MadMax2023 wrote:
What is your configuration? scummvm won't work at a playable speed on a standard falcon;
I'm on CT63 with Super Videl.
I downloaded Scummvm 2.0 and yes, now Simon 2 performs absolutely playable. So there must have been some optimizations in 2.0. Nice!
Oh nice. That makes it really worthwhile to take a look at again. I'm sure I could make a faster build than the default one, that must be not only built on a super ancient setup but also without that 8-bit colour front end.
MadMax2023 wrote:
What is your configuration? scummvm won't work at a playable speed on a standard falcon;
I'm on CT63 with Super Videl.
I downloaded Scummvm 2.0 and yes, now Simon 2 performs absolutely playable. So there must have been some optimizations in 2.0. Nice!
I tried this new version but the graphics are all scrambled. I see about 8 seperate main menus across the top, all tiny and all corrupted. But the old version is fine.. I wonder why. Using Xbios.
Falcon CT60e 060 - 256mb ram - Phantom bus and DSP accelerated // Atari TT - Thunder and Storm IDE 64mb ram - Lightning VME - USB LAN - ATI Mach64 2mb
This branch is 1636 commits behind scummvm:master.
So i guess this fork does not yet contain the last changes that make Blade Runner work. I also have no idea about the state of mikro's fork. Does it have any Atari specific changes that are not part of the main repository (https://github.com/scummvm/scummvm)?
No no, my fork shouldn't be used as the base for anything. It's just an experimental fork I did a few months ago to verify that ST MIDI is compilable and working.
The links wietze is referring to pointed to my custom builds long before this code appeared on github. So what wietze is actually asking is whether someone can send him the *binaries* originally placed on ge.tt. :)
mikro wrote:No no, my fork shouldn't be used as the base for anything. It's just an experimental fork I did a few months ago to verify that ST MIDI is compilable and working.
The links wietze is referring to pointed to my custom builds long before this code appeared on github. So what wietze is actually asking is whether someone can send him the *binaries* originally placed on ge.tt.
Ah I wrongly assumed those builds were a result of your fork then....
Do you have these changes still under source control at all?
BTW Miro, how did you manage to compile it? Did you have to make any changes? Or do you know at least the arguments used for configuring it? I cloned the original repo, and have some trouble with it. I had to hack at the configure script to prevent it from using the hosts sdl-config script when cross-compiling, amongst others. Same for freetype. Now i'm stuck with our math library missing some functions (truncf for example).
wietze: I'm attaching my original patch from 2015, as you can see, it's quite messy. I suppose I compiled it with very similar switches as above. The idea was to force it to use 8-bit rendering, which dramatically speeds up things on Falcon.
You do not have the required permissions to view the files attached to this post.
Thanks Mikro. I tried with sdl-prefix, but that does not work in my setup, because the script is in the same directory where also other m68k binaries are installed, and the configure script at leats will then try to execute them. Maybe i should open a PR that they better use pkg-config (as is already done for other packages). The missing truncf is maybe because i'm now using fdlibm instead of pml, could be that it is still missing there (but pml has much more missing functions). The functions taht you disabled seem be random and srandom, i had no problem with them so far, but maybe they have been added to mintlib after you last tried, did not check that yet.
The 8bit versions have Miros patches applied. But note that Blade Runner is not available in this mode. Also note that the executables are much larger than before (~27MB for the 8bit versions, and ~44MB for the 16bit versions).
It was compiled using an updated version of fdlibm. You will find it (and some other needed libraries) among the mint packages
You do not have the required permissions to view the files attached to this post.
The 8bit versions have Miros patches applied. But note that Blade Runner is not available in this mode. Also note that the executables are much larger than before (~27MB for the 8bit versions, and ~44MB for the 16bit versions).
It was compiled using an updated version of fdlibm. You will find it (and some other needed libraries) among the mint packages
The 8 bit version crashes on me. The 16 bit version crashes on me when pushing options from the launcher; but seems to load games when giving parameters from the commandline.
stormy wrote:He's probably referring to the libs, but I guess they can just be placed in the scummvm directory with the binary right?
The libs are only needed when you want to compile it yourself.
wietze wrote:The 8 bit version crashes on me. The 16 bit version crashes on me when pushing options from the launcher
Hm that's bad. I only did some quick checks with the free games that are downloadable from their site, and those started at least. Other than that, i don't have any experience with scummvm. Also don't forget that this isn't a release, it is just like a snapshot compile. Maybe you can try some other daily snapshot for a different OS first, most of them seem to be available, just not for Atari.
mikro wrote:Oh nice. That makes it really worthwhile to take a look at again. I'm sure I could make a faster build than the default one, that must be not only built on a super ancient setup but also without that 8-bit colour front end.