> On Mon, 14 Dec 1998, Jon M. Taylor wrote:
>
> >> [Such a driver would presumably have to prevent simultaneous access to the
> >> card for graphics purposes or we would have some funky displays whilst the
> >> machine hosed itself.]
> >>
> >> And a subsidary question: given a), is this idea insufficiently ambitious
> >> and could/should such RAM be usable as real system memory, not just fast
> >> swap?
> >
> > It probably can't even be used as fast swap. I suggested such an
> >idea in the old GGI FAQ, and someone later told me that the VRAM+memory
> >controller logic on standard video cards is designed around the assumption
> >that writing to video memory will happen much, much more often than reading
> >from video memory. I don't remember the particulars, but the upshot is that
> >bandwith reading from VRAM is maybe 1/4 of bandwidth writing to VRAM. As
> >such, VRAM reads will be so much slower than VRAM writes that any gains you
> >might have gotten from the extra RAM evaporate. If you have "normal" VRAM
> >I/O and maybe accels and DMA too (i.e. something is actually using the video
> >card for its intended purpose) going on simultaneously the situation gets
> >even worse.
>
> Are you saying that swap to VRAM would be slower than swap to
> hard disk?
No. I am saying that you'd better not be wanting to do any
serious video work with the video card if you do use it for swap. And
normal system [S]DRAM is cheaper and sits on a faster bus with no
contention from other I/O devices. AGP and/or UDMA architectures will
render this idea obsolete anyway.
Jon
--- 'Cloning and the reprogramming of DNA is the first serious step in becoming one with God.' - Scientist G. Richard Seed
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/