Re^2: Random Numbers under XP: Translating XS to Win32::API

Super! That's just what I needed.

I'll probably eliminate the use of the function 'renaming', and just call SystemFunction036 (with appropriate comments for clarity).

I noted your use of the 'V*' directive for the unpack call. Although with random data, distinguishing between big-endian and little-endian is superfluous.

Thanks for the code!

Remember: There's always one more bug.

Comment on Re^2: Random Numbers under XP: Translating XS to Win32::API

Replies are listed 'Best First'.
Re^3: Random Numbers under XP: Translating XS to Win32::API by BrowserUk (Patriarch) on Jun 29, 2005 at 20:20 UTC
I noted your use of the 'V' directive for the unpack call. Although with random data, distinguishing between big-endian and little-endian is superfluous.* True, but as the platform is little-endian, converting using V* rather than N* saves doing a lot of fiddly byte and bit-twiddling. This really shows up when you unpack a few thousand ulongs. Below I timed unpacking the same 1000 values both ways: `P:\test>471107 Name "main::SystemFunction036" used only once: possible typo at P:\tes +t\471107.pl line 14. 1 trial of N* non-native BigEndian ( 1.848ms total), 1.848ms/trial 1 trial of V* native LittleEndian ( 310us total), 310us/trial` [download] It essentially is the difference between doing nothing and doing something (fiddly). Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. Lingua non convalesco, consenesco et abolesco. -- Rule 1 has a caveat! -- Who broke the cabal? "Science is about questioning the status quo. Questioning authority". The "good enough" maybe good enough for the now, and perfection maybe unobtainable, but that should not preclude us from striving for perfection, when time, circumstance or desire allow.	[reply] [d/l]

Replies are listed 'Best First'.

Re^3: Random Numbers under XP: Translating XS to Win32::API
by BrowserUk (Patriarch) on Jun 29, 2005 at 20:20 UTC

I noted your use of the 'V*' directive for the unpack call. Although with random data, distinguishing between big-endian and little-endian is superfluous.

True, but as the platform is little-endian, converting using V* rather than N* saves doing a lot of fiddly byte and bit-twiddling. This really shows up when you unpack a few thousand ulongs.

Below I timed unpacking the same 1000 values both ways:

P:\test>471107
Name "main::SystemFunction036" used only once: possible typo at P:\tes
+t\471107.pl line 14.
  1 trial  of N* non-native BigEndian (  1.848ms total), 1.848ms/trial

  1 trial  of  V* native LittleEndian (    310us total), 310us/trial
[download]

It essentially is the difference between doing nothing and doing something (fiddly).

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

Lingua non convalesco, consenesco et abolesco. -- Rule 1 has a caveat! -- Who broke the cabal?

"Science is about questioning the status quo. Questioning authority".

The "good enough" maybe good enough for the now, and perfection maybe unobtainable, but that should not preclude us from striving for perfection, when time, circumstance or desire allow.

[reply]
[d/l]