Comments on: ARM compiler update http://hardwarebug.org/2010/01/15/arm-compiler-update/ Everything is broken Mon, 30 Aug 2010 09:33:38 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Mans http://hardwarebug.org/2010/01/15/arm-compiler-update/comment-page-1/#comment-505 Mans Sat, 23 Jan 2010 19:02:10 +0000 http://hardwarebug.org/?p=228#comment-505 When optimising for speed, I got these sizes for the .text section in the final <code>ffmpeg</code> executable: <table> <tbody> <tr><td>CS 2007q3</td><td>4.41 MB</td></tr> <tr><td>CS 2009q1</td><td>4.50 MB</td></tr> <tr><td>CS 2009q3</td><td>6.05 MB</td></tr> <tr><td>4.3.4</td><td>4.48 MB</td></tr> <tr><td>4.4.2</td><td>5.22 MB</td></tr> <tr><td>RVCT</td><td>3.77 MB</td></tr> <tr><td>TMS470</td><td>3.79 MB</td></tr> </tbody> </table> Compilers should never generate Thumb code when not asked to. When optimising for speed, I got these sizes for the .text section in the final ffmpeg executable:

CS 2007q3 4.41 MB
CS 2009q1 4.50 MB
CS 2009q3 6.05 MB
4.3.4 4.48 MB
4.4.2 5.22 MB
RVCT 3.77 MB
TMS470 3.79 MB

Compilers should never generate Thumb code when not asked to.

]]> By: funman http://hardwarebug.org/2010/01/15/arm-compiler-update/comment-page-1/#comment-504 funman Sat, 23 Jan 2010 17:39:20 +0000 http://hardwarebug.org/?p=228#comment-504 What about adding code size comparison of compilers ? It can affect cache usage, and in the case of rockbox every bit is very precious as it saves battery life when using hard disks. Not sure if the difference would be very noticeable though. Do some compilers generate thumb code even if not explicitely requested? What about adding code size comparison of compilers ?

It can affect cache usage, and in the case of rockbox every bit is very precious as it saves battery life when using hard disks.

Not sure if the difference would be very noticeable though.

Do some compilers generate thumb code even if not explicitely requested?

]]>