TheSin wrote:
acctually so the debs stay the same, fink needs to have a virtual pkg for altivec and then have a lame-altivec version which would depend on altivec and would could then use -faltivec...like i did with ffmpeg, but with ffmpeg I forced it on, not really great but since there is no way to detect altivec yet... anyhow just my 2 cents, not to mention -faltivec from my understanding only affects .s files (asm)...does lame even have altivec code?

No idea, I am glad to read about all the interest in that ArsTechnica forum, so no altivec code in lame, I'm just a happy consumer who really is experiencing more than a 200% increase with the compiled version I picked up from the BlackTree forum...Actually a 400% increase (3.2X versus fink's 0.7 when my 1Ghz Cube is not doing anything else but running the optimized-lame process on Duke Ellington Jazz tunes from the CD).


Some of those 'random' flags evidently had "quite" an effect on performance for evidently most Macsters who downloaded/used the iTunes-Lame plugin app or the compiled lame binaries, and if there were ever altivec code introduced to lame 4.0 what a speed demon lame would become :)

Why can't the package for lame in fink be set to compile with those flags deemed to enhance the performance on G4's running 10.3.x code?

On 13-Nov-03, at 9:23 PM, Alexander Strange wrote:


On Nov 13, 2003, at 1:38 AM, Gordon Mackenzie wrote:


It's sooooo slow when compared to the compiled versions I've found elsewhere for the Mac.

I have found an increase of about 200% (fink-lame speed of 0.7 versus altivec-compiled-lame of 2.2/2.3) when lame is compiled as these fellows have done (on G4's):

http://forums.blacktree.com/viewtopic.php?t=58&start=15



That is not altiveced. GCC does not automatically Altivec anything.
Also, all of those flags are overkill.
'-fast -mcpu=750 -fsingle-precision-constant' is more than enough for LAME. This isn't Gentoo ;)


What you should actually be looking at is this thread where I helped AltiVec parts of LAME (and I'll submit the patch eventually, with code to auto-detect Altivec, but it's extremely low priority for me):

http://arstechnica.infopop.net/OpenTopic/page? a=tpc&s=50009562&f=8300945231&m=1640932885


That actually gives a useful speed up as opposed to just sticking random compiler flags on.







------------------------------------------------------- This SF. Net email is sponsored by: GoToMyPC GoToMyPC is the fast, easy and secure way to access your computer from any Web browser or wireless device. Click here to Try it Free! https://www.gotomypc.com/tr/OSDN/AW/Q4_2003/t/g22lp?Target=mm/g22lp.tmpl _______________________________________________ Fink-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-devel

Reply via email to