Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: LAME Speed on a Mac (Read 2826 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

LAME Speed on a Mac

I use LAME 3.91 on my Macintosh G4 running OS X 10.1. This is a 400 MHz G4 system--granted, not the fastest out there, but I get the feeling it should be a heck of a lot faster than it is. Using --alt-preset fast standard I average around 1.5x, and with --alt-preset standard it's not even half that. This, on a system that in most other respects is at least as fast as the 733 MHz Pentium III sitting right next to it. LAME is compiled using the tools provided on the most recent Developer Tools distribution.

Suggestions are appreciated (but please don't suggest I simply use the PIII  )

--Ken

LAME Speed on a Mac

Reply #1
Quote
Originally posted by Rumkeg
LAME is compiled using the tools provided on the most recent Developer Tools distribution.


Which compiler comes with this package?

The issue of LAME running slowly on the G4 has been discussed here before actually, and I believe it's not so much a problem with LAME but a problem with gcc.  gcc produces about 30% slower code than the Intel compiler even on the x86 if I remember correctly, and according to some, the situation (gcc being slow that is) is much worse on the PPC, at least in regards to LAME.

TrNSZ has told me that in his experience in the past, Codewarrior has produced binaries up to 40% faster (or more) than gcc.  All this and Codewarrior isn't even a vectorizing compiler (vectorizing meaning it would make automatic use of special instructions like mmx, sse, or in the case of the G4 it would make use of altivec)... so that could mean that with a vectorizing compiler you could get perhaps even up to an 80% or more speed increase over gcc.. heh.

I've done a bit of research and found this compiler:

http://www.psrv.com/altivec.html

Which is a vectorizing compiler for the G4.  Unfortunately I don't have access to a G4 or to this compiler and don't know anyone that does.  If I were looking to produce faster binaries for the G4 platform though, and had the resources, this is where I'd look.

 

LAME Speed on a Mac

Reply #2
i'm running lame in osx on a dual 867 (oc'd to get there, of course). i encode using --alt-preset standard @ about 1.0-1.4x, and since there's no smp in lame (yet?), i can encode two sources simultaneously at that speed (rather than doubling it).

i have the latest (dec 2001) osx devtools installed. here's our compiler:

Code: [Select]
[localhost:~] bharding% gcc -v

Reading specs from /usr/libexec/gcc/darwin/ppc/2.95.2/specs

Apple Computer, Inc. version gcc-934.3, based on gcc version 2.95.2 19991024 (release)


i actually saw you mention the vast compiler before, dibrom, and i've been looking for a copy to test on lame ever since -- assuming, of course, that i can figure out what to do once i get it. i put out a call for help compiling with codewarrior (which i have access to presently) to no avail a few weeks ago.

brett.