Fast Red Zone and other fastmem-relocated projects

[jotd]

I think I know what the problem is. If you can test again the zoned update (only slave changes)

Also what's so bad about that game? the controls? the fact that you lose for stupid reasons ? because the game is trying to stick to the real event ? the fact that Tom Cruise is not featured ?

Personally I think that the way the controls are designed suck. You can's steer in diagonal while accelerating. Plus there should be some kind of automatic gear shift option.

Sometimes you're stuck in the pit lane, but if you reverse, you lose !

[Superman]

Just tried the update and it still crashes out on the Furia when you leave the pits. I can provide a whdl core dump file if required.

[jotd]

Yes, please

[Superman]

Originally Posted by jotd

Yes, please

Check the zone

[jotd]

doesn't help. Can you try with NOCACHE?

[Superman]

Originally Posted by jotd

doesn't help. Can you try with NOCACHE?

Still the same issue with NOCACHE. I noticed however that there is a NOVBRMOVE tool type set by default and if I remove that then it works with NOCACHE but not without.

[jotd]

this NOVBRMOVE is from whdload CD, I didn't notice that. It's probably here since forever, long before my release.

So it's already working with NOCACHE for you! good.

[jotd]

Nice, found a serious issue thanks to your report. Wrong patch, but different behaviour depending on the CPU & memory layout, now fixed.

Zoned a new version. Only slave changes

[Superman]

Latest version works ok for me on both the A1200 and A600 without messing with tooltypes.

[jotd]

thank you both for testing. Time for a release then.

[Reynolds]

I did not rolled back all 16 pages of this topic, but was this game mentioned?

https://hol.abime.net/1917/screenshot

Shuttle: Space Flight simulator

[jotd]

Shuttle has been fixed with whdload to (partially) use fastmem. It's coded like crap, uses some custom memory management system, with a lot of self-modifying code, and if you use 32-bit mem it can work and run pretty well, or lock up...

[Reynolds]

Again, Vector Grafix... ehh... Can't believe how they just ruined such great opportunities as they did...

Anyway, I haven't seen any updates of the fastmem relocation things we discussed earlier, shall I roll back in this topic, 'cause missed any great news? (I'm somewhere at the EPIC fix and middle of Fighter Bomber's project...) Thanks

[jotd]

The spreadsheet is up to date: https://docs.google.com/spreadsheets...N-A/edit#gid=0

[alexh]

Wowsers you jumped on the Space Shuttle Simulator quickly.

[Reynolds]

Originally Posted by jotd

The spreadsheet is up to date: https://docs.google.com/spreadsheets...N-A/edit#gid=0

Ah, thanks!

[alexh]

I just noticed that you did a FastMem WHDload patch for Combat Air Patrol - Psygnosis 20 years ago!

[jotd]

Correct But not the same league, as Combat Air Patrol code is naturally using fast memory when available, no need to fully reverse source relocation offsets & tables like in more recent projects.

The thing I remember is the self-modifying code fix required to avoid glitches on high-end CPUs (self-modifying code makes faster code on 68000 but should be avoided on processors with cache!)

[Reynolds]

Originally Posted by PascalDe73

Hi,


For me latest slave is really better :


1) No need more tooltypes like NoVBRMove or NoCache (A500 ACA500PLUS 68000 & A1200 68030).



2) Really a better speed on my A500ACA500Plus.


Best regards,


Pascal

I wonder how faster could it be after this fix... anyone can make a vid?
Dunno when will I have time to try it out, I'm behind my plans to check every recent stuff came out... :I

[Superman]

Originally Posted by Reynolds

I wonder how faster could it be after this fix... anyone can make a vid?
Dunno when will I have time to try it out, I'm behind my plans to check every recent stuff came out... :I

This is how it runs on the A600 with Furia accelerator

[ Show youtube player ]