pcmhacking.net

Ok so! Doing some testing with Gampy's kernel.

The OBDX Pro GT and VX worked without a single issue for reading and writing.

The OBDX Pro VT required either:
1) Pulldown resistor on VPW line to ground to run stable at 4x
2) Running it at 1x without resistor or anything.

My personal verdict is the OBDX Pro VT should be hard coded to 1x for the P04's.

Successfully Cloned my other PCM to my Junkyard PCM.

There are (probably known to devs) issues with the PCMHammer options for P04 PCMs. Calibration writes do not work. I saw somewhere that the Cal data is intertwined with parts of the logic and it definitely shows in the logs. I did a calibration write and it finished in less than a second so I looked deeper and noticed it didn't write anything even though I changed 150 tables or so.
I then tried write parameters and then I checked the log and saw the sectors where it was actually different when I tried a clone.

So it seems for now P04s have to clone every time there's a "tune?"

Anyway, that needs to be a disclaimer for P04s as well.

Also, this was done on a bench. Still afraid to try in car. I'll probably get a FOURTH PCM and give it a whirl as bench writing is quite awful when trying to actually tune the car. It's even worse than eeprom burning (which I actually prefer over both!)

Yep, the P04 has changed the game. If the reading and writing is good, we need to spend some time on the interface. Help welcomed.
We were living in the dream world with the maturity of the P01, P59 and newer. A lot of the nice things about the platform are not so nice in the older P04 and P08

What are you finding different about the P08 then say the P01 ... I find they are pretty much the same from a flashing point of view.

-Enjoy

It's the older generation CPU with its alignment crashyness we had to code around that the P01 and newer didn't suffer from. You can see the generational increments and by the P01 *everything* is pretty nice. The segmentation being properly implemented in the P08 is better, for sure. As well as its slightly limited but still present recovery boot sector. I think we might still need some PCMHammer changes to get on to its A2 00 recovery mode. No A2 01/A2 02 modes like the later which do work with current code. At least when I was playing on the bench I could trigger A2 00 but I couldn't get in to it due to timing and collisions. But I didn't try for long as it was scope creep for the task at hand.

antus wrote:It's the older generation CPU with its alignment crashyness we had to code around that the P01 and newer didn't suffer from.

That is not true, as far as flashing and coding goes it is as solid as the P01, it does not have the same issues with alignment (or whatever it is) that the P04 suffers from, the problems you are having are self inflicted, that is why I stepped in.

antus wrote:As well as its slightly limited but still present recovery boot sector. I think we might still need some PCMHammer changes to get on to its A2 00 recovery mode. No A2 01/A2 02 modes like the later which do work with current code. At least when I was playing on the bench I could trigger A2 00 but I couldn't get in to it due to timing and collisions.

I have not even looked into recovery, not in my picture ATM, doesn't need to be, it has nothing to do with the Kernel, that is all I am working on, is the Kernel.

-Enjoy

OK, we'll need to agree to disagree. I have seen changes, probably not the ones you are talking about, break and fix p08 which have no effect on P01 and newer. Some of my results were self inflicted, but I am 100% sure not all. It cannot run all the same opcodes the same way. When you release your code I'll look to see how you have done the flash setup, and with what opcodes. As we are talking about maturity of the platform as a whole, the bootloader maturity is relevant to the point. It doesn't matter now anyway, I await your kernel branch to be submitted to go in to the app after its had enough testing, and I still have a call for help on the PCMHammer front end. (and @NSFW, if you see this, to merge some branches and get at least what's in develop right now out, too). I think the kernel is under control with your branch, gampy, but a front end developer willing to help on a form to manually choose PCM type for read and write, and help re-think the write options now we don't always have a 1:1 relationship between flash segments and purpose on all PCMs would be great. Overhauling the unlock and execute logic to make the flow clearer for read/write/recovery would be great too. And of course there is bubbas CAN test branch hanging, it'd be great to review that and get it in, even if we dont have any CAN supported PCMs. Stop the branch getting stale and leave the door open for E38 or others.

Flashed in car a few times now. Successful at 1x. Not ever going 4x.

Thanks again guys. You’re amazing.

ProfessWRX wrote:Flashed in car a few times now. Successful at 1x. Not ever going 4x.

Thanks again guys. You’re amazing.

Great to hear!!!!