AmigaWorld Errors
MICROBOTICS THE VICTIM OF FAULTY REPORTING IN AMIGAWORLD
In the recently published issue of AmigaWorld (September 1993) there is a
"roundup" style article entitled "A1200 EXPANSION: CHOOSING THE RIGHT PATH".
The article starts on page 21. On page 24 there is a boxed table giving the
aggregate results of AIBB 5.5 benchmark testing on various A1200 expansion
boards and accelerators. Unfortunately for MicroBotics, the results reported
for its new M1230XA accelerator are seriously in error and make the MicroBotics
product look like a poor performer in comparison to other '030 based
accelerators. This is a direct result of not running the AIBB test using XA's
32-bit memory. The author claims the test would not run when XA's memory was
configured, so he and AmigaWorld chose to report XA's results using only slow,
PCMCIA-based ram. Neither the author nor AmigaWorld contacted MicroBotics
before publishing this extremely damaging, misleading report. Since we actually
*use* AIBB to test XA's before shipping and actually included AIBB 5.5 on the
installation diskette we sent AmigaWorld, it is especially puzzling and
disturbing that they failed to at least ask for some explanation from
MicroBotics before damning us in print. We feel that AmigaWorld has caused us
serious harm as a direct result of this weak reporting and editorial work.
Moreover, they have greatly misled and misinformed their readers. This is an
extremely unpleasant and regrettable incident.
In investigating why the author couldn't run AIBB 5.5 on XA, we found that the
M1230XA returned from AmigaWorld was returned *wrongly configured* for *eight
megabytes* of memory installed when the unit had actually only 4 megabytes
installed. Naturally, this would cause almost any program that used FastRAM to
crash. When the board was originally sent to AmigaWorld, it was *correctly*
configured to 4 megabytes (indeed the last step before shipping was to run AIBB
5.5 on the board). So it left Texas correctly set up and came back wrongly
configured. We can only conclude that the author (or someone at AmigaWorld)
made improper use of XA's setup software to tell the system it had more memory
than it actually did and thus caused the failure himself. Further supporting
this surmise is the statement in the article that "each system was configured
with…8MB of 32-bit RAM on board."
We are demanding that AmigaWorld publish a full and prominent apology and full
correction to their mistake in the next available issue. Unfortunately, no
correction can really undo all the damage AmigaWorld has done to MicroBotics
and the XA already. Nevertheless, we publicly call for maximum redress to the
wrong inflicted upon us.
In the table below, we show the aggregate AIBB numbers as reported by
AmigaWorld compared to the *real* results that are obtained when running a
properly configured M1230 XA. The third column is the percentage of error
introduced by AmigaWorld's misreporting. As one can see by the gross average
accuracy summary of only 42 percent, the reader would have been better off
flipping a coin than relying upon the numbers in AmigaWorld. We plead for the
understanding of prospective XA customers with regard to this unfortunate
situation. Accurate AIBB load modules and benchmark information for the M1230
XA are available on the CompuServe and BIX networks as well as directly from
MicroBotics.
AMIGAWORLD ACCURACY RATING CONCERNING
THEIR SEPTEMBER 93 "REPORT" ON
MICROBOTICS M1230 XA ACCELERATOR
AIBB 5.5 INTEGER RATING
AmigaWorld Accurate PERCENTAGE ACCURACY
Erroneous Test of AmigaWorld Report
Report Report
A500 3.93 12.04 32.64%
A2000 3.83 11.73 32.65%
A3000 0.63 1.94 32.47%
A4000 0.31 0.96 32.29%
AIBB 5.5 GRAPHICS RATING
AmigaWorld Accurate
Erroneous Test
Report Report
A500 4.54 5.76 78.82%
A2000 3.37 4.27 78.92%
A3000 1.91 2.42 78.93%
A4000 0.82 1.04 78.85%
AIBB 5.5 FLOATING POINT RATING
AmigaWorld Accurate
Erroneous Test
Report Report
A500 4.52 27.34 16.53%
A2000 3.05 18.46 16.52%
A3000 0.32 1.94 16.49%
A4000 0.11 0.65 16.92%
AGGREGATE PERCENTAGE ACCURACY OF AMIGAWORLD 42.67%