On Sat, 25 Mar 2000, Wandered Inn wrote:
> So, continuing this discussion. Since I'm getting the PCI ide
> controller card, I don't have enough irqs to move just the new hard
> drive to the new card. I'll need to move something else as well. That
> would be another hard drive (33MBps), a cdrom or a cdrw. Since the low
> end usage device would be the cdrw, any idea what kind of throughput it
> commands?
On a CD-ROM or CD-RW, the throughput required is whatever its speed (e.g.
16X, 24X, 32X, 40X etc) times 150kB/s. Note the k, not M. It might
actually be a shade higher, because the X factor is based on the number of
times faster than the throughput of an audio CD. An audio CD requres
44.1ksamples/second, each sample is 16 bits X 2 channels = 4 bytes, so an
audio CD requires 176.4kBps of throughput. Some of this is lost on CD-ROM
to error checking/correcting so 150 is a fair estimate. Based on that, 2X
= 300kBps, 4X = 600kBps, 8X = 1200kBps, 12X = 1.8MBps, 16X = 2.4MBps,
24X = 3.6MBps, 32X = 4.8MBps, 40X = 6.0MBps.
> Starting to wonder if this is all worth it.
May very well be.
> Maybe I should have gotten
> another ata33 drive rather than the 66?
Not likely. 33's are hard to find. 66's, on the other hand, can talk at
33, so if you had a non-66 IDE pad, you could still use it.
> Is the possibility of pci
> saturation going to eliminate the gain of the new drive?
I don't think so. I think you may have missed my point before. I was
trying to demonstrate to you that saturation is still a long way off.
> I've got a lot of stuff on my pci bus, as noted before.
Right, but the ATA66 HDD controller is going to be the only really big
resource hog.
> Any idea what I might gain by going to an agp card and ridding this
> machine of my current video card setup (pci matrox millennium + 3d pci
> card)?
Ah, now that's a difficult question. It really depends on the
characteristics of the card itself, and how you use it. Let me detail
this a bit:
First off, even if you had an unaccelerated card, (the card you have is
accelerated) you would only have to count the necessary bandwidth to
repaint the are of the screen being updated at any given time. Repainting
the whole screen may take a couple of seconds, but just redoing some of
one window uses very little bandwidth under those circumstances. When
updating the screen in this mode of operation, you will pass 1-4 bytes per
pixel, depending on what your colour depth is (1B=256 colours, 2B=65536
colours, 3B=1.6Mcolours, 4B=True Colour).
Then there are accelerations.
The accelerated cards for 2D graphics have basic line/circle/square/fill
rendering implemented in hardware, taking some of the load off the bus
because you can tell the driver to draw a square, rather than telling it
where to put the pixels.
The accelerated cards for 3D also have algorithms for doing 3D rendering,
which doesn't do anything for the bus AFAIK, but it does save CPU time.
Thirdly, there are accelerated cards for MPEG/DVD rendering, and they take
load off of both the bus and the CPU by being able to take an MPEG stream
and render it in a given rectangular range.
Some cards do these accelerations well, others not so well, and some
combine them. For example, my card is a couple of years old, but it is an
ATI 3D Expression + DVD, so it has the 3D accelerations, it has the 2D
accelerations by default of having the 3D, and it has the MPEG
accelerations as indicated by the DVD flag in the model name.
I hope this helps!
--
">Critter@Wizvax.Net
Don't Steal - The government hates competition.
PGP key available at http://www.wizvax.net/critter/pgpkey.html.
--
To unsubscribe: mail ">majordomo@ale.org with "unsubscribe ale" in message body.