exxos's Atari, Amiga & retro forum

OK, here are the results... One issue, when running the "Performance Tests" the results were often very screwy, as if the timer wasn't running correctly in the background, e.g. getting a CPU score of 0.4xx. Re-running the test could work or another metric would get a similarly wrong result.

Anyway, here are the screenshots:

agranlund wrote: 01 Jul 2022 20:07

frank.lukas wrote: 01 Jul 2022 19:03 ST-High: mouse works, but moves too slowly
Graphiccard: mouse works but moves very very very slowly

I set the mouse from 8 to 0 in the inf but it do not help

Hmmn, setting it to 0 should make the mouse completely unmovable.
That fact that it doesn't change anything could be some hint at least, thanks!
If you have some spare time at some point it would be super helpful to know which of the older builds was the last one that worked ok for you.

The latest version, uploaded just now, has an ACIA write fix, but based on what you are saying I doubt that'll do anything for the issue you are having.
(It does fix a mouse issue which happens *sometimes* when toggling zoom mode in ST or TT low resolution though)

220318b = mouse speed ok
220324b = mouse speed ok
220408 = mouse speed not ok
220701 = mouse speed not ok

I use with my PAK030 MSTE and the Nova VME NVDI5 ET4000 as driver at the moment.

stephen_usher wrote: 01 Jul 2022 21:45 OK, here are the results... One issue, when running the "Performance Tests" the results were often very screwy, as if the timer wasn't running correctly in the background, e.g. getting a CPU score of 0.4xx. Re-running the test could work or another metric would get a similarly wrong result.

Ah yes, you're correct. Thanks for point it out, I was able to see the same thing here and know what's going on :)

frank.lukas wrote: 02 Jul 2022 16:21 220318b = mouse speed ok
220324b = mouse speed ok
220408 = mouse speed not ok
220701 = mouse speed not ok
I use with my PAK030 MSTE and the Nova VME NVDI5 ET4000 as driver at the moment.

Thank you so much, this is great!
So it turns out the "new" mouse code never worked on your MSTE.

I'll add some logging in the next build to try and pinpoint where it goes wrong on your machine. Potentially also a special test build.
Since mouse_speed doesn't make any difference for you it could be as early as in the settings, or later when values are calculated, or worst case in communication with ACIA.

Hey @frank.lukas , would you mind trying this build and send me the log file?
It should help me narrow down where the mouse speed goes wrong (hopefully.. :D )
http://www.happydaze.se/wp-content/uplo ... 220704.zip

@stephen_usher it also "fixes" the tanking scores in speedometer.

Wall-of-text:
The emulator does some background housekeeping every once in a while.
One of these is checking if the emulated xpram has been modified, and if so save it to disk.
This check happens once every minute.

The very first xpram check, which always triggers a save, happens much sooner now.
Before it was pretty much guaranteed to happen during the benchmark if the first thing you did was run Speedometer.

@agranlund log file ...


Mouse do not work right with the 220704

frank.lukas wrote: 05 Jul 2022 12:37 @agranlund log file ...
Mouse do not work right with the 220704

@frank.lukas , thank you!

The numbers are correct, I am starting to suspect if the IKBD absolute mouse scale feature is broken in certain revisions of its ROM?
Does anyone know?
I am guessing not many programs set it to absolute mouse mode to begin with, and fewer still are probably using the scaling feature..


It works on ST, TT, Falcon and Emulators. It almost works on this Mega STE - it sounds like setting and using absolute mode works but the scale seems to be ignored.
There's really not that much that can go wrong in the code, there's hardly any of it, it's the equivalent of a two-liner:

cmd $09 - SET ABSOLUTE MOUSE POSITIONING
cmd $0C - SET MOUSE SCALE

After that, everything about the mouse position is handled internally by the IKBD and all we get is the position when we ask for it.
If it ignores the scale command then it's going to be very slow moving indeed. 8 times slower than normal to be exact.

I think I'll just do the scaling on the CPU instead of letting the IKBD do it.


(The reason you didn't have mouse issues before is that those old builds used the default relative mouse mode instead of changing it to absolute)

Perhaps some people are using an Eiffel, a CosmosEx, anything other than a genuine keyboard/IKBD...

I use always a Eiffel Interface in any setup ...

@agranlund ... with an original Atari Mega ST keyboard and mouse, the mouse movement is completely normal. So the Eiffel Interface is the problem with the new versions of Basilisk from 220324b to.