On Tue, 02 Nov 2004 01:10:40 +0100, Jenda Krynicky <[EMAIL PROTECTED]> wrote:
> It seems though that the constant doesn't matter much. I've tried to > copy a 709MB file using a tiny script that looked like this: ... > and it took 171s with 8*1024 and 176 with 5000 and 32*1024 byte > chunks. > > OTOH converting it to sysopen/sysread/syswrite made a big difference. > THe time went down to 79s with 32*1024 byte chunks but 180-185s with > 5000 byte ones. well, you are doing the OpSys work -- more or less -- for it. However, bug question is -- which method was more system freindly? or was this project copy simply a once-in-while thing you wanted to get done quickly? No matter what the system load performance cost was? -- WC -Sx- Jones http://youve-reached-the.endoftheinternet.org/ -- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] <http://learn.perl.org/> <http://learn.perl.org/first-response>
