On Tue, 27 Apr 2004, Rick Moen wrote:
> Hmm, I'd have to look that up. That was based on materials I have
> at home, before I left for the day.
There's a spec on serial-ata.org for port-multiplexing, presumably to
allow one to have intelligent backplanes that then just need only one
S-ATA connection to the host. (or even two, if the host/devices
support that pending multi-path spec).
> [Future doubling of theoretical SATA bus bandwidth limits]
You brought that one up.
> > Yeah, only way beyond SCSI. but hey.
> See, _right here_, this is the single most spectacular pons
> asinorum that ATA advocates get stuck on: A theoretical bus speed
> limit is significant only if there's some prospect of ever being
> bottlenecked by it.
Did I say otherwise? Also, if you have a multiplexing SATA backplane,
the extra bandwidth _will_ be useful. No single drive may be able to
use 1.5GB/s, but 15 of them will get very close with todays high-end
Indeed, why exactly is it you never see SCSI buses with more than 4
devices on them, even though the bus can address 15 devices (+1
host), least not on any sane setups. Even U320, the absolute fastest,
could take 8 drives, at best, given not-too-fast drives before
> Since ATA cannot support disconnected operation (only one device
> per chain can carry out operations at a time),
This is irrelevant to SATA as it is point-to-point, except for the
case of multiplexing, see above. There is no other device on the link
to the host with which one could contend with, where disconnected
commands would help.
> saturate ATA/100.
What has ATA/100 got to do it with?
> Minor amounts of device-connection latency aside, there is thus
> _zero benefit_ from higher ATA bus-transfer speed ceilings, almost
> certainly for the medium future (2-3 years).
Who is talking about P-ATA? I wouldnt use P-ATA for multi-disk arrays
in a machine (unless each disk is behind its own SCSI bridge), I
wasnt suggesting such use either. Stop it.
> saturate the bus under real-world conditions. And guess what? It
Niall, care to step in? please.. :)
However, we were not discussing P-ATA, we were discussing S-ATA.
S-ATA at last lets one attach multiple disks in a sane fashion.
> When U320 starts being obsolete, I'll be out there looking at what
> makes sense then for real computing. Could be FC, could be
> Infiniband, could even be some outgrowth of IEEE-1394.
Fibre Channel, yum. Insane for home use though. Good luck finding an
FC controller with linux support with anyway decent drivers.
> And I have no more time for what is clearly a fruitless discussion.
Absolutely, no small part due to your insistence on talking about
P-ATA. Very annoying.
Paul Jakma paul at clubi.iepaul at jakma.org Key ID: 64A2FF6A
warning: do not ever send email to spam at dishone.st
Life is a hospital in which every patient is possessed by the desire to
change his bed.
-- Charles Baudelaire
Maintained by the ILUG website team. The aim of Linux.ie is to
support and help commercial and private users of Linux in Ireland. You can
display ILUG news in your own webpages, read backend
information to find out how. Networking services kindly provided by HEAnet, server kindly donated by
Dell. Linux is a trademark of Linus Torvalds,
used with permission. No penguins were harmed in the production or maintenance
of this highly praised website. Looking for the
Indian Linux Users' Group? Try here. If you've read all this and aren't a lawyer: you should be!