Re: PATA sometimes faster than SATA?
- From: Alex Fraser <me@xxxxxxxxxxx>
- Date: Fri, 28 Mar 2008 19:48:41 +0000
A. J. Moss wrote:
A computer I'm looking at has an Abit NF7-S2 motherboard (socket A)
and a 300GB Seagate 7200.8 hard disk (ATA-100). I'm considering
swapping the hard disk for a SATA-150 one of the same type and
capacity.
I haven't yet checked to see if the IDE interfaces are connected
straight to the chipset and the SATA ones to the PCI bus, but if
this is the case, could it follow that the SATA solution (which
would use up about half the PCI bandwidth in sustained transfer
of large files) could actually be slower than the IDE one?
I am not familiar with the motherboard but it seems likely that you will indeed find that the SATA is connected via PCI. And so, yes, it could be slower than IDE, but I doubt it will matter much (if at all) unless you have other PCI devices taking lots of bandwidth (ie other storage controllers or gigabit networking).
On a similar note, one of my older computers has an Abit NF7-S v2
motherboard (socket A) and a Silicon Image two-port IDE RAID PCI
card, to which are attached four 500GB Seagate 7200.9 hard disks
(ATA-100) in a RAID-0. It will only be storing multi-GB files.
To optimise the throughput, the drive order is set to PM-SM-PS-SS,
instead of the default PM-PS-SM-SS. Would it also be beneficial
to set the stripe size to 512KB, rather than the default 128KB?
Performance with different configurations seems to depend on the card, so I would advise benchmarking it. That said, I have a feeling that (somewhat counter-intuitively) a small stripe size tends to be better for large files.
Alex
.
- References:
- PATA sometimes faster than SATA?
- From: A. J. Moss
- PATA sometimes faster than SATA?
- Prev by Date: Re: PATA sometimes faster than SATA?
- Next by Date: Re: Watching media PC to TV
- Previous by thread: Re: PATA sometimes faster than SATA?
- Next by thread: Re: PATA sometimes faster than SATA?
- Index(es):
Relevant Pages
|
|
