Wireshark-dev: Re: [Wireshark-dev] Startup speed up - assembler usage
From: "Kukosa, Tomas" <[email protected]>
Date: Wed, 21 Nov 2007 09:39:02 +0100
Hi,

sorry for my delay but I was little bit busy during last days.

What I have done:
I have renamed and made few new functions in proto.c for more
comfortable profiling.
  g_strcmp() renamed to wrs_strcmp()
  g_str_hash() from glib coppied into proto.c with new name
wrs_str_hash()
  some parts of proto_register_field_init() put into new functions
wrs_check_charset() and tmp_check_assert()

See attached results:
  presultr0.txt - all implemented in C
  presultr1.txt - wrs_strcmp() and wrs_str_hash() implemented in
assembler
 Note: you will not see individual times for wrs_strcmp() and
wrs_str_hash() in the presultr1.txt as my assembler (NASM) does nut
support MS debug information but you can compare time of caller
functions and total time

My intention is not to have implemented much code in assembler but just
few simple and critical functions.



Mailcode: Ph5U5F3p
> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of Guy Harris
> Sent: Thursday, November 08, 2007 1:22 PM
> To: Developer support list for Wireshark
> Subject: Re: [Wireshark-dev] Startup speed up - assembler usage
> 
> Kukosa, Tomas wrote:
> 
> > E.g. I was able (with about 80 lines in assembler) to reduce time
> > spent in libwireshark.dll during startup from cca 2600ms to 
> 1300ms on my PC.
> 
> Which code did you change from C to assembler to get that speedup?
> _______________________________________________
> Wireshark-dev mailing list
> [email protected]
> http://www.wireshark.org/mailman/listinfo/wireshark-dev
> 

Attachment: result.zip
Description: result.zip