On Tue, 25 Aug 1998, Lasciate ogni speranza, voi ch'entrate! 25-Aug-1998 2018 +0100 wrote:

> "jbender@ezol.com" "Jim Bender" wrote
> >>
> >>The SDI bus be clocked at up to 22Mbits/sec or so with a max cable length of
> >>maybe 80 feet (plus some allowance in the start and end cabs). The RA81
> >>(introduced in 1982?) has a peak transfer rate of 17.4Mbits/sec and a capacity
> >>of 456MB. By way of contrast the RA92 (1.5GB) runs at a peak transfer rate of
> >>22.16Mbits/sec.
> >
> >Then why were they so damn slow even when connected to a "fast" VAX on a
> >fast HSC or even a high end SDI controller?    So slow we were wishing we
> >had our Massbuss disks back.    :-(
> 
> Assuming this isn't a rhetorical question:
> 
> Measuring disk "speed" with a single number is hard if not impossible. The peak
> transfer rate is the best anyone could get shoving data downhill with a
> following wind. You won't get anywhere near that in practice.
> 
> The RA81 had an average access time of 36.3ms (about the same as an RD54!). The
> RA92 got that down to 24ms, the (SCSI) RZ29B is down at 12.2ms, the Fireball in
> my PC is 9.5ms.

So the upshot is, that while both ends of the BUS (host controller and drive
circuitry) can move data at a pretty good clip, the physical drives DEC 
attached to it were not up to the task.   I have never seen good transfer 
rates from ANY RAxx series drive on ANY DEC CPU under ANY OS.   
 
And it's not like they traded speed for reliability either...   Everyone
remember what pieces of junk RA81's  were/are?    We had 12 of them running
at once at one point and it was a nightmare keeping them going.    The
RA9x's we replaced them with were a godsend as far as reliability (no probs
except for toasted power supplies) but no improvement in speed.   

> The same deal holds with these 32x, 36x, 40x CDROM drives. The earlier versions
> had (relatively) poor access times, so they were not that much faster than a
> good8x or 12x drive. The latest crop of 32x drives have access times in the
> 60ms region and feel much better (at least to me).
> 
> So what on earth possessed you to get rid of your Massbuss drives :-)

No Massbuss adapter for BI machines...  at least not the ones we were
running.     :-(
 
> Antonio
> 
> Antonio Carlini                            Mail: carlini@marvin.enet.dec.com
> DECnet-Plus for OpenVMS Engineering
> Digital Equipment Corporation              Worton Grange, Reading, England
> 

Jim