Path: chuka.playstation.co.uk!news From: "Rad" Newsgroups: scee.yaroze.programming.2d_graphics Subject: Re: Accessing VRAM directly Date: Wed, 30 Jun 1999 12:03:35 +0100 Organization: PlayStation Net Yaroze (SCEE) Lines: 36 Message-ID: <7ld4b7$ko22@chuka.playstation.co.uk> References: <37789961.AB473632@mcneish.freeserve.co.uk> <7lafvn$11115@chuka.playstation.co.uk> NNTP-Posting-Host: modem-115.name3.dialup.pol.co.uk X-Newsreader: Microsoft Outlook Express 4.72.3110.5 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 Toby Sargeant wrote in message ... >On Tue, 29 Jun 1999 13:54:55 +0100, Rad wrote: >>Bruce McNeish wrote in message >><37789961.AB473632@mcneish.freeserve.co.uk>... >>>Just a quick question. >>> >>>Is it possible to read/write directly to/from the video buffer to >>>generate real-time images. >> >> >>Nope, at least not with the libs. > >Nope, full stop. VRAM isn't mapped into the CPU address space, which >means that the only wat to access it is via DMA or the GPU, which is >what LoadImage and StoreImage already do. For single pixel operations >This is going to be horrendously slow. > >The other thing to consider is that @ 33MHz, if you want to individually >update every pixel on the screen, you're looking at about 10 cycles per >pixel (@ 50 Hz frame rate), which is just plain not enough. As I figured out in my real-time fire demo which (looking back at it now) looks quite bad with the low resolution. I should have included a usage instructions "stand back and squint your eyes - looks much better that way!" Anyway, I want to rewrite it when I have time, I have an idea on mucho speed increase I'd like to try out. >Toby. Rad. rad@cyberdude.com http://www.netyaroze-europe.com/~radpsx