How To: Enable SLI on pre-i7/i5 hardware

[bluevelvetjacket]

From the way you talk about the code I'd almost suspect you might fully know how to bypass the SLI routines W1zzard but haven't done so for legal or motivational reasons . I really wouldn't doubt it since you are the one who has done so much digging in the driver code to develop GPU-Z and looked at specifications.

And I've tried many routes also, but what do we breakpoint hmm - and when?

 

[Steevo]

check your check?

 

[W1zzard]

if you claim you know where sli is enabled but there are so many code integrity checking functions just run with a debugger like softice. put a breakpoint on memory read on that code location so the debugger stops every time something tries to read from that memory address. then you sit in the code integrity routine because no other code will access the instruction bytes of the sli enable function.

 

[Mussels]

if you claim you know where sli is enabled but there are so many code integrity checking functions just run with a debugger like softice. put a breakpoint on memory read on that code location so the debugger stops every time something tries to read from that memory address. then you sit in the code integrity routine because no other code will access the instruction bytes of the sli enable function.

none of that means a thing to me, but i thank you for sharing that advice.

 

[opieum]

I know this thread is a bit dated, But is there any progress on this? Or is the attempt being abandoned until a cracker decrypts the routine? Apparently ASUS and certain other partners have unlocked versions of these drivers. So there has to be something that makes it possible at the driver level. I am guessing a BIOS hack is probably not the way to go as has been suggested elsewhere.

 

[Solaris17]

to many changes to the drivers this project is on hold.

 

[opieum]

Hmm shame. I am guessing once the x58 support is released it might be a little easier to hack? It would seem so since there will be setups in the wild that will be easy to compare against.

 

[Solaris17]

Hmm shame. I am guessing once the x58 support is released it might be a little easier to hack? It would seem so since there will be setups in the wild that will be easy to compare against.

no need to hack it in X58 both ATI and SLI are supported through physical chip and bios tattooing

 

[Mussels]

no need to hack it in X58 both ATI and SLI are supported through physical chip and bios tattooing

i think his point was supposed to be that with x58 supporting it natively, people could simply compare x58 drivers to other chipsets, and make a hack to cover the changes.

 

[Solaris17]

i think his point was supposed to be that with x58 supporting it natively, people could simply compare x58 drivers to other chipsets, and make a hack to cover the changes.

the drivers are no diffirent they just changed the integrity check methode to check for the chip (hasnt changed) and to check the bios (has changed) in eather case i cnnot hack a bios on the board to do that...i had originally thought i cold do his but apon attempting it with my early tests i came to the conclusion that not only was i unable to isolate the string in the bios that allowed it but i would need to mod the bios of every single system that someone asked me to mod it for because chipsets bios types and sizes vaery so much between boards that their was no way for me to make a "generic" change

 

[Mussels]

the drivers are no diffirent they just changed the integrity check methode to check for the chip (hasnt changed) and to check the bios (has changed) in eather case i cnnot hack a bios on the board to do that...i had originally thought i cold do his but apon attempting it with my early tests i came to the conclusion that not only was i unable to isolate the string in the bios that allowed it but i would need to mod the bios of every single system that someone asked me to mod it for because chipsets bios types and sizes vaery so much between boards that their was no way for me to make a "generic" change

look at the vista OEM hack, that was a bootloaded that faked BIOS stuff. it may be possible to do the same.

 

[Solaris17]

look at the vista OEM hack, that was a bootloaded that faked BIOS stuff. it may be possible to do the same.

good call but all in good time i have too many projects atm.

 

[opieum]

Yea that was my point. To do a comparison. It should be MUCH easier to do a code comparison. I may wait till next payday and pick up an X58 setup to test out this theory. BIOS hacks of any kind almost seem more trouble than their worth BUT seeing how it worked so well at between the BIOS and OS level with Vista and even Mac OS (different implementations but same idea) it might be a more lasting solution to try. Hell if it could be rolled into the default Vista setup then it makes it so no matter the driver version (because if the key gets changed then it just a matter of installing a patch that updates to a new working key)

Solaris, if you have any pointers and what I could do to get started, I would be happy to try at this with some guidance It would be easier to just get this going over all. PM me if you have any docs or info that I can reference to get started (I have some ideas but anything you have with the work you have done is welcome as well)

 

[opieum]

HA Here is an update. Via what should be called a penecillin hack It was discovered by accident.

http://netkas.org/?p=91 This is a hacker for the Hackintosh scene. But he was working specifically on a hack for EFI and DSTD to enable a Mac Retail CD to boot unmodified and found out how the certification works. Have fun. Sounds like this could be universally applied (seeing as you can run Mac OS on white boxes because of some of his hacks) Mind you the discovery is OS independent. So that should apply across the board.

 

[ShadowFold]

So you can SLI on a 790FX/GX board with those?

 

[zerg32]

I've repeated the steps from that post for GA-EX58-UD3R-SLI bios.

There is an interesting piece of code:
If (LEqual (FUNC, 0x41494C53))
{
Return (\_SB.PCI0.SLIC)
}

I've looked in the nvlddmkm.sys using IDA Pro for this hex code.
There is something that looks like a WMI call with this code. I'm not able to trace what this function does, cause I'm not experienced IDA user. But in my opinion for IDA/Assembler Guru it'll not be a problem to find how the protection works.

But for X58 it should work already, you just have to modify the acpi table. But I dont have required hardware to test with.

If anyone is interested I can write a short guide how to unpack the bios and edit this acpi tables.

 

[cdawall]

i'm interested on getting SLi running on my 790GX board any guide would be amazing

 

[Fitseries3]

can you hook me up with quad SLI on p45?

if your really talented maybe you can get septo SLI working for me

 

[BUCK NASTY]

can you hook me up with quad SLI on p45?

if your really talented maybe you can get septo SLI working for me

Mine's easier. I need just regular Sli on my p45. It would be nice now that I have dual 260's:shadedshu

 

[zerg32]

I'll try to write a guide for X58 boards, but it wouldn't be very usefull, because many of them have already sli support.

Is here anybody with X58 board without sli support wanting to try modded bios ?

P.S.
I have P45 myself and hope the some guru will hack the driver.

 

[opieum]

So you can SLI on a 790FX/GX board with those?

The general idea is that you can edit your BIOS to add this in. I would save the page just in case it goes down. Still this is great. I like nvidia cards, but hate their chipsets.

 

[eidairaman1]

if you hadnt read the news Gigabyte Recently enabled SLI across their Entire x58 Chipset Boards.

I've repeated the steps from that post for GA-EX58-UD3R-SLI bios.

There is an interesting piece of code:
If (LEqual (FUNC, 0x41494C53))
{
Return (\_SB.PCI0.SLIC)
}

I've looked in the nvlddmkm.sys using IDA Pro for this hex code.
There is something that looks like a WMI call with this code. I'm not able to trace what this function does, cause I'm not experienced IDA user. But in my opinion for IDA/Assembler Guru it'll not be a problem to find how the protection works.

But for X58 it should work already, you just have to modify the acpi table. But I dont have required hardware to test with.

If anyone is interested I can write a short guide how to unpack the bios and edit this acpi tables.

 

[opieum]

if you hadnt read the news Gigabyte Recently enabled SLI across their Entire x58 Chipset Boards.

Right but the point is the certfication or the so called "cookie" has been found and could be implemented on ANY board anywhere with the right tools. I am going to try this on one of my older non-sli boards I am not using. I will post an update once I have it working. If it works I can post a howto to enable anyone to do this on their own setup

 

[opieum]

@zerg Please write a guide. It will save everyone the time. Maybe over time this can be scripted to be simplified.

 

[e.v.o]

I've repeated the steps from that post for GA-EX58-UD3R-SLI bios.

There is an interesting piece of code:
If (LEqual (FUNC, 0x41494C53))
{
Return (\_SB.PCI0.SLIC)
}

I've looked in the nvlddmkm.sys using IDA Pro for this hex code.
There is something that looks like a WMI call with this code. I'm not able to trace what this function does, cause I'm not experienced IDA user. But in my opinion for IDA/Assembler Guru it'll not be a problem to find how the protection works.

But for X58 it should work already, you just have to modify the acpi table. But I dont have required hardware to test with. Also there is no "nvlddmkm.sys" file on my system.

If anyone is interested I can write a short guide how to unpack the bios and edit this acpi tables.

This is only the return function. There is some more "code" with the cert ahead of it. Look at the UD4 BIOS (F3 and F4).

We only need to know now what else the driver is checking... since the driver is not only looking for the certificate...