[mythtv-users] SATA and DMA (SIIG 3112)

Brian Wood beww at beww.org
Fri Mar 31 23:54:44 UTC 2006



On Mar 31, 2006, at 4:21 PM, leland sindt wrote:

>
>> I record on a RAID-0 software  array with 2 SATA drives:
>>
>> /dev/hda:
>> Timing cached reads:   2608 MB in  2.00 seconds = 1303.92 MB/sec
>> Timing buffered disk reads:  184 MB in  3.00 seconds =  61.33 MB/sec
>>
>>
>> /dev/md0:
>> Timing cached reads:   2648 MB in  2.00 seconds = 1323.92 MB/sec
>> Timing buffered disk reads:  338 MB in  3.03 seconds = 111.62 MB/sec
>>
> Those look like good hdparm numbers, how does your system respond? How
> hard are you pushing it? I am looking to sustain 2 SD and 1 HD
> recordings, (at least) one front end using XvMC, and mythcommflag.
>

This is an Athlon64 3700+ system w/1GB RAM, I have four PVR-150s in  
it. Using an nVidia 5700 video card with the 7676 drivers, not using  
XvMC. No HD in this system.


I once had four recordings happening, 2 mythcommflag jobs running at  
"medium" CPU usage and was watching a recording playback, all with no  
noticeable problems, obviously that was a "worst case" test. CPU load  
was not high, maybe 30% at peaks, usually less.

2 SD and one HD as you describe would be an even higher load on the  
drive system, assuming that by "HD" you mean 1080i. XvMC would take  
some load off the CPU but wouldn't help the stress on the drive  
system, you still have to move the data off the drive(s) at the same  
speed no matter how it's getting translated to the screen.

For comparison, another system has the following numbers:

/dev/hda:
Timing cached reads:   3796 MB in  2.00 seconds = 1897.88 MB/sec
Timing buffered disk reads:  170 MB in  3.01 seconds =  56.51 MB/sec

/dev/md0:
Timing cached reads:   3708 MB in  2.00 seconds = 1853.88 MB/sec
Timing buffered disk reads:  342 MB in  3.00 seconds = 113.99 MB/sec

This is an AMD FX-60 dual-core box with the same drives, you can see  
the faster CPU. This box does *not* run Myth, that would be overkill.




More information about the mythtv-users mailing list