[890FX Dlx3] HOI-HOI-HOI! Mmmmery christmas from the "mad scientist"!

[Hemi]

No spare Bios chip, but if you get it stable I will give it a try. My mem is Gskill eco stuff. Cas 2 / 8 meg . not the best for overclocking but very stable.

[Ket]

I'll see what I can do. Since being able to pull apart the 1B module I'm 99% sure I've figured out what parts of it hold the memory support. Only problem is the memory tables I'd like to use are from a Asus BIOS, and they have really twatted about with the 1B module combining stuff so I'll need to find out if I can just use the combined memory support module or if I need to either A: modify the AGESA module, or B: find a way to uncombine the memory support tables. In the meantime I do have other memory tables I'll be using temporarily for people to experiment with.

[Ket]

You are going to have to hold your breath for a bit guys. I've run into problems when trying to replace components within the 1b module using ami_1b_combiner. Specifically; whenever I attempt to simply list the components within the 1b module to get the offset data (essential) the combiner spits the below error at me. Until this issue is sorted, there won't be any fun mBIOSes with stuff like much better memory and CPU support.

[arroyo]

@Ket

I just don't know what you are doing wrong. First you must extract 1B part in uncompressed form using MMTool:


And then just extract the part you want (splitter=extract, combiner=inject):


Use version of MMTool I've posted and splitter from my previous post. I've tested on Windows 7 and Windows XP. The error, that you have is from using compressed form 1B ROM part (As is in the ROM file). Or maybe you tried to decompress Agesa from whole Asrock ROM using splitter?

[Ket]

I'll start over from scratch again, but that error just shouldn't be popping up as I did extract the 1B module in uncompressed format.

[Ket]

Ok, for whatever reason it just appears for me its a complete no go. Heres the steps I've repeated several times;

- Extracted 1B module from Deluxe3 BIOS in uncompressed format with MMTOOL 3.22 mod
- Running CMD as admin typed ami_1b_combiner --list 1B
*up pops the errors*

I've even taken the source code for these utilities and compiled them myself - same result. So if you get a moment arroyo theres a few things I'd like you to do for me with ami combiner

[arroyo]

Try to use MMTOOL v3.22 BKMOD from HERE.

Can you send me your BIOS (or upload it somewhere). Maybe the problem is the bios itself. I'll try to extract it and check what's wrong.
I will help you ... no problem. Just not today. Since this is new years eve ... I'm benchmarking myself. 1.3 promile of alcohol in my blood and still standing.

Happy New Year

[Ket]

EDIT - Don't worry arroyo I got it working now. The issue (very very weirdly) was because when running CMD I was doing it start>run>cmd instead of pasting CMD directly to C:\ and running it from there.

[Ket]

FYI arroyo simply replacing the agesa code does not seem to work. I've experimented with different ways but nothing works. My board just gets stuck on POST code 07, which is undocumented. Probably going to have to open up agesa components from the msi and asrock bios, look for CPU only stuff and paste it into the asrock agesa. Even that probably won't work still :/

[Ket]

After much research and trying different things, replacing the AGESA and still having a BIOS that works is proving to be very difficult. Your turn now arroyo, maybe you will think of something I'm currently not thinking of. ONce you have something to try let me know and I'll test it.

[Ket]

A further post for you arroyo.

I've dug around in the 1B module some more and can confirm the AGESA code asrock are currently using (for the Deluxe 3 anyway) is 3.7.0. AGESA code being used in the MSI GD70 is 3.7.1. There are a lot more similarities with the Asus CIVF 1B code than there are with the MSI, so you will have more luck pulling the code from there. There are some notes however;

Asus build their 1B module very differently. You will find the AGESA code in the component DIM_CSEG. In the asrock BIOS, you will find the AGESA code in component RUN_CSEG. Lastly, the CPU.AGESA code used in the asrock Deluxe 3 is 3.13.67, in the Asus 3.13.71.

[arroyo]

Hello Ket,

There are some things that I found about AGESA:
- AGESA version goes always with compatable MEMINIT version
http://www.bios-mods.com/forum/Thread-MEMINIT-BIN
- bios-mods have Agesa resource base
http://bios-mods.com/resources/Optio...s/AMD%20Agesa/
- guys from bios-mods have proper working Agesa replacing procedure (they are using AGESA from ASUS Crosshair)
http://forums.guru3d.com/showthread.php?t=286970&page=9 post 203

I'm still waiting for spare BIOS chip for my mobo to experiment with home-made bioses.
That is a shame for Asrock, that they old motherboards like ALiveNF6G-DVI have newer Agesa than 890FX or 890GX ones!

[Ket]

Interesting stuff. I'll dig around some more. I have a bit more time before I move to 1155/2500k combo. I also have some technical papers to write, so time permitting.

[Formula350]

Thought maybe your internet blew up, Ket! :O

... Or I'm blocked on MSN, which is equally likely! lol

I nabbed the latest 1.9m, might give it a whirl later on. Anything different between it and 1.9pct you sent back around October?

[Ket]

lol no my interwebz isn't dead. The reason your having trouble getting me on MSN is because my 11 year old (or more) hotmail account just stopped working. Contacting M$ about the issue was of course absolutely useless and I'm yet to make a new hotmail account. As for the mBIOS, its just something I fiddled around with a bit. Works perfectly fine, and for me gave some small, but significant improvements such as going from a board that would not POST past 295MHz or so, to being able to boot up to 318MHz.

[NEOAethyr]

Last time I knew one of these ami bios tools would supposedly corrupt the bios on save.
I assume that's not true anymore is it ?

[Formula350]

As for the mBIOS, its just something I fiddled around with a bit. Works perfectly fine, and for me gave some small, but significant improvements such as going from a board that would not POST past 295MHz or so, to being able to boot up to 318MHz.

Think it'd help/break/fix the AM3 Deneb DDR3-1800MHz wall? I think I told you a few months back, but that Nanya ECC I got a year ago for cheap (as back up) and is only 1066MHz, you had said "Server RAM is crazy good quality stuff". SO I thought being that I was bored, we'll see what she'll do lol Pulled the OCZ, plugged in the 2x1GB, and went to work! It's 1.65V stuff and pretty darn lax timings (but it's ECC and meant for stability so...), but I set it at 1.73V and "Auto" timings, amazingly managed to get that 1066 up to 1780! I tried 1800, but it wouldn't post. It was posting fine alllllll the way up to when I jumped to 1800 (from 1750 or so) and then NOTHIN Remembered 1800 was an issue so I just tried 1780 quick to see what'd happen and it POSTed, then also passed DOS MemTest w/o an error. So I didn't bother with attempting anything more than 1780 or adjusting timings (they were like 10-11-11 something), just chalked it up as "I hit that wall". Anywho think that might help? Oh BTW, it shouldn't have any problem with that Zosma 840T should it? I'll be finding out at the end of the month if I can get my hands on one

Oh and you've got PM!

[felix_w]

Excellent thread guys...Ket & arroyo, thank you for all the interesting info...

I have something in mind to discuss here (Gigabyte AMD mb), but for the moment , spare time for this is very limited to gather all the detailed info....

[Ket]

Think it'd help/break/fix the AM3 Deneb DDR3-1800MHz wall? I think I told you a few months back, but that Nanya ECC I got a year ago for cheap (as back up) and is only 1066MHz, you had said "Server RAM is crazy good quality stuff". SO I thought being that I was bored, we'll see what she'll do lol Pulled the OCZ, plugged in the 2x1GB, and went to work! It's 1.65V stuff and pretty darn lax timings (but it's ECC and meant for stability so...), but I set it at 1.73V and "Auto" timings, amazingly managed to get that 1066 up to 1780! I tried 1800, but it wouldn't post. It was posting fine alllllll the way up to when I jumped to 1800 (from 1750 or so) and then NOTHIN Remembered 1800 was an issue so I just tried 1780 quick to see what'd happen and it POSTed, then also passed DOS MemTest w/o an error. So I didn't bother with attempting anything more than 1780 or adjusting timings (they were like 10-11-11 something), just chalked it up as "I hit that wall". Anywho think that might help? Oh BTW, it shouldn't have any problem with that Zosma 840T should it? I'll be finding out at the end of the month if I can get my hands on one

Oh and you've got PM!

It probably won't help with that. The limitation is going to be the IMC. Did you try setting the command rate to 2N? You might be able to sqeak 1800MHz at something like 9-9-9 or 10-10-10 then. Overall the vanilla 1.9 BIOS Asrock released is very good, they did most stuff I recommended, they just didn't refine the little things, hence the 1.9 mBIOS.

[Ket]

Last time I knew one of these ami bios tools would supposedly corrupt the bios on save.
I assume that's not true anymore is it ?

I haven't experienced any problems with MMTOOL 3.22 or AMIBCP 3.37 when modding BIOSes.

Excellent thread guys...Ket & arroyo, thank you for all the interesting info...

I have something in mind to discuss here (Gigabyte AMD mb), but for the moment , spare time for this is very limited to gather all the detailed info....

When you get a chance, tell us whats up and if theres any insight we can give we will digging through deeply in machine code certainly isn't my strong suit, but I can usually recognise what I'm looking for.

[felix_w]

If you have time, check this out....Raid Rom has already been updated after my communication with GB Tech....when i have time, i'll check again for modifying the values of CPU-NB Voltage

[Formula350]

Excellent thread guys...Ket & arroyo, thank you for all the interesting info...

I have something in mind to discuss here (Gigabyte AMD mb), but for the moment , spare time for this is very limited to gather all the detailed info....

I've tried with MODBIN and the Award BIOS for my 890GP-UDH3, but 2 things I found seemed to happening:
-MODBIN didn't seem to know how to handle the recompression and would corrupt the menu when re-loaded in MODBIN. (would load w/o error though, and menu layout was OK, just the text went all garble)
-There is a form of CRC/HEX check or something after flashing and upon reboot it says it's failed, then flashes the main chip from the backup BIOS I think it is related to the first thing because when it gets saved it shows modification of some files in the BIOS (HT.DLL), so seems as if maybe they are getting corrupted or something.

I contacted Gigabyte about it, but they obviously were of NO help -_- There was a lot of cool I wanted to unhide (unlock), but since I don't know how to read HEX, I couldn't compare anything to try and fix it.

I also attempted to mod the BIOS, save it (it did save fine a few times), then rip the menu module out and insert it into the original with CBRom. That didn't work either. I couldn't get the module order back to how it was supposed to be, second in line, and instead was last

[felix_w]

Formula350, u can also check the link i attached to my previous link....my objective was quite similar to yours as far as i can see now...maybe we can exchange our information...

[prznar1]

for the wallblock try to get memory tables from m4a89gtd pro and c4f. those boards can do 2k ram with deneb.

[Ket]

^^ Only if the IMC is good enough. Theres probably also some PCB differences to help get those speeds as well such as better layout for the traces and additional phases for the DIMM slots.