Assembly Language eComic - Issue 7 now available

[NormanDunbar]

Hello to all (both?) my faithful readers. The latest edition of the somewhat irregular eComic about QL Assembly Language Programming is out now. Get it from:

https://github.com/NormanDunbar/QLAssem ... ag/Issue_7

You only need the PDF (to read) and the Code.zip file, to save you typing. Enjoy.

In this issue there is an article by Tobias on the Q68, plus exciting stuff about the UTF8 character set encoding and how it can be used on the QL - or at least, how I can use it! Two world class (ahem!) utilities are supplied to enable conversion from the QL to UTF8 and back again. There's even, wait for it, a table of contents!

Help yourselves and enjoy.

Cheers,
Norm.

[NormanDunbar]

I've fixed a small bug in the code submitted by Tobias in Issue 7. There was a use of A7 in a register list that should have stopped at A6. (That's the sort of mistakes I make!) Wolfgang spotted it and let me know.

I've uploaded a fresh copy of Issue 7 to the URL above. If you downloaded your copy prior to 19:00 BST tonight, aka GMT +01:00, aka UTC + 01:00, then help yourself to a fresh pdf download to get the error corrected version.

Cheers,
Norm.

[retro_collector]

Thank you Sir

[tofro]

I've fixed a small bug in the code submitted by Tobias in Issue 7. There was a use of A7 in a register list that should have stopped at A6. (That's the sort of mistakes I make!) Wolfgang spotted it and let me know.

I guess whatever you (I) do, there's always that last error you (I) miss

Sorry for the inconvenience,
Tobias

[NormanDunbar]

Or even, no matter how many times you (I) proof read something, errors only appear totally obvious once the thing has been distributed!

No problem at all Tobias, thanks for the contribution.


Cheers,
Norm.

[Peter]

In this issue there is an article by Tobias on the Q68, [...]

Almost missed this interesting read. Very nice, thank you Tobias and Norman!

As for the loop unrolling: If a few percent of performance don't matter, one might not unroll 256 instructions, but 64 for example, spending 75% less memory. Having 32 MB SDRAM this seems pointless, but if the code is executed in Q68 fast RAM or (not yet released) cache, size matters.

Loop unrolling is always a good idea on the Q68 where speed matters. E.g. I also used loop unrolling in the SD card driver. While the CPU can execute simple instructions in a single clock cycle if running from SRAM, the cost for branches are relatively high. I wonder if the SMSQ/E memory copying routines in the screen driver could be optimized in a similar manner.