Thruput of Sun/Mercury Combo

[fnddr at acad3.fai.alaska.edu]

We are looking at a setup which essentially takes a 16-bit sample stream
and performs ffts on it.  One proposed solution would have a Sun-4/330
with a Mercury array processor and adc on its VMEbus.  The Mercury specs
claim it can do a 1024-point FFT in 2.7ms.  However, some Mercury
competitors say that, when the bus i/o overhead and the SunOS i/o overhead
are figured in, the average time required to do the FFT on a data stream
is much much longer.  In particular Sky is pushing their RISC+AP board as
solution to this problem.

So, if anybody is running such a system, I'd like to hear about the actual
performance levels.  Any comments on the reliability of the APs (MTBF)
would also be welcome.

Thanks,
Don Rice                                  Internet: fnddr at acad3.fai.alaska.edu
Geophysical Institute                     E-mail:   fnddr at alaska.bitnet
University of Alaska                      Phone:    (907) 474-7569
Fairbanks, AK 99775                       Loran:    64.86N 212.16E