Re: gcc 3.4.3 performance problem illustrated

[Paul Pluzhnikov]

Kenneth Massey <kmassey@masseyratings.com> writes:

> BTW, are there any tools to actually count cache misses?

  valgrind --tool=cacheprof
  tsprof (http://www.bitwagon.com)

> 1) is this observed only on AMD64, or also x86 ?

On "Intel(R) Pentium(R) 4 CPU 1500MHz", compiled with 
"-O3 -march=pentium4", my results are:

g++3.2.3: 2.03
g++3.4.3: 2.19
g++4.0.0: 2.15

On "Pentium III (Coppermine) 850MHz", compiled with -O3

g++2.95.3: 6.95
g++3.2.3:  6.75
g++3.4.3:  6.78
g++4.0.0:  6.68

P.S. There is an off-by-one bug which actually causes this to crash
when compiled with g++2.95:

>   for (int i=0;i<=k;i++)

Better make that "for (int i = 0; i < k; i++)"

Cheers,
-- 
In order to understand recursion you must first understand recursion.
Remove /-nsp/ for email.