Conference hydra::amiga_v1

****************************************************************
*                                                              *
*          SCSI Controller Performance Evaluations             *
*                                                              *
****************************************************************


                                     by:  E.J. Lippert II
                                          C Ltd.





Philosophy and Mechanics of the Testing Procedure



Disclaimer:

    These  tests  were  conducted  by  C Ltd.  at C Ltd.  and all readers should
therefore  consider  the probability that some of bias  may  be  induced  into
the results of the tests.  While all possible effort was expended to maintain
objectivity, it is quite probable  that unintentional advantages may have been
provided to the  company's  own  products due to the fact that the person (Ed
Lippert)  conducting  the tests was simply more familiar with the operation of
the C Ltd.  products.

    Additionally,  no  attempts  were  made  to  tamper  with the software  or
configuration  data  supplied with each of the SCSI controllers.   In all cases,
except as noted in the test results' notes, the initialization and formatting
software supplied by the manufacturer was used intact and exactly as provided.


Equipment Used:

    In  addition to the specific Hard Drives and SCSI controllers (as  specified
within  each test) the equipment used included an Amiga  A-2000  (Rev  4.3
mother  board/Rev.  1.2 Roms) with dual floppy  drives  and  a  Micron  2
Megabyte memory expansion board installed  in  slot  5.   The  monitor used was
an Amiga 1084 RGB monitor.   Unless  specifically  stated within the summary of
the test  results,  no other hardware was used at any time during the testing.


Test Procedure:


    For each test:

        All  Hard  Drives  were configured with a SCSI address of
        Zero.

        The  specific  SCSI controller and Hard Drive combination
        to be tested was mounted in the Amiga-2000.

        The  system  was  booted  with  the  disk provided by the
        manufacturer of the SCSI controller under test.*

        The  Hard  Drive  under  test was formatted and otherwise
        configured  as  per  the  instructions  provided  by  the
        manufacturer of the SCSI controller under test.

        The  system  was  then  re-booted as per the instructions
        provided by the manufacturer of the SCSI controller under
        test.

        The  Devspeed speed test program was run with output from
        the program redirected to a file on a floppy disk in DF1:
        as in the example below.

            DF1:Devspeed >DF1:KRONOS_ST-225N CLtd.device 1000 0


*Note:   In  some cases when a particular Hard Drive could not be configured
when  using  a  specific  SCSI  controller additional attempts  were  made
(using the testers `best efforts') to enable the  specific  SCSI controller/Hard
Drive combination so that the test  could  be completed.  In these cases, the
specific problems encountered  and  the  actions  taken  are  noted within the
test results for the specific SCSI controller/Hard Drive combination.


Operation of the Devspeed Program:


     When  executed,  the  Devspeed program calls specific device driver
software  supplied  by  the  manufacturer  of  the  SCSI controller  (similar
to calling the AmigaDOS trackdisk.device to read from floppy disks) with a
request to read a specified number of  sectors  of  data from the DEVICE being
tested into a buffer.  The  time required to read the data is saved to be used
latter to calculate  the  speed  of  the device.  Then Devspeed again calls
trackdisk.device,  but  this  time  the program requests that the data  in  the
buffer be written to the device.  The time required to  write  the  data  is
saved to be used latter to calculate the speed  of  the device.  This procedure
is repeated five times and the  read  and  write  times are then averaged and
the calculated performance  (in  BYTES  of  data  per  second) is written to the
screen.   This  procedure  is  then  repeated  using successively larger blocks
of data.


    Devspeed is also a non-destructive test in that it reads data from  the
drive  and  then  writes the data back out to the same sectors from which it
came.  This, therefore allows users to test drives  that  contain  data without
having to clear any space for the test.

Testing Objectives:

    The  testing  procedure  outlined  herein  was used on a wide variety  of
SCSI  controller/Hard  Drive  combinations  for  the purpose  of determining the
answers to two questions.  Which SCSI controller provided the best overall
performance with the variety of   Hard   Drives  tested?   Which  SCSI
controller/Hard  Drive combination provide the fastest overall performance?


Why the Devspeed Program was Used.

    There  are  literally  hundreds of thousands of objective and subjective
ways to test the performance of SCSI controllers.  Our first objective was to
eliminate as much subjectivity as possible from the testing procedure.  Our
second objective was to make the test  simple  to administer and consistent in
its outcome so that the  results  could be easily confirmed by any outside
individual or company wishing to do so.

    The  Devspeed  program  was  written  specifically to be used within this
series of tests because it fulfilled the above stated objectives  better  than
any other available method.  The reason that  Devspeed  was  used  instead of
the more prevalent Diskperf program  was  that  Devspeed  reads  and  writes
directly to the AmigaDOS  Trackdisk.device  which eliminates any contamination
of the results which may be caused by the normal Amiga File System's optional
buffering  scheme  and other non-controllable variables like disk fragmentation.

    The biggest single drawback of the Diskperf test (as has been vocalized  by
many of our and other manufacturers' customers) was that  the  results  are  not
easily  confirmed.   For example, a Diskperf  test  of  a system done by a
manufacturer could yield a result  of  XXX bytes/sec because the manufacturer
took an lot of time  to change the various buffer sizes and methods to yield the
best results on that specific (Diskperf) test, while a user using exactly  the
same  equipment, but using the default buffer sizes etc.,   could   not
reproduce   the  speed  as  stated  by  the manufacturer.

    When  testing  a  C Ltd.  SCSI controller, for example, there are  255
possible buffer settings within the C Ltd.  SCSI driver software  and  two
optional  buffering  methods.   There  are an additional  255 possible buffer
sized available within the filing system (set from the mountlist) and then
another 255 buffers that can  be  added  using the AmigaDOS AddBuffers command.
In almost every  instance, these various buffer sizes are interactive so it is
impossible  to  accurately  predict  the results of using any specific  set  of
values.   It  is  even unlikely that one could predict  changing  one  value
within  a  set  of values based on previous   test  results.   Therefore  to
optimize  the  system, 33,162,750  (255  *  2  *  255 * 255) tests should be run
on each system  to  determine the best over all preforming combination of
values.  (In reality, some values, and even ranges of values, can be  eliminated
making  the  task some what reasonable.  But even allowing  for  those
eliminated tests, the task is a substantial one  requiring  no  less  than  120
man/hours  to  complete  the optimization of just one SCSI controller/Hard Drive
combination.)

    An  interesting  side note is that the set-up values obtained through  all
of  the  above  testing  is SELDOM IF EVER the best set-up   to  use  for
normal  Hard  Drive  operations.   Systems optimised  for  Diskperf  are  not
generally even close to being optimised for `real world' operation.


    Because  the  Devspeed  test  bypasses  all  of  the AmigaDOS buffering
schemes,  it  provides  a true and `repeatable' number which is an accurate
representation of the `actual' data transfer rate  of  the  hardware used within
the test.  Because additional overhead  involved  in  the AmigaDOS filing
system, Diskperf test results  SHOULD  be  lower that those obtained with
Devspeed.  In some cases, Diskperf results may be higher than those obtained by
devspeed.   It  is, however, impossible to read and write data TO THE  HARD
DRIVE  faster  through  the  DOS  filing  system using Diskperf   than  through
actual  device  driver  software  using Devspeed.   The  only conclusion that
can be reached here is that the  Diskperf  test result is skewed because some
(or all) of the data  is  actually  written to a buffer in the computer
somewhere and  then  to the drive, while reading the data is then only from the
buffer (which is still holding the written data) and not from the drive.

    Devspeed is also a non-destructive test in that it reads data from  the
drive  and  then  writes the data back out to the same sectors from which it
came.  This, therefore allows users to test drives  that  contain  data without
having to clear any space for the test.



DMA vs. non-DMA SCSI Controller Myth:

    There  has  been  a  pervasive  myth that holds that DMA SCSI controllers
are, by design, faster than non-DMA SCSI controllers.  While  it  is  true that
in some cases the DMA controllers tested were  faster than non-DMA controllers,
it is equally as true that in some cases the non-DMA controllers tested were
faster than DMA controllers.  The truth is:

  "On a standard Amiga, well designed DMA and non-DMA controllers will  provide
roughly equal performance.  DMA controllers can suffer  substantial performance
degradation during high levels of DMA activity, including, for example, the
display of hi-res over-scan  video.  Non-DMA designs, however, can suffer either
(as  determined by interrupt settings) performance degradation from high levels
of processor activity or cause a slow down of data  processing during Hard Disk
activity.  When higher speed 68020/68030  processors  and/or co-processors are
added to the system, the edge in performance tips substantially towards the
non-DMA controllers because they are able to take advantage of the  faster
processors  while  the  DMA controllers are still limited to moving data at the
(un-changed) DMA bus speed."



Exploding Other Myths:


    While  we are on the subject of DMA/non-DMA SCSI controllers, it  should  be
noted  that,  mis-leading advertising hype to the contrary,  the  GVP SCSI
controller IS NOT a DMA controller.  The GVP  controller  uses the DMA mode of
the SCSI controller chip to transfer  data  into a buffer on the GVP controller
and then uses 68000  processor  to  transfer  the  data  from the buffer to the
Amiga's  memory.  (This is the exact same design that has been in use  by C
Ltd.'s SCSI series controllers since their introduction in  1986,  almost  two
years  prior  to  the  release of the GVP products.   This similarity of design
may account for some of the similarities  in  the  performance  of the GVP
controller and the older C Ltd.  SCSI series controllers.)   This type of
controller can  be  called  interrupt  driven,  processor controlled or even
pseudo-DMA, but it is NOT a DMA DEVICE.


The Hard Drive Average Track Access Time Myth.

    Much  ado  has  been  made  in  advertising  lately about the `average
track  access  time' of Hard Drives.  There are several points that need to be
made on this subject.  First, the `average track  access  time'  of a Hard Drive
is a measure of the average time  required  for  the  Hard  Drive  to  move
from the current (random)   track   being   accessed  to  another  (random)
track containing  data  on the Hard Drive.  `Average track access time' IS  AN
ATTRIBUTE OF THE HARD DRIVE AND HAS NOTHING WHATSOEVER TO DO  WITH  THE  SCSI
CONTROLLER  CONNECTED TO IT.  So if one Hard Drive  is faster than another, it
will generally be faster on any SCSI  controller  to  which  it  is  connected.
The fact that a controller is sold with a FAST Hard Drive, should not be taken
to infer that the controller is fast, in fact, the contrary is quite often the
case.  A manufacturer of a SLOW controller might decide to  mate  a  FAST  Hard
Drive  to  the  controller  in  order to compensate for the slow performance of
the controller.

     A second and very important point is that the `average track access  time'
of a Hard Drive is not the only measurement of the performance  of  a  Hard
Drive.   As  can  be seen from the test results, the Rodime RO-652 with an
`average track access time' of 65ms.   performed  equal to or better than most
of the 40ms.  and 28ms.   drives  tested.   Don't be fooled into believing the
MYTH that  a  faster `average track access time' automatically means a faster
drive.

    The  third and final point to be addressed on this subject is nothing  more
than the logical conclusion that the specifications related  to a Hard Drive
mechanism may not be a true indicator of the performance of the complete Hard
Drive system as installed on an  Amiga  computer.   Systems  and  controllers
can  in fact be designed  to perform better (or worse) with specific products
and therefore the only meaningful specifications are those that refer to  the
performance of the complete system  (ie. the combination of  the  Hard  Drive,
the SCSI controller and the computer) when tested as a complete functioning
unit.









****************************************************************
*                                                              *
*          SCSI Controller Performance Evaluations             *
*                                                              *
****************************************************************





SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-225N
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  315076  Writes  307200  b/sec
 153 K    Reads  332108  Writes  326230  b/sec
 307 K    Reads  338201  Writes  342920  b/sec
 512 K    Reads  343240  Writes  341333  b/sec
 768 K    Reads  343240  Writes  343880  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Rodime
Model:            RO-652
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  585142  Writes  585142  b/sec
 153 K    Reads  519211  Writes  504986  b/sec
 307 K    Reads  554345  Writes  538160  b/sec
 512 K    Reads  574205  Writes  574205  b/sec
 768 K    Reads  587006  Writes  587006  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Syquest
Model:            SQ-555
Capacity:         44 Meg.
Avg. Access:      28ms.


NOTES:  This is a removable media cartridge type Hard Drive.


  51 K    Reads  558545  Writes  558545  b/sec
 153 K    Reads  567138  Writes  542117  b/sec
 307 K    Reads  599414  Writes  567138  b/sec
 512 K    Reads  596504  Writes  590769  b/sec
 768 K    Reads  608316  Writes  596504  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-157N
Capacity:         48 Meg.
Avg. Access:      28ms.


  51 K    Reads  614400  Writes  614400  b/sec
 153 K    Reads  658285  Writes  646736  b/sec
 307 K    Reads  658285  Writes  670254  b/sec
 512 K    Reads  667826  Writes  667826  b/sec
 768 K    Reads  670254  Writes  667826  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-277N-1
Capacity:         65 Meg.
Avg. Access:      28ms.


  51 K    Reads  768000  Writes  558545  b/sec
 153 K    Reads  819200  Writes  737280  b/sec
 307 K    Reads  921600  Writes  857302  b/sec
 512 K    Reads  884028  Writes  847448  b/sec
 768 K    Reads  873554  Writes  849400  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Maxtor
Model:            XT-3170
Capacity:         135 Meg.
Avg. Access:      23ms.


  51 K    Reads  512000  Writes  512000  b/sec
 153 K    Reads  542117  Writes  567138  b/sec
 307 K    Reads  554345  Writes  594580  b/sec
 512 K    Reads  558545  Writes  626938  b/sec
 768 K    Reads  568888  Writes  616454  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          KRONOS/2000 (NEW!)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Kalok
Model:            Octo-30
Capacity:         30 Meg
Avg. Access:      65ms.


NOTES:   This drive is an ST-506 drive and was tested
         with an OMTI 3527 SCSI/ST-506 device controller.


  51 K    Reads  361411  Writes  438857  b/sec
 153 K    Reads  414202  Writes  455111  b/sec
 307 K    Reads  469605  Writes  519211  b/sec
 512 K    Reads  481882  Writes  568888  b/sec
 768 K    Reads  488912  Writes  581451  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-225N
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  201442  Writes   26947  b/sec
 153 K    Reads  200347  Writes   27026  b/sec
 307 K    Reads  207684  Writes   27076  b/sec
 512 K    Reads  206868  Writes   27125  b/sec
 768 K    Reads  205714  Writes   27105  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Rodime
Model:            RO-652
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  231849  Writes  201442  b/sec
 153 K    Reads  222072  Writes  196085  b/sec
 307 K    Reads  231849  Writes  198193  b/sec
 512 K    Reads  233612  Writes  202105  b/sec
 768 K    Reads  236005  Writes  203893  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Syquest
Model:            SQ-555
Capacity:         44 Meg.
Avg. Access:      28ms.


NOTES:  This is a removable media cartridge type Hard Drive.


  51 K    Reads  372363  Writes  211862  b/sec
 153 K    Reads  423724  Writes  223418  b/sec
 307 K    Reads  446836  Writes  224097  b/sec
 512 K    Reads  443610  Writes  224233  b/sec
 768 K    Reads  447378  Writes  224506  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-157N
Capacity:         48 Meg.
Avg. Access:      28ms.


  51 K    Reads  351085  Writes   54857  b/sec
 153 K    Reads  354461  Writes   54694  b/sec
 307 K    Reads  361411  Writes   54734  b/sec
 512 K    Reads  363550  Writes   54710  b/sec
 768 K    Reads  362834  Writes   54678  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-277N-1
Capacity:         65 Meg.
Avg. Access:      28ms.


  51 K    Reads  423724  Writes   54613  b/sec
 153 K    Reads  449560  Writes   54613  b/sec
 307 K    Reads  469605  Writes   54694  b/sec
 512 K    Reads  467224  Writes   54661  b/sec
 768 K    Reads  472615  Writes   54661  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Maxtor
Model:            XT-3170
Capacity:         135 Meg.
Avg. Access:      23ms


  51 K    Reads  384000  Writes  204800  b/sec
 153 K    Reads  433694  Writes  209454  b/sec
 307 K    Reads  455111  Writes  213086  b/sec
 512 K    Reads  450109  Writes  212595  b/sec
 768 K    Reads  452874  Writes  212595  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     C Ltd.
Product:          SCSI/2000 (Old)
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Kalok
Model:            Octo-30
Capacity:         30 Meg.
Avg. Access:      65ms.


NOTES:   This drive is an ST-506 drive and was tested
         with an OMTI 3527 SCSI/ST-506 device controller.


  51 K    Reads  396387  Writes  198193  b/sec
 153 K    Reads  396387  Writes  207101  b/sec
 307 K    Reads  416542  Writes  216847  b/sec
 512 K    Reads  425190  Writes  214450  b/sec
 768 K    Reads  425681  Writes  216338  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-225N
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  372363  Writes  307200  b/sec
 153 K    Reads  341333  Writes  317793  b/sec
 307 K    Reads  341333  Writes  329142  b/sec
 512 K    Reads  341333  Writes  330322  b/sec
 768 K    Reads  333309  Writes  329142  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Rodime
Model:            RO-652
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  614400  Writes  585142  b/sec
 153 K    Reads  542117  Writes  472615  b/sec
 307 K    Reads  554345  Writes  508468  b/sec
 512 K    Reads  571534  Writes  536593  b/sec
 768 K    Reads  565398  Writes  542117  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Syquest
Model:            SQ-555
Capacity:         44 Meg.
Avg. Access:      28ms.


NOTES:  This is a removable media cartridge type Hard Drive.


  51 K    Reads  585142  Writes  534260  b/sec
 153 K    Reads  576000  Writes  512000  b/sec
 307 K    Reads  589824  Writes  554345  b/sec
 512 K    Reads  590769  Writes  556018  b/sec
 768 K    Reads  587006  Writes  574205  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-157N
Capacity:         48 Meg.
Avg. Access:      28ms.


  51 K    Reads  512000  Writes  614400  b/sec
 153 K    Reads  550208  Writes  658285  b/sec
 307 K    Reads  576000  Writes  614400  b/sec
 512 K    Reads  590769  Writes  636683  b/sec
 768 K    Reads  587006  Writes  626938  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-277N-1
Capacity:         65 Meg.
Avg. Access:      28ms.

  51 K    Reads  146285  Writes  261446  b/sec
 153 K    Reads  200347  Writes  542117  b/sec
 307 K    Reads  210051  Writes  609322  b/sec
 512 K    Reads  200784  Writes  579622  b/sec
 768 K    Reads  206868  Writes  574205  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Maxtor
Model:            XT-3170
Capacity:         135 Meg.
Avg. Access:      23ms.


Notes:  This drive caused the Amiga to lock-up, displaying a grey
        screen   every   time  the  computer  was  powered-up  or
        re-booted.   The  only  way the system was finally tested
        was  to  connect  a  second  Amiga computer with a C Ltd.
        controller to the system in a SCSInet fashion and turn on
        both   computers,   boot   the  Amiga  with  the  C  Ltd.
        controller,   boot   the  Hard  Drive  from  the  C  Ltd.
        controller  and  then  finally  boot  up the system under
        test.   There  is  no way that an average user could ever
        install and/or use this drive with the 2090A.


  51 K    Reads  614400  Writes  614400  b/sec
 153 K    Reads  614400  Writes  576000  b/sec
 307 K    Reads  594580  Writes  635586  b/sec
 512 K    Reads  614400  Writes  636683  b/sec
 768 K    Reads  606315  Writes  635586  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Commodore/Amiga
Product:          2090A
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Kalok
Model:            Octo-30
Capacity:         20 Meg.
Avg. Access:      65ms.


Notes:  This is an ST-506 drive controlled by the 2090A's
        ST-506 controller.


  51 K    Reads  256000  Writes  219428  b/sec
 153 K    Reads  256000  Writes  231849  b/sec
 307 K    Reads  253360  Writes  243326  b/sec
 512 K    Reads  256000  Writes  248242  b/sec
 768 K    Reads  253884  Writes  249081  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-225N
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  151703  Writes  240941  b/sec
 153 K    Reads  151703  Writes  237832  b/sec
 307 K    Reads  151392  Writes  236307  b/sec
 512 K    Reads  151330  Writes  238601  b/sec
 768 K    Reads  150958  Writes  238139  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Rodime
Model:            RO-652
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  153600  Writes  245760  b/sec
 153 K    Reads  151081  Writes  237832  b/sec
 307 K    Reads  152330  Writes  239376  b/sec
 512 K    Reads  151143  Writes  237678  b/sec
 768 K    Reads  151081  Writes  238139  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Syquest
Model:            SQ-555
Capacity:         44 Meg.
Avg. Access:      28ms.


NOTES:  This is a removable media cartridge type Hard Drive.


  51 K    Reads  332108  Writes  236307  b/sec
 153 K    Reads  357902  Writes  233316  b/sec
 307 K    Reads  366805  Writes  335127  b/sec
 512 K    Reads  367904  Writes  329436  b/sec
 768 K    Reads  369378  Writes  338823  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-157N
Capacity:         48 Meg.
Avg. Access:      28ms.


  51 K    Reads  512000  Writes  614400  b/sec
 153 K    Reads  550208  Writes  658285  b/sec
 307 K    Reads  576000  Writes  614400  b/sec
 512 K    Reads  590769  Writes  636683  b/sec
 768 K    Reads  587006  Writes  626938  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-277N-1
Capacity:         65 Meg.
Avg. Access:      28ms.


NOTES:   It  is  obvious  from these test results that interleave
         value  for  this  drive is inappropriate.  GVP, however,
         provides  no means to alter the default interleave value
         so  these  results are a true measure of the performance
         that  an  average user would experience from this drive.
         Too bad!


  51 K    Reads  180705  Writes  204800  b/sec
 153 K    Reads  191005  Writes  196085  b/sec
 307 K    Reads  192501  Writes  194021  b/sec
 512 K    Reads  193207  Writes  194430  b/sec
 768 K    Reads  193207  Writes  193410  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     Great Valley Products
Product:          GVP Impact-2000
Controller Type:  Pseudo DMA/Processor Controlled

Hard Drive Tested:

Manufacturer:     Maxtor
Model:            XT-3170
Capacity:         135 Meg.
Avg. Access:      23ms.


Notes:   This  drive  was never recognized by the system whenever
         the  computer was powered-up or re-booted.  The only way
         the  system  was  finally tested was to connect a second
         Amiga  computer  with a C Ltd.  controller to the system
         in  a  SCSInet  fashion and turn on both computers, boot
         the  Amiga  with  the  C Ltd.  controller, boot the Hard
         Drive  from the C Ltd.  controller and then finally boot
         up  the  system  under  test.   There  is no way that an
         average  user  could  ever install and/or use this drive
         with the GVP controller.


  51 K    Reads  323368  Writes  332108  b/sec
 153 K    Reads  347773  Writes  354461  b/sec
 307 K    Reads  354461  Writes  361411  b/sec
 512 K    Reads  358250  Writes  361411  b/sec
 768 K    Reads  359298  Writes  362121  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-225N
Capacity:         20 Meg.
Avg. Access:      65ms.


  51 K    Reads  384000  Writes  384000  b/sec
 153 K    Reads  347773  Writes  351085  b/sec
 307 K    Reads  354461  Writes  354461  b/sec
 512 K    Reads  353103  Writes  353103  b/sec
 768 K    Reads  346466  Writes  346466  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Rodime
Model:            RO-652
Capacity:         20 Meg.
Avg. Access:      65ms.


NOTES:  This hard drive repeatedly failed to be recognized by the
        Microbotics  HardFrame and required several attempts each
        time  the system was powered up before the drive could be
        used.


  51 K    Reads  614400  Writes  534260  b/sec
 153 K    Reads  594580  Writes  576000  b/sec
 307 K    Reads  599414  Writes  599414  b/sec
 512 K    Reads  611343  Writes  611343  b/sec
 768 K    Reads  610331  Writes  612358  b/sec
1024 K    Reads  612867  Writes  612867  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Syquest
Model:            SQ-555
Capacity:         44 Meg.
Avg. Access:      28ms.


NOTES:  This is a removable media cartridge type Hard Drive.


  51 K    Reads  558545  Writes  215578  b/sec
 153 K    Reads  594580  Writes  409600  b/sec
 307 K    Reads  614400  Writes  475664  b/sec
 512 K    Reads  605320  Writes  516302  b/sec
 768 K    Reads  614400  Writes  543716  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-157N
Capacity:         48 Meg.
Avg. Access:      28ms.


  51 K    Reads  614400  Writes  614400  b/sec
 153 K    Reads  658285  Writes  658285  b/sec
 307 K    Reads  676403  Writes  676403  b/sec
 512 K    Reads  667826  Writes  667826  b/sec
 768 K    Reads  675164  Writes  675164  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Seagate
Model:            ST-277N-1
Capacity:         65 Meg.
Avg. Access:      28ms.


  51 K    Reads  768000  Writes  768000  b/sec
 153 K    Reads  768000  Writes  801391  b/sec
 307 K    Reads  837818  Writes  837818  b/sec
 512 K    Reads  824697  Writes  819200  b/sec
 768 K    Reads  797922  Writes  797922  b/sec


*******************************************

SCSI Controller Tested:

Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Maxtor
Model:            XT-3170
Capacity:         135 Meg.
Avg. Access:      23ms.


Notes:   This  drive  caused  a  GURU every time the computer was
         powered-up  or  re-booted.   The only way the system was
         finally  tested  was  to connect a second Amiga computer
         with  a  C  Ltd.   controller to the system in a SCSInet
         fashion  and turn on both computers, boot the Amiga with
         the  C Ltd.  controller, boot  the Hard Drive  from  the
         C Ltd.   controller and then finally  boot up the system
         under  test.  There is no way that an average user could
         ever install and/or use this drive with the HardFrame.


  51 K    Reads  614400  Writes  614400  b/sec
 153 K    Reads  658285  Writes  658285  b/sec
 307 K    Reads  682666  Writes  689046  b/sec
 512 K    Reads  694237  Writes  690337  b/sec
 768 K    Reads  687761  Writes  687761  b/sec


********  END  OF  TEST  DATA  *********





****************************************************************
*                                                              *
*          Conducting Your Own Tests Using Devspeed            *
*                                                              *
****************************************************************



    Included  with  this  file  are  two  programs.  The Devspeed program which
can be used to test the read and write speed of any device,  and  the  TCB
program  which lists all devices that are currently  active  on your computer.
To test your system, follow the procedure outlined below.


1.  Execute the TCB program by typing:

          TCB  d     <RETURN>

2.  Review the list of devices and  locate the name of the device you want to
    test.  Make note  of  the  EXACT  spelling of the device name and case of
    the letters used.  (Device names  ARE case sensitive.)

3.  Look  in  the  mountlist  in the Devs:  directory,  or in the documentation
    provided  by  the manufacturer for the UNIT number and  the FLAGS value
    associated with the hardware device that you want  to  test.  Devices
    mounted using the AmigaDOS Mount command get  the information  from  the
    devs:mountlist and will have an entries  in  the  mountlist  called  UNIT
    and FLAGS.  Information about  devices  mounted  with  the  BindDrivers
    command should be located in the expansion drawer.

4.  Having located and noted the required information, execute the Devspeed
    program by typing:

          Devspeed  [device]  [unit]  [flags]   <RETURN>

     [device] is the EXACT name of the DEVICE to be tested.
     [unit] is the UNIT number to be tested.
     [flags] is the flags value specified for use with the UNIT.


NOTES:  

1.   Devspeed  loads  data  into  RAM and will use up to 1 Meg of memory if it
     is available.

2.   The  program  may  crash  if  supplied illegal UNIT or FLAGS values.

3.   REMEMBER device names ARE case sensitive.


    

    John,
    
    Could you post the devspeed program or do you think it is copyrighted?
    
    thanks,
    
    steve

    DevSpeed pgm and source are available on the CLtd BBS for general
    use.  I'll upload the program, docs and source to 
    FSDEV1::Disk$User1:[JBERNARD.AMIGA]DevSpeed.*
    
    By the way, I have both the old and new controller up on the system
    and, although I didn't have much time to play with the DevSpeed
    program, the figures posted by CLtd are very close to what I get.
    
    I'll post my results also.  Most of my drives are ST506 with an
    Omti 3527 (RLL) SCSI adaptor.  I do have an ST225N imbeded scsi.
    
    John
    

    DEVSPEED does not appear to work. Might be better is you ARC it
    first and then upload it. I tried FIXOBJ and that didn't fix it
    either.
    
    -Paul

    DEVSPEED.ARC is now available in  FSDEV1::DISK$USER1:[JBERNARD.AMIGA]
    Docs and source included.
    

    I attempted to run it, but it caused my 2000 to hang.

    I ran devspeed last night with the following configuration:

    A2000 + A2058 + Microbotics Hardframe + Quantum Prodrive 40S (40MB)
    
    Lots of utility type of programs were running in the background.
    The drive was about 35% full but I don't think this matters for
    this test.
 
    The results are attached below.  Any idea why the results varied with
    block sizes so much for my configuration?  The ones posted in .1
    don't look like this at all.  While it seems a bit slow for smaller
    block sizes, it is incredibly fast for the larger sizes.  It seems to
    me that I should be able to improve the performance at the low end.
    Any suggestions as to what should be changed to do this?  I've believe
    I've got addbuffers dh0: set to 35.
    
    thanks,
    
    steve
    
--------------------------------------------------------------------------------
SCSI Controller Tested:
    
Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Quantum
Model:            ProDrive 40S
Capacity:         40 MB
Avg. Access:      19ms


   6 K    Reads    13771  Writes    13771  b/sec
  15 K    Reads    31779  Writes    31779  b/sec
  51 K    Reads   105931  Writes   105931  b/sec
 153 K    Reads   315076  Writes   317793  b/sec
 307 K    Reads   635586  Writes   630153  b/sec
 512 K    Reads  1059310  Writes  1059310  b/sec
 768 K    Reads  1575384  Writes  1588965  b/sec
1024 K    Reads  2118620  Writes  2118620  b/sec
1228 K    Reads  2520615  Writes  2542344  b/sec
1433 K    Reads  2966068  Writes  2966068  b/sec

    You might try getting rid of the background tasks and not using
    addbuffers during the test.  This may affect the lower end.  As
    for the large block transfers, they look fine!!!  I've never
    seen such a transfer rate.  It does make one suspect the program
    though...
    Do you have an 020 or 030 on the system?  This does make a difference
    but should not impact the small block transfers negatively.
    
    Running DiskPerf on the KRONOS showed about a 10% increase over
    the standard controller, running DevSpeed showed a 60% increase.
    Maybe there is some sleight of hand (program) going on here from
    CLtd.
    
    I posted the program and results so those that may be looking at
    the KRONOS as an upgrade might have a bit more information rather
    than the usual advertisement hype.  You do get some performance
    but not the  "10 times faster" than the old controller that was
    originally promised.  It does have auto-boot capability and does
    perform well.  The software allows you to use just about any drive
    on the market, supports a color scanner, and allows one to tie 2
    or more systems together, sharing resources.  These are the 
    largest advantages for me personally.  If you don't need these
    capabilities, I'd go for the HardFrame, GVP, or Overdrive, in that
    order.
    
    John
    

I'm having  a  problem with DEVSPEED. I'm a new Amiga 2000 owner. I recently
bought  a  Hardframe  2000 SCSI controller and a Seagate ST296N (80meg) hard
disk.  I  would  like  to measure the performance of this set-up, expecially
since  reading  a  note  a  few  back (I forget the number right now), about
Seagate drives being intentionally slowed down to let MACs keep up.

The problem is getting the info that DEVSPEED wants (UNIT, FLAGS, etc.). The
controller  automounts  the partitions and saves this info in the rigid disk
blocks on the HD, thus no entry in MOUNTLIST.  The DEVSPEED doc says that in
this case, the info should be in the expansion drawer.  The only thing I can
find  there  are  binaries.  Is it in one of these, and how can I get at it?
Sorry if this is trivial, I haven't gotten into the nuts and bolts yet.

Anyway, I  tried to run it last night with a device name ranging from DH0 to
DH7 (with and without colons), a unit number of 0, and a flag of either 0 or
1  or  nothing  specified. When I did this I get an error message similar to
"can't open .device".  Any ideas???

BTW, I  boot  minimally off of a floppy which transfers control to the HD to
continue  booting  and  to  run  off of (until I get time to install the 1.3
ROM).

Thanks for any help,
Gail

    DEVSPEED doesn't want a device name as the 1st parameter, it wants the
    name of the device driver! For example if you use the standard device
    driver supplied by CBM you would type something like:
    >DEVSPEED hddisk.device 1 0 - this tests Unit #1
    Hope this helps.
    
     mark

    Here is DEVSPEED for a A2000 + 2M (MEGAboard 2000) + 2090A + RD51(2)
    
    RD51 (HD0) - 10M partition (FFS)
    ==========================
    
   6 K    Reads  228205  Writes  199680  b/sec
  15 K    Reads  245760  Writes  230400  b/sec
  51 K    Reads  250775  Writes  198193  b/sec
 153 K    Reads  256000  Writes  242526  b/sec
 307 K    Reads  249081  Writes  242526  b/sec
 512 K    Reads  251803  Writes  247741  b/sec
 768 K    Reads  249756  Writes  244132  b/sec
1024 K    Reads  251031  Writes  247492  b/sec
1228 K    Reads  249502  Writes  246994  b/sec
1433 K    Reads  250775  Writes  247706  b/sec
    
    RD51 (HD1) - 10M partition (FFS)
    ==========================
    
   6 K    Reads  199680  Writes  199680  b/sec
  15 K    Reads  230400  Writes  230400  b/sec
  51 K    Reads  236307  Writes  219428  b/sec
 153 K    Reads  254234  Writes  244132  b/sec
 307 K    Reads  233316  Writes  227555  b/sec
 512 K    Reads  240469  Writes  236307  b/sec
 768 K    Reads  240626  Writes  237525  b/sec
1024 K    Reads  242606  Writes  239298  b/sec
1228 K    Reads  240548  Writes  239376  b/sec
1433 K    Reads  243498  Writes  241110  b/sec
     
     mark

    
    The command you probably need is:
    
    DEVSPEED HardFrame.device 0 0
    (note that the device name is case sensitive.)
    
    This is what I used and aside from weird results it seemed to work
    correctly.
    
    I booted with a totally empty system.  No background processes at 
    all.  I raised the priority of the process I was running in also.
    I had 0 addbuffers.  I got the same results as posted previously
    to within a few percent difference (1-2%).  I still have no idea
    why my performance improves so dramatically with block size.
    Seems to get better and better. If I had enough memory I could 
    probably read the whole 40MB in no time at all :-) :-).
    
    regards,
    
    steve

    I found my problem.  The command I gave in the last message was not 
    correct for the HardFrame the way I have it set up.  The default SCSI
    unit number is 6 not 0.  Devspeed uses DoIO and doesn't check the
    result code.  Therefore the command should be:

    DEVSPEED HardFrame.device 6 0

    This gave the numbers given below.  Still pretty fast for large blocks.
    The 64K byte cache on the Quantum is probably why the small block reads 
    are so fast.
    
    regards,
    
    steve
--------------------------------------------------------------------------------
SCSI Controller Tested:
    
Manufacturer:     MicroBotics
Product:          HardFrame
Controller Type:  DMA

Hard Drive Tested:

Manufacturer:     Quantum
Model:            ProDrive 40S
Capacity:         40 MB
Avg. Access:      19ms

       6 K    Reads 1597440  Writes  532480  b/sec
      15 K    Reads  921600  Writes  921600  b/sec
      51 K    Reads  768000  Writes  768000  b/sec
     153 K    Reads  837818  Writes  801391  b/sec
     307 K    Reads 1504653  Writes  837818  b/sec
     512 K    Reads 1293473  Writes  847448  b/sec
     768 K    Reads 2118620  Writes  841643  b/sec
    1024 K    Reads 3276800  Writes  844536  b/sec
    1228 K    Reads 3780923  Writes  842605  b/sec
    1433 K    Reads 3308307  Writes  849540  b/sec
    

    For all you Kronos owners, good news. I received the following in the
    mail yesterday. Reprinted without permission.
    ___________________________________________________________________________
    Flyer #1
    				Press Release
    
    Micro-Dyn Inc. has obtained the rights to all CLTD Inc. products. We
    are in production on all KRONOS and SCSI harddisk controllers for the
    A-1000, A-500, and A-2000 computers. We are offering service and
    upgrades on CLTD Inc. products. Some of the upgrades available are;
    KRONOS-TAPE DRIVER, AMAX-II DRIVER, and SCSIDOS 3.1.
    
    Micro-Dyn Inc.
    2011 S. Washington
    Wichita, KS 67211
    Phone 316-265-2661
    FAX   316-265-4130
    ____________________________________________________________________________
    
    Flyer #2
    
    Dear KRONOS card user
    
    	Mycro-Dyn Inc. has finally finished the AMAX-II Driver for all
    KRONOS controller cards (A1000,A2000,A500).
    	This upgrade is completely compatible with existing harddrive
    formatting. But does require a PAL change on the card as well as
    software driver installation. There is also supplied with is new
    utilities for the SCSI KRONOS controller. The purchase of the upgrade
    will entitle you to KRONOS TAPE driver software free and SCSI direct
    command when available (within 60 days).
    
    	The package price is $50.00
    
    Prepaid orders are shipped free, other UPS.
    
    Micro-Dyn Inc.
    P.O. Box 192
    Derby KS 67037
    Phone 316-265-2661
    FAX 316-265-4130
    ___________________________________________________________________________
    
    Flyer #3
    
    			KRONOS TAPE PRICE LIST
    
    KRONOS SCSI Tape drive internal (A2000)			$400.00
    
    KRONOS SCSI Tape drive Ext w/power supply and case		$500.00
    		(A1000,A500)
    
    60 Meg Tape Cartridge					$ 25.00
    
    QIC-40 Tape Backup						$425.00
    ____________________________________________________________________________
    
    Flyer #4
    
    				Price List
    
    KRONOS 2000							$150.00
    
    KRONOS 1000							$180.00
    
    KRONOS 500							$180.00
    
    SCSI 1000							$ 60.00
    
    SCSI 2000							$ 60.00
    
    SCSI 500							$ 60.00
    
    ____________________________________________________________________________
    
    Flyer #4
    
    				KRONOS CONTROLLER
    
    FEATURES
    	
    		o 16 bit operation
    
    		o static dual port ram caching on board
    
    		o FASTEST Amiga hard drive controller (950K)
    
    		o auto booting
    
    		o work with AMAX II
    
    		o available for all amiga computers
    
    		o works with CSA, CBM, GVP accelerators
    
    		o single slot used
    
    		o hd mounting frame available (3.5")
    
    		o SCSI net and drive daisy chain
    
    ___________________________________________________________________________
    
	the above is taken directly off the flyers so don't blame me for
    the confusing grammer. I am also not affiliated with Micro-Dyn, BUT
    just happen to have a KRONOS 2000 collecting dust in my file cabinet,
    anybody interested? :-)
    
    Roger
    

    re:.14
    
    Thanks for posting the good news.  I still use my Cltd SCSI controller,
    and I recently got a cheap Kronos.  Nice to see that they think there
    is still a market for them.
    
    I wonder if "auto-booting" means they finally implemented Rigid Disk
    Blocks instead of their own method.  It would be nice to get a scsi direct
    driver, I'll believe it when I see it ship in 60 days.
    
    Does this make them the first Amiga SCSI vendor to sell controllers
    for less that $100?  
    
    Dave