[fpc-pascal] Problems with valgrind
Jonas Maebe
jonas.maebe at elis.ugent.be
Mon Oct 15 15:06:58 CEST 2012
On 15 Oct 2012, at 14:08, Darius Blaszyk wrote:
> I'm trying to use valgrind for the first time seriously to profile my app. See output from console below;
>
[snip]
> After googling I found some remark about adding a -m32 flag with GCC. Is this the same issue I have here?
No, because -m32 is a Linux-specific flag. And Valgrind for Mac OS X by default supports both 32 and 64 bit binaries.
> What is the remedy?
I don't know. It works fine here on Mac OS X 10.6 with Valgrind 3.8.0:
$ fpc -gv prog.pp
Free Pascal Compiler version 2.6.0 [2011/12/30] for i386
Copyright (c) 1993-2011 by Florian Klaempfl and others
Target OS: Darwin for i386
Compiling prog.pp
Compiling unit1.pp
Assembling (pipe) unit1.s
Assembling (pipe) prog.s
Linking prog
18 lines compiled, 1.2 sec
$ valgrind --tool=cachegrind ./prog
==790== Cachegrind, a cache and branch-prediction profiler
==790== Copyright (C) 2002-2011, and GNU GPL'd, by Nicholas Nethercote et al.
==790== Using Valgrind-3.8.0.SVN and LibVEX; rerun with -h for copyright info
==790== Command: ./prog
==790==
--790-- ./prog:
--790-- dSYM directory is missing; consider using --dsymutil=yes
==790==
==790== I refs: 1,061,926
==790== I1 misses: 1,811
==790== LLi misses: 1,372
==790== I1 miss rate: 0.17%
==790== LLi miss rate: 0.12%
==790==
==790== D refs: 541,094 (384,427 rd + 156,667 wr)
==790== D1 misses: 2,997 ( 1,802 rd + 1,195 wr)
==790== LLd misses: 2,156 ( 1,326 rd + 830 wr)
==790== D1 miss rate: 0.5% ( 0.4% + 0.7% )
==790== LLd miss rate: 0.3% ( 0.3% + 0.5% )
==790==
==790== LL refs: 4,808 ( 3,613 rd + 1,195 wr)
==790== LL misses: 3,528 ( 2,698 rd + 830 wr)
==790== LL miss rate: 0.2% ( 0.1% + 0.5%
You may want to look at the Instruments application that comes with Xcode though. That one offers a lot of different profiling options based on performance counters and sampling.
Jonas
More information about the fpc-pascal
mailing list