Overview Features Instructions Performance Forum Downloads Products Reseller Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
VISIT APOLLO IRC CHANNEL



All TopicsNewsPerformanceGamesDemosApolloVampireCoffinReleases
Running Games and Apps.

Protracker Big Problem With Glitchespage  1 2 

Billy Nest

Posts 22
10 Apr 2018 11:43


Speaking with a coder friend said to me that the glitch occurs from the Apollo core.
Its difficult to build a new patch for protracker 3.61 - 3.62 versions but its more easy to fix it on the 68080 core in the next update.
I'm using the latest stable 2.9 x10 core on my V600
Can you please fix that glitch?


Olav Sørensen

Posts 3
10 Apr 2018 12:47


If the issue is in fact in Vampire's 680x0 core, then it's flat out wrong to try and make a "fixed" PT3.62 version. Fix the Vampire core instead!
 
Some people think that I made PT2.3E for Vampire users, but I did not. The reason it works nicely on Vampire and fast Amigas is because I replaced the CPU busy loops (for DMA latch and certain input routines) with a raster wait, so that it's waiting long enough regardless of the CPU and speed.
 
I've also got told that PT3.62 is most likely doing it the same proper way, so it sounds to me like it's an issue with the Vampire core. I'll get to disassembling it when I have time, and find out if it's indeed not having CPU busy loops in the replayer.
 
Wolfgang Köchl:
The reason I focused on patching the 2.x series of ProTracker and not 3.x, is simply because it's *the* best series. 3.x is a joke, they should never have touched (and ruined) the replayer logic while making it.


Gunnar von Boehn
(Apollo Team Member)
Posts 3515
10 Apr 2018 13:16


Olav Sørensen wrote:

I'll get to disassembling it when I have time,

That would be nice - thanks!
The problem is not full understood yet.



Olav Sørensen

Posts 3
10 Apr 2018 13:27


Don't expect me to bring scientific information back though, as I'm not an Amiga developer in any way. I just want to see if it uses the CPU busy loop practice or not. StingRay suggests that it uses a level 1/level 3 interrupt combo, but he's not 100% sure if he remembers correctly.
 
Stefano Briccolani:
It's not true that any PT version will run correctly on very fast Amigas by just disabling CPU caches and setting the system to PAL.


Gunnar von Boehn
(Apollo Team Member)
Posts 3515
10 Apr 2018 13:38


Olav Sørensen wrote:

it uses a level 1/level 3 interrupt combo,

Would it not use audio interrupts of LEV 4?
I can imagine that problems could be caused by bouncing IRQs.


Billy Nest

Posts 22
11 Apr 2018 07:06


There is a useful archive with protracker sources from the EAB server
 
  ftp://ftp:any@grandis.nu/TheZone/files/_2013/ProTracker Collection like ProTrackers.zip, but with more versions and sources for titobal].rar


Billy Nest

Posts 22
11 Apr 2018 07:45


Gunnar also check this video to see how the glitches affect the play of the track

EXTERNAL LINK


Billy Nest

Posts 22
15 Apr 2018 17:52


Any progress on the problem team ? Is it possible to fix that problem soon ?


Renaud Schweingruber

Posts 226
15 Apr 2018 18:12


Team is on it. No ETA.


Billy Nest

Posts 22
16 Apr 2018 10:35


At least dont release Gold 3 core without it.
It's crucial for the function of the core. Its an annoying bug and it was reported
Back since core 2.7 released.
I trust the team because the team delivers :)


Olav Sørensen

Posts 3
16 Apr 2018 12:12


I disassembled PT3.62 and had a look. It's indeed waiting for Paula DMA latch the same way PT4.00 Beta does it: EXTERNAL LINK     
 
Seems to trigger a special event interrupt on new notes that is handling DAT/LEN for sample loops, but I'm not really the correct guy to analyze this. Anyways, at least I now know that PT3.62 doesn't use the CPU busy loop method like 1.x/2.x which caused problems on faster CPUs. This also explains why it just works on faster Amigas.
 
I'll leave this for the Apollo team to figure out, maybe reading the 4.00 source code will help. I'm not even sure where in the program Vampire has problems, so take it for what it's worth.
     


Billy Nest

Posts 22
03 May 2018 22:10


any news or any development lads ? is the problem fixed with the protrackers ? any testing ?


Renaud Schweingruber

Posts 226
04 May 2018 11:43


Team is still working on it.

Answer will probably be a global core improvement than a specific fix. This takes time :-)


Jim Neray

Posts 3
14 May 2018 21:06


I want to believe ... :)

posts 34page  1 2