Which better for Scrolling Beat em up?

[Amiga Forever]

Amos Pro or Blitz 2.1?

I have got both Programming language on Amiga and I know both of them even thought I am quite good on Amos pro. IF going be Blitz 2.1 then I have to relearn them to get up to the Speed....

I guess making Scrolling beat em up shouldnt be too hard to make...should it?

[BippyM]

Blitz is more powerful and allows inline asm for any speed critical parts

[frikilokooo]

Take a look to this game made in Amos Pro:

[ Show youtube player ]

[BippyM]

Looks nice but a little sllooooow

[Retro1234]

I cant realy buy the argument that Blitz is more powerful as I think theres ways of using ASM in AMOS.

[ Show youtube player ]

Both videos use software scrolling maybe with some clever programing you could use Hardware scrolling.
I cant say which language is better I dont think anyone can - plus im no expert - best to find the one your happy with.

[Lonewolf10]

Originally Posted by Boo Boo

I cant realy buy the argument that Blitz is more powerful as I think theres ways of using ASM in AMOS.

Yes, there is.

You store your assembly code in an AMOS bank, and then use one of the AMOS commands to execute it (jump to start of said bank). The end of the ASM code must contain an RTS instruction to hand control back to AMOS.

Sorry, I can't remember the precise AMOS command used as I haven't programmed in AMOS for a few months.


Regards,
Lonewolf10

[BippyM]

Originally Posted by Boo Boo

I cant realy buy the argument that Blitz is more powerful as I think theres ways of using ASM in AMOS.

[ Show youtube player ]

Both videos use software scrolling maybe with some clever programing you could use Hardware scrolling.
I cant say which language is better I dont think anyone can - plus im no expert - best to find the one your happy with.

That is like saying Amiga Basic is also just as good.

Blitz2 with it's various commands etc is better than Amos and Amos Pro etc.. Don't get me wrong, Amos is perfectly capable at what it does, only Blitz is better and faster!

[Retro1234]

I couldnt realy say and wont go any further but better and faster is much of an example

My Dads bigger than yours

[Graham Humphrey]

It's a statement of fact, sorry

Blitz is faster and more powerful than AMOS. Again that's not to say AMOS is bad, it certainly has its merits, but for this type of game Blitz is definitely more suitable (though not as suitable as delving into assembler of course).

[Retro1234]

Its a statement that means nothing its faster? Im faster than you -Faster at what?

[frikilokooo]

It's not correct mix Amos and Amos Pro,Amos Pro is much more powerful.

There are great games,on a technical level,to show that Amos is powerful.However Blitz is better undoubtedly.

[Retro1234]

Ok Ok ill check it out sometime
Cheers

[Samurai_Crow]

Originally Posted by Amiga Forever

Amos Pro or Blitz 2.1?

I have got both Programming language on Amiga and I know both of them even thought I am quite good on Amos pro. IF going be Blitz 2.1 then I have to relearn them to get up to the Speed....

I guess making Scrolling beat em up shouldnt be too hard to make...should it?

The biggest limitations on a scrolling beat-em-up game on Amiga are the AI, and the speed of drawing transparent blits.

It's true that the AmosPro compiler generates crappy code so that would be a drag on the AI front. Blitz is OK on the AI front.

As for the speed of drawing transparent blits, you're going to be running into some barriers pretty quickly with both AmosPro and Blitz since they rely on the Blitter to do all of their transparent blitting. If you do transparent blitting with the CPU, you can cache the mask plane in a data register and speed up the graphics plotting by a factor of 5/12 for a 6 bitplane screen.

If the quality of the graphics is not critical, you can use sprite hardware though, to get past the blitter limitations. Unfortunately, the sprites are kind of useless on anything less than AGA due to the fact that they are very narrow. Also, using fewer bitplanes will speed up the blitter operations if you need to.

[Lord Riton]

Originally Posted by Samurai_Crow

As for the speed of drawing transparent blits, you're going to be running into some barriers pretty quickly with both AmosPro and Blitz since they rely on the Blitter to do all of their transparent blitting. If you do transparent blitting with the CPU, you can cache the mask plane in a data register and speed up the graphics plotting by a factor of 5/12 for a 6 bitplane screen.

If the quality of the graphics is not critical, you can use sprite hardware though, to get past the blitter limitations. Unfortunately, the sprites are kind of useless on anything less than AGA due to the fact that they are very narrow. Also, using fewer bitplanes will speed up the blitter operations if you need to.

You just need to add a 9th bitplane that is the mask of the 8 actual graphics bitplane, then you can use the blitter with that mask instead of him generating the mask each time he draws the graphics.

Sure it takes 1/8th more memory for your graphics, ah well

[Samurai_Crow]

Originally Posted by Lord Riton

You just need to add a 9th bitplane that is the mask of the 8 actual graphics bitplane, then you can use the blitter with that mask instead of him generating the mask each time he draws the graphics.

Sure it takes 1/8th more memory for your graphics, ah well

I was talking about using a tall mask with interleaved bitplane images. That is a separate mask for each bitplane. Using a single mask is even slower yet! Also, I was referring to an EHB screenmode earlier so it would cost 1/6th more chip memory to use one plane as opposed to 1/8th.

[Lord Riton]

Originally Posted by Samurai_Crow

I was talking about using a tall mask with interleaved bitplane images. That is a separate mask for each bitplane.

Could you please explain this more ? i don't understand it

Normaly a mask is a product of several bitplanes, to see wich areas they will cover. I really don't see how a mask for each bitplane can help , and by the way, a mask of just one bitplane is simply that bitplane. You got me confused

[Samurai_Crow]

Originally Posted by Lord Riton

Could you please explain this more ? i don't understand it

Normaly a mask is a product of several bitplanes, to see wich areas they will cover. I really don't see how a mask for each bitplane can help , and by the way, a mask of just one bitplane is simply that bitplane. You got me confused

When dealing with non-interelaved bitplanes they are all in separate memory locations including the mask. This is the slowest possible configuration but the most efficient memory storage. The reason it is slow is that the blitter rereads the mask plane every time it blits another bitplane.

When dealing with interleaved bitplanes under Kickstart 3.x, the bitplanes are stored in a single tall blit. The mask is stored in a single tall blit as well. This is slightly faster because the CPU only has to queue the blitter once for the entire blit. This takes nearly twice as much Chip memory and really isn't much faster.

When dealing with CPU-blitting, this ignores the presence of the blitter but is much faster because you can read the mask once and apply it to each bitplane before you read in the next longword to process. This is 7/16 the time of the interleaved blit on an 8 bitplane screen and it lets you store the mask as a non-interleaved blit. This gives you the most speed and the least memory consumption but doesn't use the blitter at all.

I hope this makes sense now.

[Lord Riton]

Ok, i understand now what you meant with: "That is a separate mask for each bitplane."

In fact that is a mask that takes as much memory as all the bitplanes together, and is just a normal mask but put 6 times in a row (if using 6 bitplanes objects) so to help the blitter to do it all in one blit. Now i understood it thanks

But it is for sure much faster than if you have to wait with your cpu for the blitter to get the first 5 bitplanes drawn. Of course you could probably use blitter interupts to remove that delay.


To do it with the cpu only is probably only faster on fast processors, especially if you have to shift (rotate) the bytes (or words) with your cpu before writing them.


What about a mix of the 2 ? you start the blitter with bitplane 1 then the cpu does bitplane 2,3,4 then you tell the blitter to do bitplane 5 and do the 6th with the cpu ? Would this slow down the cpu, and so the whole process, because the DMA channels are used by the blitter ?

[Samurai_Crow]

The best mix between the two is to let the blitter do all of the opaque blits because it can do those as efficiently as the CPU. Use the CPU to do transparent blits if you have an '030 or faster.

The GameSmith library tried to do something similar by using only idle CPU time to assist the blitter. Unfortunately GameSmith isn't very complete. I think the Games' Master System might be a better system to look at if you're making freeware.