Posted by Robert G Laurin on December 04, 2010 at 08:52:16:
In Reply to: Re: ORDINAL READ posted by Barbara Brazil on December 03, 2010 at 08:03:01:
Yeah a smaller Ordinal number worked. 4GB used to be the limit of the file size. Since the ordinal is multiplied by the record size, that should give us a decent file size but probably a far cry from the new "extended files" size.
So does the 4GB Ordinal key number work properly for files larger than 4GB ?
I used to use an 8.0 for ORDINAL years ago, but now with files reaching 100 million records... That's why I gave an arbitrary 16.0 to Ordinal when I wrote my test program in Comet 32.
Another surprise...
Used to be that ORDINAL READ was much faster than Indexed read. No true no more... I wrote a test program that did 10,000 KEY() & INQUIRE(), then 10,000 sequential read (no KEY=) and then read 10,000 using ordinal keys. Had this test time itself and repeat 4 times (keep moving forward in the same file, not reading same key twice to avoid caching). I was surprise to see that in all cases performance was very close, varying slightly up & down in all method. They all ranged in the 5.0 to 6.5 seconds for each 10,000 iteration.
So I decided not to try to re-write this massive program to try to make it faster by using the ORDINAL. Not worth the effort and the pain.
I did however appreciate the much larger variables in Comet 32, allowed me to optimize a program that now takes under 3 days instead of the planned 8+ days. Thank goodness we don't run this everyday.
Each file can be a maximum of 1MB in length Uploaded files will be purged from the server on a regular basis.