| Re: .0
I use my expansion ram as a disk during C development--I copy the compiler,
linker, libraries, and include files all to RAM:. I keep the files that
make up my program on disk.
This does speed things up quite a bit. (Nope, I don't have numbers. I'll
try some benchmarks tonight.)
I very strongly believe that it is better to get more memory rather than
more disks. More memory makes multitasking really come alive.
Don't hold your breath for Turbo C. One of Borland's many sins of late
is "announcing" products that they never write. I believe that Turbo C
was never written for any machine.
On the other hand, Turbo Pascal for the Amiga might come about. Originally,
Borland was going to release a version of Turbo Pascal for the Amiga and
took out full page ads in AmigaWorld to that effect. A few months go by,
and Borland starts sending out letters saying that they had reconsidered,
and were no longer going to do an Amiga version. Lately, Borland has been
claiming that they are (again) going to do an Amiga version. (They just
released the Mac version of Turbo Pascal, so now, at least, they do have
a 68000 code generator.) Who knows what they are going to do?
|
| I did a test compile of a reasonably good sized file. I compiled it
twice: the first time was with Lattice C, all the supplied include files,
all my include files, and the source program all in ram. The second time
was with all the proceeding on disk.
The times were:
All ram 57 seconds
All disk 139 seconds
Or the disk compile was 2.45 times longer.
I didn't time any link-times. They should be speeded up by a large factor
as well. Linking is a two pass process: all the object files are read
twice.
|