Path: utzoo!utgpu!water!watmath!clyde!att-cb!att-ih!pacbell!ames!oliveb!pyramid!prls!philabs!sbcs!root
From: root@sbcs.sunysb.edu (SBCS Systems Staff)
Newsgroups: comp.sys.amiga
Subject: Re: 512K expansion for A500
Summary: Using fastmem to store image data
Message-ID: <1107@sbcs.sunysb.edu>
Date: 2 Apr 88 03:39:02 GMT
References: <3054@gryphon.CTS.COM> <47622@sun.uucp> <684@applix.UUCP>
Organization: State University of New York at Stony Brook
Lines: 42

In article <684@applix.UUCP>, scott@applix.UUCP (Scott Evernden) writes:
> In article <47622@sun.uucp> cmcmanis@sun.UUCP (Chuck McManis) writes:
> 
> How about some clever code to swap unseen screen bitmaps to/from FAST
> memory??  I've been toying around with this idea for a few weeks
> now.  I haven't really spent the time to work thru the problem completely,
> but I can imagine this as a potential solution to the CHIP mem crunch.

	Yes.  I've been cleaning up my NeWS port for the Amiga over
	the past few days now, and this is exactly what I do down in
	the pixrect driver.  What happens is simply this:

		Try to allocate CHIP mem except when system starts
		to run dry.  Else, allocate pixrects into FAST mem.

		When blitting, I do a simple test to check if I can
		use the Amiga Blitter to do the RasterOP.  If not, I
		use the Sun routines to perform the OP.

	Works very well.  Anyways, I'll add a little value to this
	posting besides just a "YUP".  I think that it would be really
	nice if the graphics.library would make just this sort of
	decision, ie Blit when possible, otherwise emulate.  It seems
	(unless there are some horns that I haven't spotted here) that
	this would quite nicely remove the asymmetry of CHIP vs FAST
	memory, at least as far as graphics ROP'ing was concerned.  Of
	course, the technique doesn't work for the copper, etc.  Well,
	that's my $.02.
> 
> You'd have to get down and dirty replacing quite a few graphics/intuition
> library functions, and you might even have to write software equivalents
> of most of the graphics calls, or else block tasks attempting output.
> But it COULD be done (couldn't it??).
> 
> Am I completely off-the-wall? 

	not at all!
> 
> -scott

						Rick Spanbauer
						SUNY/Stony Brook