Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!orstcs!ogccse!blake!bissiri
From: bissiri@blake.acs.washington.edu (Moja Fritzah)
Newsgroups: comp.sys.atari.st
Subject: Re: HELP IDENTIFYING SEAGATE 296N WITH REV. 7 ROM
Message-ID: <3347@blake.acs.washington.edu>
Date: 23 Aug 89 00:52:15 GMT
References: <2520@orion.cf.uci.edu> <5440047@hplsla.HP.COM>
Reply-To: bissiri@blake.acs.washington.edu (Moja Fritzah)
Organization: University of Washington, Seattle
Lines: 48

In article <5440047@hplsla.HP.COM> andyc@hplsla.HP.COM (Andy Cassino) writes:
>bissiri@blake.acs.washington.edu writes:
>| 
>| Thirdly, the packing slip from the manufacturer gives the drive's specs.
>| 
>| Note that many folks with the ROM 8 296N are quite happy with
>| 2:1 interleave and about 400+k data transfer rate.  After all, most
>| things coming off your drive are relatively small.  Unless you have
>| disk intensive operations going all the time, you may be satisfied with
>| ROM 8's.
>| 
>
>I'll also verify this. My experience is that the FAT lookup is a much bigger
>bottleneck than transfer rate - even with FATSPEED. Also, when loading large
>programs, I notice that quite often it takes the program 2-3 times longer to
>wake up than it took to get it off disk. Increasing transfer rate 30% 
                                                                   ^^^^
    The math is really closer to a 70% difference:
    My Quantum at 1:1 delivers 675k/sec versus a 2:1 Seagate 296N
    delivering 400+k/sec.  To me, that's almost a 70% improvement
    with the Quantum.  

>isn't going to be very helpful. My guess is that handling very large data 
>files is where the transfer rate is going to get in your way - then you 
>want to try something like the Quantum Kevin mentioned.
>

     Or, if after the program is "awake", it still needs to pull
     stuff off the drive on a frequent basis.  

>I'd also mention that a good disk cache is really the best way to speed
>up some disk-intensive operations like compiles, short of putting all the
>libraries, files and executables in a RAM disk, that is. I use DCACHE and
>with 200-300K of cache, recompiles often only make one or two brief disk
>accesses.

     The Quantum has a programmable 64k cache/pre-fetch that is very noticeble...
     Seems like a zero seek rate.  
     I haven't tried DCACHE yet... may interfere with my drive...
     we'll see.

     My only comment about the Seagate 296N is that it was supposed to
     be a 1:1 interleave drive, and people paid money for that spec.
     As far as I know, Seagate has yet to deal fairly  with its customers
     who purchased the ROM8 version of the 296N.

-kevin
bissiri@blake.acs.washington.edu