SRFERGU@ERENJ.BITNET (Scott Ferguson) (02/21/90)
I'm using some rather large (up to 150 MBytes or so) files, which I am forced to do random access on. I'm using C, because I like structures and the style of the language. Unfortunately, if I need to do an fseek() on the file, the further the seek goes, the longer it takes. I had hoped that it could do something like increment a file pointer, but it seems to be marching through, byte-by-byte, or block-by-block. I can't use ms_$ calls, because I'm going over NFS to a non-apollo system with this stuff, which makes seek/read take that much longer (quick gripe about getting out those 100 Mbit/sec networks...) I haven't tried using the open/read/lseek set of I/O statements, just fopen/fread/fseek. Would it be any faster to switch to open/read/lseek? I haven't tried using FORTRAN direct access files, because I think my Titan FORTRAN environment might use a funky direct-access system, which is different from my sr9.7 'rec' type files. Is there a faster way to seek through a file than fseek, without using mapped segment stuff? Thanks, Scott Ferguson srfergu@erenj.bitnet