Contents | < Browse | Browse >


/// Microbotics 1230XA -- 68030 Accelerator for the Amiga 1200
    ----------------------------------------------------------
    By Tom Mulcahy


The 2nd true accelerator for the A1200 is upon us.  The default con-
figuration of an '030/50 MHz makes it the fastest accelerator currently
available for the A1200.  Other versions include 33 MHz and EC030 40 MHz
chips.  All CPUs are socketed for upgradablitiy.  There is  also an FPU
socket for a  68881/68882 floating point unit which can be clocked in-
dependently of the CPU.  If the board is turned over one will find one SIMM
socket.  This one SIMM socket is capable of supporting up to a 128 MB SIMM
giving it the largest RAM capacity of any of the other FPU/RAM boards and
accelerators.  Next to the SIMM socket is a battery for the on board clock.

There are two jumpers on the board.  The first jumper is a 3 pin jumper and
is moved to pins 2 & 3 if the cpu and fpu are the same speed and on pins
1 & 2 if the cpu and fpu are clocked independently.  No mention is made of
what to do if NO FPU is installed.  I was told by Microbotics that if there
is no FPU installed the jumper configuration has no effect.  The second
jumper is what is known as the TEST jumper.  This jumper should be left
OFF.  If it is in the ON position the configuration EEPROM (mentioned
later) will be bypassed on boot and the XA's RAM setting is set to the
slowest cycle for 128 MB.  The installation sheet states the jumper should
ON only if the EEPROM is inadvertently in a 'confused' state.  I have yet
to find out exactly what Microbotics meant by 'confused'.  Never the less
the jumper should be left off or the board will seem to be acting confused.

Back to the topic of SIMMs...  Microbotics states this board will accept
any industry standard 32-bit wide 72-pin SIMM. I and another user on BIX
have reported incompatibility problems with both a 16 MB and 8 MB SIMM.
These were standard 32-bit wide 72-pin SIMMs as well.  When logs of the
error  tests were reported to Microbotics they felt it was a bad SIMM, but
after we both tested the SIMM on other machines (I tested it on a Mac
Centris 610) -- I don't recall what machine the other user tested his SIMM
on but it was in fact in working condition -- it was safe to assume the
board appears finicky about what brand SIMMs are occupying the socket.
I'm not trying to scare anyone into thinking this board doesn't work with
8 or 16 MB SIMMs, because my local dealer has an 8 MB SIMM in his 1230XA.
If you are able to either ask to try the SIMM in the board before purchase
or order one of  the recommended SIMMS listed in the installation sheet
that comes with the board.   

The board does not autoconfig.  According to the docs the board will 
autoconfig under the upcoming WB 3.1.  Two programs, SetXA and AddMem,
allow the board to be configured.  SetXA, which writes to the onboard
EEprom, allows you to set the amount of RAM installed, speed of CPU and
FPU and type of CPU (EC030 or 030).  AddMem should be placed in the
WBstartup drawer. Upon booting the board will be configured.  Also
included on the installation disk is a Mem test program and a copy of
AIBB.  

How fast is this board?  To get a true indication of it's speed try to
obtain the appropriate AIBB module as it will give you the most accurate
benchmarks.  Using Sysinfo v3.14 though, I obtained the following:

Dhrystones 7435

Machine | CPU | SPEED | x 1230XA/50/50/w4MB faster than
-------  ----- ------- ---------------------------------
A600     68000   7mhz   12.66
B2000    68000   7mhz   10.63
A1200    EC020  14mhz    5.83
A2500    68020  14mhz    3.61
A3000    68030  25mhz    1.60
A4000    68040  25mhz     .40

CPU Mips - 7.76
FPU MFlops - 1.33
Chip Ram speed vs A600 - 5.58


Besides the potential SIMM incompatibilities, the board has a couple areas
that could have been improved upon.  The addition of one other SIMM slot
would have been nice.  Although the one socket can already accept up to
a 128 MB SIMM this can make the SIMM purchase decision tricky.  Do you
spend a bundle all at once and get a 16 MB or even a 32 MB SIMM?  Or do you
settle for 4 MB or 8 MB the first time out?  And once the decision is made
to upgrade that particular SIMM must be sold off first.  The GVP 1230
Turbo has 2 SIMM sockets and so does the DKB RAM/FPU board.  Keeping the
price of the board down may have had a part to play in this design 
decision.  The board also doesn't have a jumper to disable the board.
Not a major problem but would have helped somewhat with compatibility.
I've run into a couple even newer games that won't run with the 1230XA,
such has Pinball Fantasies, one of my favorites.  

[Editor's note:  At last check, a 128 meg SIMM costs about $12,000.00.)

What I won't complain about is the performance of the board.  It is fast,
no doubt, and provides a massive increase of CPU and floating point
performance over a stock A1200.  Along with the battery backed clock and
large RAM capacity the board presents a good value.  With upcoming 
accelerators from CSA and ICD, Microbotics will face much stiffer
competition. 

** REVIEW ADDENDUM **

This review was intended to be published in the 1.14 edition of Amiga
Report but a certain discovery has prompted me to add the following.  The
Sysinfo benchmarks given were indeed for the '030/50/50 version of the
Microbotics board.  While playing with the settings this past week in an
attempt to get some games to run, I found out that the board can be sped
up by nearly 1 MIPS, pushing almost 9.  The settings for the SetXA program
with a 50 MHz CPU should be set up for a 50 MHz CPU, naturally, and 4 RAM
 cycles.  The RAM cycles are automatically defaulted to 4 whenever you
select 50 MHz.  If you set the speed to 33 MHz the RAM cycles get
defaulted to 3.  I obtained the following results from SysInfo 3.14 with
the 33 MHz/3 cycles setting:

Dhrystones 8368

Machine | CPU | SPEED | x 1230XA/50/50/w4MB faster than
-------  ----- ------- ---------------------------------
A600     68000   7mhz   14.25
B2000    68000   7mhz   11.97
A1200    EC020  14mhz    6.56
A2500    68020  14mhz    4.07
A3000    68030  25mhz    1.80
A4000    68040  25mhz     .45

CPU Mips - 8.73
FPU MFlops - 1.33
Chip Ram speed vs A600 - 5.58


I reported this to Microbotics, and the settings that the SetXA program
default to are conservative measurements.  They don't recommend changing
them and if you do you do so at your own risk.  I also ran the RAM test
program that comes with the board, MBRtest-2, all night long and it
reported zero errors.  Now I personally don't recommend you change the
default settings either nor will I or Amiga Report take responsibility if
you do and blow something up.  I  just wanted to report this phenomenon.
This is not a quirk in the Sysinfo program.  AIBB reports speed impovements
in several areas.  I rendered a scene of Ohau in hi-res overscan with
Scenery Animator that took exactly 2:48 to render with the board set to 50
MHz and 4 RAM cycles. With the board set to 33 MHz and 3 RAM cycles the
same scene took 2:35.  If you set the board to 50 MHz and just change the
RAM cycles the board will appear to speed up to 8.70+ MIPS but then after
a few minutes of use it always goes back down to the 'normal' 7.70 MIPS, so
the RAM cycles aren't solely responsible.  As one would expect if the board
is set to 40 MHz, the benchmarks fall right in between the 33 MHz setting
and the 50 MHz setting... a little faster than the 50 MHz setting and a
little slower than the 33 MHz setting as long as the RAM cycles are set to
3 and not 4.  Of course, being the speed freak that I am I decided to push
the board farther by setting it to 25 MHz.  No go.  The 1200 will simply
refuse to boot.  If the RAM cycles are set to anything less than 3, the
1200 will tend to lock up.  The 'safety zone' seems to be between 33 MHz/3
cycles and 50 MHz/4 cycles (the recommended setting!).

Microbotics, Inc.
1251 American Parkway
Richardson, Texas 75081
(214)437-5330  Central Time Zone, 9:00 am to 5:00 pm