[vorbis-dev] CVS troubles

[Segher Boessenkool]

> I assume you mean dsptest?

Yes I do.

> 
> I'm beginning to suspect that a few of you folks have more running code than I
> do right now ;-)
> 
> > Later this week I'll send some optimizations to Monty; it's easier to test if
> > it runs a bit faster :-)
> 
> Agreed.  My own profiling shows the MDCT to be the major bottleneck (80% of 
> execution time.  The price of a simpler, non-subbanding algortihm).  The major 

I'm sorry to disappoint you, but the psych model is the bottleneck. Try
compiling with

-pg -static

It is a real eye-opener :-)

Library routines like cos, sin, atan, log, exp are not shown in a normal
gprof listing. Ouch. cos was using a _lot_ of time.

> problem there is using gcc/egc as the compiler.  On IRIX, the MIPSpro CC 
> geenrates code 4x faster than GCC and on Solaris, SunSoft CC generates code 
> 2.5x faster.  I'd assume the same is true on Intel but I don't have any 
> commercial compilers to bench against on x86.

On intel, on my mdct algorithm, gcc is about 2x faster than visual c.

**Segher

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/