Acekard 2i/R4i Ultra HW80

[RocketRobz]

Kernels: https://flashcard-archive.ds-homebrew.com/r4ultra.com/
Original kernel: https://flashcard-archive.ds-homebrew.com/r4ultra.com/old/R4iUltra-En_1.56.zip

Info: Flash dump (recently found to be 2MB, same as HW44) can be read, but not written, due to unknown commands.

Images (Click to Expand)

GM9 Untrimmed Dump: R4DSULTRA_R4DSR4_0.zip

Debug Log: ntrboot.log

EDIT 10/14/2025: Unfortunately, I longer have the R4(i) Ultra flashcard, so I'm unable to do any possible tests that'd be needed to ensure things are working properly.

[reminon]

Here is the full 2MB dump from GM9
R4DSULTRA_R4DSR4_00_00200000.zip

[reminon]

Cart can be written to using hw44 commands.
Cart appears to be a clone of an AK2.1.
Flashing a 2.1 dump onto the cart, leads to a black bottom screen and a black top screen with white loading text.

Not sure if it is ntrboot_flasher or something with the cart, but you have to flash a backup multiple times to completely restore a backup, otherwise you end up with backups with various differences.

[ApacheThunder]

The Acekard's flashing code has been super borked for years. I've also had to flash backups multiple times to eventually get the intended data on to the card and I'm using a hw81 version that I'm pretty sure isn't a fake. The official updaters have no issue flashing them. It seems like the Acekard flash code needs to be completely redone in my opinion. It's super unreliable.

[Ammako]

The Acekard's flashing code has been super borked for years. I've also had to flash backups multiple times to eventually get the intended data on to the card and I'm using a hw81 version that I'm pretty sure isn't a fake. The official updaters have no issue flashing them. It seems like the Acekard flash code needs to be completely redone in my opinion. It's super unreliable.

Well, that would explain why my card bricked when I restored to remove DSi ntrboot, and needed another reflash to be fixed.

It's very weird, though. Multiple re-flashes in a row sometimes just undoes my changes entirely and reverts the card back to what it was before I tried to write my modified file.

(Also, sometimes, you make just a small change, and yet it writes like 100 chunks)

[ApacheThunder]

All these years later I finally found out why flashing backups to my Acekard 2i (which is a hw 81 varient) was so unreliable:

#144

Turns out the flash size is just wrong. It's way too large. hw81 actually has a 2MB chip. Someone must have misread the data sheet and assumed the datasheet meant 16 Megabytes when it was denoted in megabits (16Mbits == 2Mbytes). So like hw 44, hw81 should be using a 2MB flash dump too!

Writes past the 2MB barrier were resulting in some undefined behavior and was why I was having so many issues. Both the DSi and 3DS versions of ntrbootflasher use this submodule so explains why I had the problem on both consoles.

For now you can just trim your backup to 2MB and use the DSi version of ntrbootflasher. I think 3DS versions try to preload the file into ram and may not let you flash it if it's smaller then it expects. The DSi version only loads each part of the file it's about to flash and after it flashes the final chunk of the 2Mb it will error out trying to read data that doesn't exist on the SD card so it fails gracefully. I was getting correct flashes every ime after I started trimming my flash dumps to 2MB! :D

This probably flew under the radar for so long because it largely doesn't impact ntrboot injection. I think ntrbootflasher on both 3DS and DSi only flash the areas that need to be flashed and ntrboot payload rom is pretty small so it never ends up attempting to write past the 2MB barrier. This only becomes a problem once you try to do a backup restore. On 3DS version you can sometimes get correct flashs because it's reads of the 3MB+ regions might happen to match what's in the backup so it doesn't attempt to reflash them. In those instances you will get a flash that works. But this is not guaranteed to always happen which explains why it sometimes works but sometimes ends up with the flash seemingly being reverted. I don't recall if the DSi version was doing this too. I did notice it being more unreliable there but the issue I ran into was always the same in the end though.

[Ammako]

Any guesses as to why HW44 couldn't be updated past 1.4.3? Since it was assumed for the longest time that the older revision had less space and couldn't fit the newer bootloader updates. but if they actually have the same amount of space, that doesn't hold up anymore.