CJ wrote on 03.10.08 at 23:19:39:

Hi Klaz Awesome work again :)

Thanks CJ!  :)


CJ wrote on 03.10.08 at 23:19:39:

On the STe detection issue, ignore pperror.

After looking at my code again and while googling a bit, I must admit that there actually may be a bug in my STE detection code. :-[ My code presumes that _MCH value on an STE is exactly $00010000. But there's a source (toshyp) that mentions another value for _MCH on an STE with IDE hardware ($00010008).
OTOH, other sources claim that $00010008 identifies an ST Book (like Cyclone's/X-Troll article here).
So I guess I will have to make further investigations... ;-)


CJ wrote on 03.10.08 at 23:19:39:

I am currently re-coding ULS from the ground up - please feel free to drop by IRC and discuss anything you feel should be "changed" or added to it, or drop me a PM, or post here - You are without a doubt the most extensive field tester of it we have :) What we currently have: 100% pc-rel ULS "core" Dynamic buffer allocation based on pc-rel address of init routine *much* faster buffer copy routine for swapping Caches/16mhz enabled during ULS if available (Possibly BLiTTER if machine has one *undecided*) Much simpler integration

Sounds great!
Switching on the 'turbo' when shuffling RAM around is a nice idea.
I don't know if you still have the source (Metrocross?) I send to you half a year or so ago. Some of the points you mention have already been implemented there (PC relative ULS buffers, faster copy rout).
What do you think of making the clean exit "quit option" a part of ULS?
Since ULS provides the foundations for it, it might fit in there nicely? :-)

PS: my apologies to thorn for hijacking this thread.  :-[