?

If I included a link, do you think you might be able to help guide me if you are not too busy?

Thanks for all the help

 

[borealiss]

bios code, especially closed source, is a tricky animal. most bios' are decompressed from rom, so you'd have to extract the raw image as it's loaded from the lpc bus. even then, some bios vendors encrypt their code. there's also the driver interaction you will have to deal with, and understanding the interaction between the cpu and the rest of the system in doing a frequency change. you're doing a voltage and clock ramp, so you're going to have to know what delays to expect/program in the firmware, and this is a major problem in mobile systems that do dynamic clock ramping. if you're not familiar with clock settling delays or how the whole disconnect process works on an intel platform you could get corrupted data across the bus, or latch on garbage data. this is usually programmed in acpi code that is present in the bios. and that's another thing, knowing where in the acpi tables to program this stuff. i think this is pretty much an impossible task if you do not have access to any nda documentation or the bios source code.

 

[ArchStudent]

Originally posted by: Peter
ArchStudent, go to developer.intel.com and fetch the tech docs for the mobile P-III. That'll point you to the speedstep mode sense pin and its properties.

Thanks for the help Peter. I did download the tech document for the mobile PIII, but I can't seem to discern which pin controls the speedstep mode sense.

 

[Peter]

Point me to the document that lists the pins and their function.

 

[Lynx516]

OK the Speedstep pin is GHI# and to get it into the faster bus ratio you need to drive this with an open drain with no pull up. I.e. you need to connect it to gnd if you want the faster speed mode. That is if I read it correctly .

Peter if you want to check up on me the doc is This one

 

[Peter]

Right. However, as chapter 2.2.10 indicates, tying GHI# low will not do anything, since the CPU always fires up in "battery optimized" mode. GHI# state will only be recognized upon wakeup from Deep Sleep processor state.

This means the pin needs to be connected to whatever general purpose I/O pin is available on this particular mainboard design (through a voltage converter that brings it from 3V down to the required 1.5V signalling, apparently); and BIOS ACPI code needs to be written that does the transition by operating the pin and then triggering a sleep/wakeup sequence.

 

[ArchStudent]

Lynx516 and Peter thank you for your replies

This means the pin needs to be connected to whatever general purpose I/O pin is available on this particular mainboard design (through a voltage converter that brings it from 3V down to the required 1.5V signalling, apparently); and BIOS ACPI code needs to be written that does the transition by operating the pin and then triggering a sleep/wakeup sequence.

I think I get what you are saying Peter and do you mean to say that there may be more to it than playing with some of the pins on the CPU? BIOS may need to be "modified", and some other changes to the mainboard?

 

[Peter]

Yes. The way things look from that datasheet, you need that pin to be programmable, and you need a BIOS mechanism that operates this. It would have been easy if the processor read the pin state upon powerup to decide about its default state - but it doesn't.

 

[Dinominant]

Check out BIOSMods.com
Maybe sombody there can help you

 

[jhu]

on some motherboards you can replace the bios with linux.

 

[Peter]

LinuxBIOS or other BIOS, it needs to be written for the very machine it is going to run on - down to the finest details of the mainboard layout. This is why LinuxBIOS is limited to a handful of very popular, simple to reverse engineer, and long-term manufactured mainboards.