Re: No need to optimize in assembly anymore
From: Matt Taylor (para_at_tampabay.rr.com)
Date: 05/21/04
- Previous message: Matt Taylor: "Re: Newbie Question. How to determine DLL Export Prototype Parameters?"
- In reply to: Stephen Sprunk: "Re: No need to optimize in assembly anymore"
- Next in thread: Stephen Sprunk: "Re: No need to optimize in assembly anymore"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Fri, 21 May 2004 07:54:23 +0000 (UTC)
"Stephen Sprunk" <stephen@sprunk.org> wrote in message
news:010901c43edf$5b6b14f0$6401a8c0@stephen...
> "Matt Taylor" <para@tampabay.rr.com> wrote in message
> news:PfXqc.2409$6%6.1638@twister.tampabay.rr.com...
> > "Randall Hyde" <randyhyde@earthlink.net> wrote in message
> > news:xpVqc.2074$Tn6.1682@newsread1.news.pas.earthlink.net...
> > > You're better off with SCSI than a crazy number of SATA drives (at
> least,
> > > if performance is what you're seeking).
> > > A *parallel* version of SCSI (note that SCSI has three physical
layers:
> > > parallel, SCSI over SATA, and SCSI over firewire; currently parallel
> > > SCSI is the highest performing version).
> >
> > Aren't SCSI and fibrechannel intimately related as well?
>
> FC is a different physical medium with a superset of the SCSI control
> protocol (additions being needed for routing, redundancy, etc).
>
> > I would expect 1394 to underperform since the protocol has relatively
high
> > overhead and low transfer speeds. However, I am curious to know how far
> > apart U320 and SCSI-over-SATA are.
>
> I think you mean Serial-Attached SCSI, which has nothing to do with SATA
> except sharing the same physical connectors.
I mean whatever the original poster meant by "SCSI over SATA." I've never
heard of such a thing, but it isn't beyond belief either. As I understand,
ATAPI has some relationship with the SCSI protocol.
> > SATA seems competitive at least, though
> > at the moment most drives and chipsets use the converter chips.
Supposedly
> > the conversion adds something like 5% overhead; I presume that is
latency.
>
> Most SATA controllers coming to market in the last 3-6 mos are native, and
> native SATA drive electronics are no longer rare.
When I last checked around March, the only drives on the market that I found
with native SATA support were the Seagate disks (which have been native
since their initial launch). Even the Raptor uses the converter chip despite
its positioning as a high performance drive. I also looked at a number of
SATA RAID cards, and most had the Marvell converter chips clearly visible. I
have not seen these chips on motherboards, but then I presume (for other
reasons) that this is built-in to the south bridge.
Motherboard change far more frequently, and I can't really comment on them
since my newest board is an Abit IC7-G. That board is a year old, and I'm
fairly certain it converts SATA -> ATA; it is possible to construct a RAID
array with an ATA disk and an SATA disk, and besides that one ATA channel is
not usable when SATA disks are connected. Perhaps newer boards do support
SATA natively, but how would you know? The only boards I've ever seen with
the Marvell chips were boards using pre-SATA chipsets.
> > Medium aside, you're not going to find 15K rpm SATA disks.
>
> Maybe not today, but 10krpm drives are available. 15krpm drives are still
> rare in the SCSI/FC space. The interface is orthogonal to the drive
motor;
> the lack of 15krpm drives today is purely a marketing/profitability
> decision.
I recognize this, but I am pointing out that 15K rpm disks aren't an option
yet in the SATA space, and there isn't any indication that they will be in
the near future.
There aren't a lot of options with 10K disks, either; you either have the
Raptor I or the Raptor II. Somebody else (Hitachi?) was working on 10K rpm
disks, but I don't recall who it was or when they were going to hit the
market.
-Matt
- Previous message: Matt Taylor: "Re: Newbie Question. How to determine DLL Export Prototype Parameters?"
- In reply to: Stephen Sprunk: "Re: No need to optimize in assembly anymore"
- Next in thread: Stephen Sprunk: "Re: No need to optimize in assembly anymore"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|