1.6 128MB RAM Upgrade Tutorial

[Bowlsnapper]

1 hour ago, sweetdarkdestiny said:

No XISO support for EvoX. I know that Rocky5 has made a BFM EvoX XISO but I don't think that I can patch that. Maybe if you ask Rocky5 for an updated version of his BFM bios he will do it. 

And the MakeMHz versions will come later. I will send you a PM with them to test and when you say they are good, I will drop them here. But, atm I need more coffee to come to life.  

Youˋre a beast sweet. I wish I could code like that.

[sweetdarkdestiny]

13 minutes ago, Bowlsnapper said:

Youˋre a beast sweet. I wish I could code like that.

Wait wait wait. I didn't code anything!

The 128MB EvoX patch and ofc. the XBoxHDMI/HD+ patches are from the MakeMHz Team. And the TITAN patches are from Gaasedelen. 

I have only patched the bios files in the right order in different flavors, nothing else. All credits belong to the guys above.

[Bowlsnapper]

18 minutes ago, sweetdarkdestiny said:

Wait wait wait. I didn't code anything!

The 128MB EvoX patch and ofc. the XBoxHDMI/HD+ patches are from the MakeMHz Team. And the TITAN patches are from Gaasedelen. 

I have only patched the bios files in the right order in different flavors, nothing else. All credits belong to the guys above.

Oh shit, sorry. I read the post and made the lazy assumption that you were involved in the project. I need to fuckinˋ pay attention before I just go handing out unsolicited credits like that.

[sweetdarkdestiny]

42 minutes ago, Bowlsnapper said:

....

NP. I still take the one for being a beast. Usualy girls say that to me.

[sweetdarkdestiny]

Alright. Droping a non final bios pack. (There is only the 1.0-4 & 1.6: " EvoX Titan (UDMA2/4/5) MakeMHz XBoxHDMI 128MB " patched bioses missing (So for the old HDMI thingy thing).) 

ATM 108 pre patched bioses in total right now.  And ofc I will see that I patch the ones for the old HDMI asap. And a new updated version of the OGXBox Bios Flasher is also on the way. ^^ 

Download : https://1fichier.com/?hq4acxhna8j0wxsdvztw

EvoX 128MB
EvoX 128MB 1.6
EvoX 128MB HDplus
EvoX 128MB HDplus 1.6
EvoX 128MB XBoxHDMI
EvoX 128MB XBoxHDMI 1.6
EvoX Titan 128MB
EvoX Titan 128MB 1.6
EvoX Titan 128MB HDplus
EvoX Titan 128MB HDplus 1.6

All bioses are available as  as yesDVD (YD) and 
as noDVD (ND) version. 

The TITANS are available in UMDA2 + 4 + 5 (So in UDMA11)

And you may guessed it, they are also available as 256kb, 512kb & 1024kb version. 

 

Credits and a HUGE thanks for the patches fly out to: MakeMHz and Gaasedelen.

And also a HUGE thanks to Donnie-Burger for testing.

Edited June 12, 2022 by sweetdarkdestiny
Typos. -.-

[Donnie-Burger]

11 hours ago, sweetdarkdestiny said:

Alright. Droping a non final bios pack. (There is only the 1.0-4 & 1.6: " EvoX Titan (UDMA2/4/5) MakeMHz XBoxHDMI 128MB " patched bioses missing (So for the old HDMI thingy thing).) 

ATM 108 pre patched bioses in total right now.  And ofc I will see that I patch the ones for the old HDMI asap. And a new updated version of the OGXBox Bios Flasher is also on the way. ^^ 

Download : https://1fichier.com/?hq4acxhna8j0wxsdvztw

EvoX 128MB
EvoX 128MB 1.6
EvoX 128MB HDplus
EvoX 128MB HDplus 1.6
EvoX 128MB XBoxHDMI
EvoX 128MB XBoxHDMI 1.6
EvoX Titan 128MB
EvoX Titan 128MB 1.6
EvoX Titan 128MB HDplus
EvoX Titan 128MB HDplus 1.6

All bioses are available as  as yesDVD (YD) and 
as noDVD (ND) version. 

The TITANS are available in UMDA2 + 4 + 5 (So in UDMA11)

And you may guessed it, they are also available as 256kb, 512kb & 1024kb version. 

 

Credits and a HUGE thanks for the patches fly out to: MakeMHz and Gaasedelen.

And also a HUGE thanks to Donnie-Burger for testing.

What an epic pack.  Thank you for sharing!

[big F]

39 minutes ago, Donnie-Burger said:

What an epic pack.  Thank you for sharing!

Ditto on that.

 

Just wondering what happens if you use one of these Bios's on a not yet updated to 128 board.

I am guessing they wont go bang!!

[sweetdarkdestiny]

8 hours ago, big F said:

Just wondering what happens if you use one of these Bios's on a not yet updated to 128 board.

I am guessing they wont go bang!!

Not testet but I would assume that the "EvoX 128MB" would work like any other non patched bios. 

But I would not try to use the HD+/XBoxHDMI patched ones on a XBox without the HDMI mod. 

[big F]

37 minutes ago, sweetdarkdestiny said:

Not testet but I would assume that the "EvoX 128MB" would work like any other non patched bios. 

But I would not try to use the HD+/XBoxHDMI patched ones on a XBox without the HDMI mod. 

Was thinking the same, lazy me was just wondering if I could get away with flashing all my boxes the same regardless of memory installed.

That way if/when I update them its no bother. 
 

I have a couple HDMI modded and a couple more waiting to be done, normally I try anything substantially different on a mod chip first, that way if it goes pear-shaped you can switch banks and recover. 

[sweetdarkdestiny]

Alright. Here is the final bios pack : 

 

[Prehistoricman]

I've been busy...

Can you guess what I'm investigating? The probes on the left connect to a logic analyser.

Sadly this didn't work as I hoped. It introduces some kind of interference, signal reflection, or something and makes the Xbox very unstable. For example:

The Xecuter Live Config was also rather corrupted with the spinning logo having some crazy polygons.

Xblast was blasted, usually with just some white sprinkles around the screen but occasionally something interesting. Looks like the BIOS failed to extract all its own images.

Looks like it is trying to display all the titles at once, the text is smaller than usual, and I didn't make 2 changes.

[KaosEngineer]

1 hour ago, Prehistoricman said:

I've been busy...

Can you guess what I'm investigating? The probes on the left connect to a logic analyser.

Sadly this didn't work as I hoped. It introduces some kind of interference, signal reflection, or something and makes the Xbox very unstable. For example:

The Xecuter Live Config was also rather corrupted with the spinning logo having some crazy polygons.

Xblast was blasted, usually with just some white sprinkles around the screen but occasionally something interesting. Looks like the BIOS failed to extract all its own images.

Looks like it is trying to display all the titles at once, the text is smaller than usual, and I didn't make 2 changes.

The input/output lines of the SDRAM chips are SSTL_2 compatible signals.  What type of input/outputs signalling does your logic analyzer support?

See: SSTL Interface Description, Stub Series Terminated Logic for DDR, DDR2, DDR3 Electrical layer, SSTL-2, SSTL-18, SSTL-3 (interfacebus.com)

[Prehistoricman]

58 minutes ago, KaosEngineer said:

The input/output lines of the SDRAM chips are SSTL_2 compatible signals.  What type of input/outputs signalling does your logic analyzer support?

See: SSTL Interface Description, Stub Series Terminated Logic for DDR, DDR2, DDR3 Electrical layer, SSTL-2, SSTL-18, SSTL-3 (interfacebus.com)

It's a DSLogic U2Basic upgraded to Plus capability. The input threshold can be set by the user between 0 and 5V in 0.1V increments. My problem is not related to the capability of the analyser but some physical property of this setup. If I plug in the probes, but leave them disconnected from the analyser, I still see issues.

Possibly a series resistor on each probe at the header would solve this issue.

Before building this setup, I had much longer wires connected to the same analyser and this was more stable.

[vitor]

Thanks for tutorial. One thing that i am thinking is about heat issues with the memories one over other. One secure way to prevent it? Thanks

[Donnie-Burger]

1 hour ago, vitor said:

Thanks for tutorial. One thing that i am thinking is about heat issues with the memories one over other. One secure way to prevent it? Thanks

Done 3 so far and heat is no issue.

[vitor]

2 hours ago, Donnie-Burger said:

Done 3 so far and heat is no issue.

Glad to know

Thanks

[Prehistoricman]

Teaser #2

Here I've lifted the A8 and A9 pins and swapped them. Repeated for all 4 chips.

It works as long as the BIOS is patched to flip the function of these pins. 

[Prehistoricman]

The BIOS I put in the tutorial is bad. It works for checking the RAM but it shouldn't be used aside from that. The source code was updated since the 0.56 version that works well. It's some mix of my compiler and the source code changes that have broken it. The known issues are:

1. HDD flash does not work. The BIOS can't read the HDD.
2. Net flash does not work. The page times out before the upload completes.
3. CD flash may be dodgy. It works for me but not for someone else.

Unfortunately a modchip supplier started shipping Aladdins with this BIOS flashed on, effectively preventing any users from installing another BIOS! What a mess. I'm working with them to fix this.

This is why you don't download people's dodgy .bins from the internet. And this is why I shouldn't download dodgy source code from the internet

 

On the plus side... teaser #3: Xbox with 96MB RAM??

[Prehistoricman]

I fixed the BIOS and called it version 0.60.

The 0.56 source code was good except for one thing: the net flash was super slow. This also happens to be an issue with the Xecuter FlashBIOS for some (or all?) users. According to the commit history of XBlast, this issue had been fixed already but I couldn't see any changes in the source that would be responsible for the fix.

I enabled debug output on xblast, then hackily patched all the broken and segmented debug support. After a lot of testing, I found an easy way to test for the issue. In a ping test, the BIOS would respond instantly to every other ping. Thanks to the debug output, I could see that the functions to send the packet were actually being called on time but somehow a packet was not transmitted. That brought me to the lowest depths of the ethernet driver: forcedeth. The function to write the packet to the hardware registers and send it is forcedeth_transmit. At the end of this function, pci_push is called which is helpfully commented "force out pending posted writes". That's exactly the bit that sounds like it's not working. So I just duplicated this function call and magically the issue was gone. Turns out the nForce just needs a couple clock cycles before it's ready to transmit the packet. The compiler would sometimes sneak in an instruction to delay pci_push and this is what makes Xblast 0.56 work properly.

[xgekox]

hello,

first of all great job! I wanted to ask you if you also had the source code of the new version of xblast (0.60) which if I understand correctly, works like 0.56 but in addition it has support for xbox 1.6, right?

Thanks

[Prehistoricman]

6 hours ago, xgekox said:

hello,

first of all great job! I wanted to ask you if you also had the source code of the new version of xblast (0.60) which if I understand correctly, works like 0.56 but in addition it has support for xbox 1.6, right?

Thanks

I wasn't going to publish the source since it's barely any different from the original. And I don't want responsibility

To reproduce what I have in 0.60:

1. Install a git client. I use TortoiseGit since it has a nice UI
2. Clone the git repo from https://bitbucket.org/psyko_chewbacca/lpcmod_os/src/master/
3. Roll it back to version 0.56 to fix FATX and networking bugs
4. Modify menu/textmenu/ToolsMenuInit.c as described in the tutorial
5. Duplicate line 1111 in etherboot/drivers/net/forcedeth.c. This line should say pci_push(base);
6. Change the version string in include/config.h

[zzattack]

Interesting upgrade. Are you looking into using different RAM types entirely, such as DDR-333/DDR-400?

I'm gonna see if I can complete a first 1.6 with 128MB + 1.4GHz combo. 

[Prehistoricman]

32 minutes ago, zzattack said:

Interesting upgrade. Are you looking into using different RAM types entirely, such as DDR-333/DDR-400?

I'm gonna see if I can complete a first 1.6 with 128MB + 1.4GHz combo. 

I was thinking about it, yeah. However I'm not sure it the system would support the RAM clock being faster than the NVCLK at 233MHz. I'm also not sure how to change the timings for the memory. DDR-400 would require a higher CAS latency, for example.

[big F]

Wouldn’t the higher rated chips be more cool/stable at a lower speed. Could be useful for emulated systems.

[Donnie-Burger]

So I have gifted about 4 xbox 1.6s with 128mb ram and many are saying games are black screening freezing etc.  I'm going to try and pinpoint if its a Titan/Cerbios/UDMA5/ or 1.6 - 128mb issue.  This might be a 1.6 - 128mb issue as sega chihiro games also have issues and i'm pretty sure I played some of those on my non 1.6s 128mb.  If anyone has any info please share.

 

I'm upgrading a 1.6 now so I can test.  I have a feeling this is bios related but need to be one hundo.