PureBasic Forums - English

Yes, Filemapping is the way to go, see here:
http://www.purebasic.fr/german/viewtopi ... 94#p152894

Thanks very much, I didn't know about that

I ran the ts-soft test code (no debugger) on a 3.1g file, it couldn't open. PB opens it ok. I tried on a 183mb file and pb was around 10% slower. I was expecting a bigger difference than that. (PB 4.60b3)

Hm... I haven't tested it on very large files yet

I'm working on a SHA 224-256-384-512 FileFingerprint implementation and I discovered that my posted code worked fine for files with size <= #MAXLONG but failed for larger sizes. So I worked with the code until I had a version that seemed to work correctly. (unposted as yet) I dowloaded a free file hash generator, hashslash, and tested it against my code. No differences. So I was happy until I discovered that hashslash was processing large files in about 1/3 of the time my code took. Ouch. I know my code is efficient (it completes in around 60% of the time the posted version takes) so I looked to the filemapping code by ts-soft for a possible solution to the speed difference and so far I only see an improvement of around 10% for filesizes <= #MAXLONG and no success opening files larger than that.

Maybe using ts-soft's ReadData_M instead of ReadByte_M would help? (sorry if you already tried this)

Ragards,