MM and JSW patched versions...

[Mysterion]

Looking through my old stuff, it seems I use that area in JSW for the 48k POP-PUSH mode.  In MM, the same routine is located at 37708.

 

One thing that puzzled me when doing the JSW patch was that there's a decent amount more useable space going spare, despite it being a much bigger game.  JSW benefits just a touch more in 48k mode than MM as a result.

 

 

M

[Spider]

Looking through my old stuff, it seems I use that area in JSW for the 48k POP-PUSH mode.  In MM, the same routine is located at 37708.

 

One thing that puzzled me when doing the JSW patch was that there's a decent amount more useable space going spare, despite it being a much bigger game.  JSW benefits just a touch more in 48k mode than MM as a result.

 

 

M

 

Sorry for the delay I had a bit of trouble finding the relevant information it seems. :) Credit to Mark Woodmass for said versions. To quote from Dr Broad's archived page a little bit:

 

Mark Woodmass posted "marginally faster" versions of Manic Miner and Jet Set Willy to alt.binaries.comp.sinclair on 27th January 2002, which speed up the frame-rate by replacing the slow LDIR instructions with a stack method to copy the screen-data to the Spectrum's video-RAM. You can download them here. Thanks to Philip Bee for finding them and sending me a copy.

 

The content's readme in the zip:

 

From: "Woody"

Subject: Manic Miner and Jetset Willy taps (0/1)

Date: 27 January 2002 21:49

 

After an idea put forward in CSS about replacing the slow LDIR

instructions in the MM/JSW game engines to speed up the framerate I've

modified both of the games here to use a stack method to copy the

screen data to the Spectrum's video memory.

 

It does make the games marginally faster but not as much as I'd hoped

for. Ah well.

 

Zip of said files:

 

fast.zip

 

:) Interestingly the code is in the Basic loader and a second interesting point (one I appreciate too as its not that common) is the 'base' for the MM version is the Software Projects version.

Edited April 17, 2018 by Spider

[Mysterion]

Thanks, dude.

 

It's certainly a very worthwhile improvement over the original game.  Interesting that he stuck the code in with the graphics data for the front screen.  Presumably, he didn't have the benefit of the Skoolkit disassembly showing him where all the unused areas of memory are. B)

 

Do you know if this one also patches the two attribute buffer copies as well as the pixels?  Turbo MM and JSW both do this for an additional gain.

 

Doing some quick testing by timing the number of seconds it takes poor Willy to asphyxiate when the air runs out, the original game, with the music off, times out in 2m 21s.  The version above ends after 1m 43s.  Turbo MM in 48k mode is good for 1m 31s and 1m 22s in 128k mode.

 

There is a second or two still up for grabs for the 128k mode and I will eventually get round to doing it.  It won't be too noticeable when playing it, but it'll make me feel good knowing that it's in there... :thumbsup:

 

 

M

[Spider]

You're welcome. :)

 

I don't *think* it patches the attribute buffers but I'll be honest I'm not 100% sure. My coding standards are not quite where they should (probably) be unfortunately. :blush:

[Spider]

@Mysterion , despite my efforts I was not really able to find any more relevant information about these 'fast' versions. :( :) Not as of the moment at least.

[Mysterion]

Just had a quick peek around inside it here, dude.  And, it does indeed speed up the attribute copying, using the same routine at 40963 as is used for the pixels.

 

It's very nicely done.  Certainly far more elegant than my version.

 

 

M