[conspire] How to check if a mobo is compatible with

Rick Moen rick at linuxmafia.com
Sun Mar 30 14:18:41 PDT 2008


Quoting David Fox (dfox94085 at gmail.com):

[snip a lot of good advice, with which I concur and which I appreciate.]


> Well, that's a 150 - so it should work. The MB in question does
> support SATA up to 150 meg/sec - surely fast enough, but the MB
> doesn't advertise specifically SATA 3.0, which is the newer standard
> And 7200 is fast enough for most stuff - here, the bottleneck is
> likely to be disk transfer speeds, not the interface.
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Quite.  It's important to realise that, with ATA (PATA or SATA), the
theoretical bus-speed limit is almost never the limiting factor.  The
bus ceiling is typically so extremely high compared to the speed of
physically reading data from the hard drive in all but very artificially
contrived situations (and maybe not even then) that, e.g., 150 vs 300
MB/sec speed limits on the HBA bus really don't make a damned bit of
difference.

(If you're driving a Citroen 2CV, it doesn't make a whole lot of
difference whether the speed limit is 55 or 70 MPH.)

The other things to know about hard drives (aside from the 10kRPM = runs
hot datum that David also mentioned) is that the slowest thing a drive
can do, by far, is "seek", i.e., have to move the heads from track to
track.  The second slowest operation, but an order of magnitude or two
faster, is spin to bring the correct sectors back underneath the heads.

So, 10kRPM is better than 7200 RPM is better than 5400 RPM is better
than 3600 RPM on account of lower "rotational latency".  Drives with
lower average seek times are (much) better on account of seek latency.

...and clustering frequently-visited parts of your Linux filesystems on
either side of the swap partition in the order likely to minimise the
average seek distance helps performance and lowers disk wear.

...and having two drive "spindles" (physical disks) to split one's Linux
system between is even better in that department, because it lets you
further minimise average seek time/distance and split the swap activity
between the two drives.  (Just create swap on each drive.  Linux's
swapper process then uses both very intelligently, without further
help.)





More information about the conspire mailing list