Re: WORKAROUND: transcode-1.0.0-1.2.fc4.i386.rpm - segmentation fault

I have AMD Athlon XP 3200+.  I read different postings from a google search and
from what I gather unless you have an Athlon 64 all bets are off (and since I
don't have one of those, I won't vouch that works either).  I guess the key
phase is "compatibility" and what exactly that means.  I also noticed something
else interesting when experimenting with the --accel option.  If I use "--accel
autodetect" it also works.  If I use SSE or SSE2 as an option, transcode fails
with a "segmentation fault".  I did notice in the comments by Matthias for one
of the beta builds:  "Explicitly disable MMX for now, it makes the build fail
:-(".  Could that have anything to do with this?  For now, since I have an AMD
chip, I am using "--accel 3dnow".

Quoting Ralf Ertzinger <[email protected]>:

> Hi.
>
> "Gerald B. Cox" <[email protected]> wrote:
>
>> It appears that there is something wrong with the acceleration
>> autodetection function.  If you look at my previous posts you will find
>> that when using the transcode-0.6.14-1.1.fc3.fr.i386.rpm that the
>> acceleration used was "mmx".  The new version defaults to "sse".
>> Apparently SSE/SSE2 isn't supported on AMD ATHLON processors
>
> Depending on what generation athlon we are talking about they support
> everything from MMX to SSE3. Unless I am very much mistaken all
> Athlon class processors (including Duron) support at least MMX
> and a partial implementation of SSE. From the Palomino/Morgan core
> onwards (that's in the 2001 timeframe), SSE support is complete.
> What processor are you using?
>
> --
> Q. What's the difference between a viola and a trampoline?
> A. You take your shoes off to jump on a trampoline.
>