Gustin Johnson wrote:
> Ralf Mardorf wrote:
> <snip>
>>>> but making hard disks sleep is a PITA on Linux,
>>>>
>>> indeed.
>>>
> Really? I have I used this ability under Linux for the better part of a
> decade.
>> :(
still, setting it up properly is not trivial.
>>>> hdparm doesn't work on many machines, so the hard
>>>> disks are running all the time, even if I only need one of two devices.
>>>>
> It has worked on pretty much every machine I have ever owned. For USB
> connected devices I have used sdparm. The syntax is different so you
> will need to RTFM.
>
>>> It's not only hdparam but also the 'noatime' mount option,
>>> /proc/sys/vm/dirty_writeback_centisecs (or /etc/sysctl.conf) and ext3's
>>> journal_data_writeback option (`man tune2fs`).
>>> ..and beware: in case of power-loss it can be dangerous to have fiddled
>>> with those!!!
>
> noatime is not dangerous,
right, sorry. I should have said the latter two are dangerous.
> but does not really gain anything either since
> it happens when you are accessing a disk.
..or when re-accessing some file that is already in the disk-cache.
It probably does not have any impact on audio workstations, but it does
matter when listening to mp3s on the road for example. It spins up the
disk again.
> Messing with journal options
> is probably not the best idea, especially on a DAW where data integrity
> would presumably be important.
;-)
> Usually you only see the noatime and nodatime options when using write
> sensitive media like compact flash. In those cases you would not be
> using ext3 anyway.
>>> I'm not sure if /etc/hdparm.conf accepts wildcards, but I'd hazard a
>>> guess that
>>> /dev/discs/disc*/disc {
>>> spindown_time = 240
>>> }
>>> may be a good default for 64studio.
>>>
>
>>>
>>>> It's written that Intel can't scale each core of a CPU independently,
>
> Anything older than the Core i7 does not scale independently. I did not
> know that AMD could do this, perhaps they got something right with AM2
> after all.
>
_______________________________________________
64studio-users mailing list
[email protected]
http://lists.64studio.com/mailman/listinfo/64studio-users
