> > PATA isn't terminated, and is "designed" (cough) to work with cables
> >  between 10 and 18 inches.  If the cable is shorter than 10 inches or
> >  longer than 18 inches then you are more likely to lose data.
> 
> That is interesting but:
> 
> 1.    Notebooks have much shorter cables, as short as 2 inches.

Do they do something special to make this work reliably?

> 2.    2 inch long 40 Pin IDE cables are probably also available.

Just because something is offered for sale doesn't mean it works.

> Iirc, If A transmission line is short enough, it doesn't need to be
> terminated.  Short enough is determined by data rate and rise time.

IIRC, if it is short enough it isn't a transmission line.

Is 2 inches short enough?

> > PATA doesn't have error detection & correction on everything so your
> >  data is at risk.  SATA does have error detection & correction on 
> > everything.
> 
> Nice but you need a SATA transceiver for each Flash Card.

So they don't make SATA flash cards yet?

> > SATA allows cables up to 1 meter long.  (I don't know what the 
> > minimum is.)  I think eSATA allows 2 meters?
> 
> Is SATA terminated.

I don't know, but I assume it is.  There's no reason not to.
Perhaps the PATA "designers" (cough) were trying to spare users
the horrible task of adding/removing terminators depending on
where the drive was on the cable.

> > SATA allows port multipliers, which are like a power cube tap, or a 
> > ethernet/firewire/usb hub.  A port multiplier can fan out one SATA 
> > controller port to 15 drives.
> 
> That might be a good point.  However, Flash Cards are still parallel so
> you are going to need a serial to parallel converter for each card.

I'm talking about PATA vs SATA in general, not just flash cards.

> Have you looked at SiliconImage's chips?
> 
> http://www.siliconimage.com/products/product.aspx?id=27

I'm not familiar with the SiI3124.  I have a 3512.  It works fine, but the
max throughput I can get out of it is about 37 MB/s with one drive or 40 MB/s
total with both drives.   The same model drives get up to 65-70 MB/s per drive
with the nforce4.  I've seen complaints about slowness with some other
SiliconImage SATA chips, I don't recall the exact part numbers.  If the SiI3124
fixes the throughput problem and has a documented protocol for NCQ it would
probably be a good choice.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to