Conference hydra::amiga_v1

Re: -.1

Hi Vince,

we (the Software Brewery) will be stocking up our storage space on several of
our systems with the 2614S drives in the week from the 7. to 13. January.
If nobody else should respond 'till then, I'll surely be able to do so. ;-)

Cheers,

<CB>

    Hi Christian,
    
    By fooling around, I found the mystery...
    My WorkBench 1.3 upgrade book had an error by saying that
    the mountlist entry "MaxTransfer" being the max "SECTOR" i/o
    per transfer.  It is infact the max BYTE...
    MaxTransfer should be 130560(?) if I remember correctly.
    DiskPerf or DiskSpeed were transfering only ONE sector in
    the buffers which were showing no change from 512 bytes buffer
    to 256K.
    
    Now with 256K buffer the fujitsu 2614S is going way over 500Kb/sec...
    
    Pretty good considering the controler being a A2090A. I am eventually
    planning to buy a HardFrame controler. (As soon as I finish paying
    the hardDrive...)
    
    The drive is working like a charm, not a single bad spot out of
    182 Meg.
    
    Oh, before I forget, the formatter keep jamming (just hang there)
    at cyl 1309. I guess that 1309 to 1334 must be reserved for bad
    sectors/self test diags?.  Anyway, I put my MaxCyl in the mountlist
    at 1308.  Let me know if you can get beyond 1309...
    
    Enjoy!
    Vince.

Re: -.1

>    MaxTransfer should be 130560(?) if I remember correctly.
With the A2090(a) there's no need to specify a MaxTransfer value.
MaxTransfer is needed for braindead drivers, that don't know how to limit
transfer request to a size they can handle themselves. 
Don't limit your controller by artificial boundaries.

However, you might wan't to set your MASK value to something that stops the
controller even considering writting directly to chip memory. I'll bring
my mountlist along one of these days.
    
>    Pretty good considering the controler being a A2090A. I am eventually
>    planning to buy a HardFrame controler. (As soon as I finish paying
>    the hardDrive...)

The problems with the A2090(a) are:
a) DMA contention with deep hi-res screens and
b) insufficient/non-existent SCSIdirect and RigidDiskBlock support.

While a) can only be partly remedied by software, b) can. Both problems will
be addressed and eventualy solved this year...(don't quote me ;-)
The A2090(a) isn't a slow controller, it has a pretty solid design. The only
controllers I'd prefer to it are the A2091, GVP-II and HardFrame (in that 
order). But all of them don't give you ST-506 support and those two RD53's
are quite nice to have. ;-)
    
>    at 1308.  Let me know if you can get beyond 1309...
I'll do.

Happy new year,

<CB>

    Hi <CB>,
    
    You're saying that MaxTransfer is needed only for braindead drivers,
    but by specifying 130560, I got >10 times more troughput. Obviously,
    I use the commodore harddisk.device.  Do you mean that I should not
    specify any maxtransfer in the mountlist?.  But anyway, it works great
    with 130560, and I have a philosophy: "It works, don't mess with it"...
    BTW: the mask is 0xfffffe (by memory, or very close to that value.)
    
    I also tried to play with the interleave and it doesn't change a thing.
    I assume that SCSI couldn't care less about interleaving, since it
    access by block number isn't it?  The "low level" format doesn't allow
    to interleave anyway (PREP).  I am not sure about the hardframe low level
    format?
    
    Vince.

Re: -.1

>    Do you mean that I should not specify any maxtransfer in the mountlist?
Exactly...

>    BTW: the mask is 0xfffffe (by memory, or very close to that value.)
My mask value is 0x007ffffe, a recommendation by Andy Finkel and an Amiga
memory map will show you that it makes sense...
    
>    I also tried to play with the interleave and it doesn't change a thing.
>    I assume that SCSI couldn't care less about interleaving, since it
>    access by block number isn't it?  The "low level" format doesn't allow
>    to interleave anyway (PREP).  I am not sure about the hardframe low level
>    format?

Yup, SCSI drives couldn't care less about interleave values in the mountlist.
However, other lowlevel format tools allow you to force the HD to conform your
formatting wishes. But any half-witted SCSI drive (which are about any drives
besides the older and slower Seagates) will know best on how to format itself,
so it should be allowed to do that. ;-)

<CB>

I know that this really isn't related, but does anyone know if it is possible 
to disconnect that fan in an Seagate ST277-N drive. The noise of both the Ami
2000 fan AND the seagates is DEAFENING!!!!!

Cheers, dave

    Does yours have a fan in it?  Mine doesn't.  It still makes a lot of
    noise, but that is the drive motor.  The power supply chassis the drive
    is mounted in DOES have a fan, but I'll leave that there to prevent
    overheating.
    
    Tony

    
    ST-277N's do not have fans.

Re: .2

Nope, I can't get it go beyound 1308, too. My drive got pretty confused and
refused further cooperation until I reseted the system.
I'm checking if this is a hddisk.device or a Fujitsu feature.
BTW, our drives turned out to be the ESA type which is a tad better then the
plain S drives, having an average access time of 20ms instead of 25ms.
DiskSpeed 3.0 reports the drive r/w transfer at ~650KB/s max.
So far, I'm very pleased with the drive (speed, noise, power consumption).

Cheers,

<CB>

    Hi <CB>,
    That's what I figures also,  if you ever find out about
    the 1308 maximum cyl, let me know.  I have the fujitsu
    catalog and it doesn't show the ESA version. Is it the
    only difference?
    Mine give me troughput at 550 using diskspeed 3.0, but
    I can't complain since I paid about $520 US for close
    to 182 megaBytes, it's still a good price. Just for
    curiosity, how much more is the ESA version?
    
    One last question: Is your drive stop once every 5 minutes
    (more or less) for about 1 or 2 seconds, make some sort of
    re-positioning noise and keep on going? (Even if not being
    used it does it)  It is not a defect since there are no
    error message regardless of the program running. Not that
    it bothers me really, but do you know why it does it.
    
    Vince.

RE: -.1

>    Mine give me troughput at 550 using diskspeed 3.0, but
>    I can't complain since I paid about $520 US for close
>    to 182 megaBytes, it's still a good price. Just for
>    curiosity, how much more is the ESA version?

That price is indeed very good at least judging from an European point
of view. The 2614 is advertized for about $1200 in German magazines and we
were able to get hold of them for half of that. ;-) 
    
>    One last question: Is your drive stop once every 5 minutes
>    (more or less) for about 1 or 2 seconds, make some sort of
>    re-positioning noise and keep on going? (Even if not being
>    used it does it)  It is not a defect since there are no
>    error message regardless of the program running. Not that
>    it bothers me really, but do you know why it does it.

Yup, it does. This is a feature of most modern drives. They try to
recalibrate themselves when they get in doubt of their own integrity.
The frequency of the Fujitsu is somewhat higher than what I _heard_ of
other drives, but as long as it doesn't increase dramatically everything
should be fine.

We'll be getting the technical documentation of that drive shortly, maybe
it also addresses the differences between the ESA and S types.

<CB>