Phenom II on Gigabyte GA-M57SLI-S4

[TheWiz]

OK, thanks for going through my "experiment" As soon as I crack the ncpucode.bin case (which i think will be soon , then ill post back.

TheWiz

[haarp]

No problem. Good luck!

The problems with the USB controller seem unrelated. At least I think so. After using your BIOS, I can't overclock the graphics card anymore. This usually happened at boot for me. Attempts to do so result in an insta-freeze and the USB-controller freaking out like described before, forcing me to power-cycle. This happens on both BIOSes, so I'm not sure if it's related to me trying your BIOS. I believe that somewhere, some hardware address got scrambled, resulting in hte overclocking tool trying to write to the USB controller or something... strange.
Either way, don't worry. I don't blame you

Another interesting quirk I noticed is that your BIOS changes my onboard network's MAC address while it's active
It goes from 00:1d:7d... to 00:20:ed... both ranges belong to Gigabyte though.

[gabefauver@yahoo.com]

Hey haarp,
I like the concept of switching from one chip to another. Could you post some pics and details on your dual bios/switch setup as I am interested in doing this on my board.

[haarp]

Hey haarp,
I like the concept of switching from one chip to another. Could you post some pics and details on your dual bios/switch setup as I am interested in doing this on my board.

The setup is specific to the M57SLI-S4. You can find a Howto with a nice, big pic here: http://stuge.se/m57sli/

To quote the author of this mod:

Gigabyte owns a patent for Dual-BIOS, where they have two flash
chips, a switch controlled by a timer and a special handshake. Board
powers on first time with one chip active, if there is no handshake
within the timeout then the board is reset again, the other chip is
activated and reset released.

The pads are there for this circuit on many Gigabyte boards, but it
isn't populated very often.

We found out the traces for the pads involved in the mod and I
designed the manual switching circuit that replaces the timer.

I don't remember if I wrote in the instruction, but the switch
switches 3.3V, so it's a good idea to put some heatshrink over the
contacts to avoid a short circuit by touching the case or so.

All flash chips have a CS# signal which is normally connected to the
flash chip bus master. On m57sli this connects to E on both NPN
transistors. Collector goes to flash chip CS# pin. CS# pins are
pulled high by 100k. Both transistor B are pulled low by 100k. 3.3V
is switched into one of two 4k7 and then to a transistor B.

I would have prefered to switch GND but the pads did not allow that
without adding extra components outside the pads. I decided it was
nicer to only use existing pads.

[TheWiz]

OK, harrp, thanks for your patience. This is a complete rewrite with all new microcode, so here's the link and best of luck. (finally able to get to these as was able to extract the AMD microcode today) I tried a couple different "techniques" so I ahve listed the mods in the perferred flash order. So try the first link, it it works great, if not move on to the next. make sure when you flash you rename the file to bios.bin.

http://dl.dropbox.com/u/1525839/Bios/M57...281%29.bin
http://dl.dropbox.com/u/1525839/Bios/M57...282%29.bin
http://dl.dropbox.com/u/1525839/Bios/M57...283%29.bin

TheWiz

[haarp]

Hello and thanks for your efforts!
While all of those BIOS are booting fine, none of them seem to be able to circumvent the multiplier-lock. CPU stays at 800MHz

[TheWiz]

Does it detect the correct cpuid?

TheWiz

[haarp]

Yes, but even the official BIOS does that. At bootup, I get the correct model name and CPU-Z has no problems recognizing the CPU.

[TheWiz]

OK, is it alright if I try and set a manual multiplier?

[haarp]

Well, I've already tried changing the multi. Or do you mean hardcode a multi into the BIOS? That could work. The default multi should be 16 for this CPU.

Now that I think about it, the whole problem could have to do with P-States. After all, the CPU defaults to the lowest P-State and discards the others it seems. Additionally, It refuses to accept any changes to this P-State. (You can modify the states saved inside the CPU with some tools like K10stat) Don't quote me on that though