Re: 32 Bit vs 16 bit I/O

[0 <fruitless orange net>]

tony d wrote:

> tony d wrote:
>
> > > /dev/hda:
> > > multcount = 16 (on)
> > > IO_support = 1 (32-bit)
> > > unmaskirq = 0 (off)
> > > using_dma = 1 (on)
> > > keepsettings = 0 (off)
> > > readonly = 0 (off)
> > > readahead = 8 (on)
> > > geometry = 7476/255/63, sectors = 120103200, start = 0
> > > [root localhost tony]# /sbin/hdparm -tT /dev/hda
> > >
> > > /dev/hda:
> > > Timing buffer-cache reads: 128 MB in 0.16 seconds =800.00 MB/sec
> > > Timing buffered disk reads: 64 MB in 1.17 seconds = 54.70 MB/sec
> >
> >
> >
> >
> > And heres what 16 bit I/O gives
> >
> > /dev/hda:
> > multcount = 16 (on)
> > IO_support = 0 (default 16-bit)
> > unmaskirq = 0 (off)
> > using_dma = 1 (on)
> > keepsettings = 0 (off)
> > readonly = 0 (off)
> > readahead = 8 (on)
> > geometry = 7476/255/63, sectors = 120103200, start = 0
> > [root localhost tony]# /sbin/hdparm -tT /dev/hda
> >
> > /dev/hda:
> > Timing buffer-cache reads: 128 MB in 0.16 seconds =800.00 MB/sec
> > Timing buffered disk reads: 64 MB in 1.18 seconds = 54.24 MB/sec
> >
> > am i missing somthing here or is there just not much difference in 
> > the two I/O
> > modes.
> > i read thru the man-hparm pages does any one have a link to more HD 
> > tunning
> > info?
> > Thanks
> > Tony 
>
>
> so after some more looking around i tried these setting and got this
>
> /dev/hda:
> multcount    = 16 (on)
> IO_support   =  3 (32-bit w/sync)
> unmaskirq    =  1 (on)
> using_dma    =  1 (on)
> keepsettings =  0 (off)
> readonly     =  0 (off)
> readahead    =  8 (on)
> geometry     = 7476/255/63, sectors = 120103200, start = 0
> [root localhost tony]# /sbin/hdparm -tT /dev/hda
>
> /dev/hda:
> Timing buffer-cache reads:   128 MB in  0.16 seconds =800.00 MB/sec
> Timing buffered disk reads:  64 MB in  2.20 seconds = 29.09 MB/sec
>
> Hmm not so good about a 35% slower disk read
> Tony
>
> > >
> >
> >
> > _______________________________________________
> > Redhat-install-list mailing list
> > Redhat-install-list redhat com
> > https://www.redhat.com/mailman/listinfo/redhat-install-list
> > To Unsubscribe Go To ABOVE URL or send a message to:
> > redhat-install-list-request redhat com
> > Subject: unsubscribe
>
>
>
> _______________________________________________
> Redhat-install-list mailing list
> Redhat-install-list redhat com
> https://www.redhat.com/mailman/listinfo/redhat-install-list
> To Unsubscribe Go To ABOVE URL or send a message to:
> redhat-install-list-request redhat com
> Subject: unsubscribe
Setting hdparm -c3 uses extra CPU overhead for the sync signal (see man 
hdparm) so the lower throughput might be down to that. Setting -c1 is 
thus better if your disk/controller can do it. As for general speed 
increase over default 16bit io (-c0), I didn't get an improvement 
either. But hey, thats life (or the possibility its not actually 
working). You can take comfort in the the fact that your buffered disk 
read is faster than my buffer-cache read (renewed sense of upgrade urgency).

Stu C