nelson_p@apollo.uucp (05/31/88)
I have an application that's very floating point-intensive.
It is written in (M.S. QuickC) C and run on my Leading
Edge Model 'D' XT clone. Since I don't have an 8087,
the code uses the M.S. floating point emulator code.
(BTW I have V1.01 which fixes the famous precision error
in V1.00.)
My question is this: How does the speed of the software
floating point code compare with doing it with an 8087
running at the same clock speed as the CPU? I am especially
interested in single-precision multiplies and divides.
I need to decide if an 8087-2 will give me enough of a boost
to justify its ~ $150 pricetag. Than you in advance.
--Peter Nelson