FB II Compiler
Consider high precision numbers
I need very high precision for some mathematical calculations I'm trying to do, and FB is bombing out. I need two high-precision vectors, with eventual dimensions of 10,000 elements each, and these elements should be about 800 floating point digits.
In working my way up to this goal, I have been encountering problems. First, I cannot seem to DIM double-precision arrays; for example, DIM f#(500) fails, but DIM f!(500) succeeds. Also, while the program runs with DIM f!(500), g!(500) it fails if the only change is to make it DIM f!(1002), g!(1002) (the failure is "The program has quit with a Type 3 error")
Currently I have my preferences set at 200 double-precision digits, 150 single-precision digits, but as Imentioned I eventually want to set these numbers at about 800 (did you know you can accomplish this by simply typing into the preferences dialog the number 800 followed by the TM symbol, option-2 ?).
Any thoughts are appreciated. Perhaps I should call MOREMASTERS? My FB allocation is currently 4M, I can easily increase it to 25 M if necessary.
Sounds like you need to increase your memory partition. A 200 digit precision float is pretty big. Multiply that by 500 or 1002 or more and you'll run out of memory quick.