[PD] Pd on Athlon64
IOhannes m zmoelnig
zmoelnig at iem.at
Thu Jan 17 21:59:22 CET 2008
Andy Farnell wrote:
> I was thinking of trying 64studio as a new main Pd machine.
> A few things are holding me back from upgrading;
> 1) The reported table/array problem, is that fixed now?
> 2) Extended - when you say there's no pd-extended, does this mean
> that most externals can be compiled 64 bit but none are collected
> together so you have to do each one yourself?
Pd itself does not have any problems any more with 64bit.
there might still be a few externals that are incompatible with 64bit
mode (that have to be ported, because they use low-level tricks), and
more externals that still miss the "-fPIC" flag (this is rather trivial
to solve when compiling)
> 3) Does the 64 bit version benefit from any computational accuracy
> improvement, or is it just running as 32 bit?
both the type of numbers in messages (t_float) and signals (t_sample)
are still single precision floating point.
In Pd-0.41 the types t_float and t_sample are used almost everywhere
(where applicable), i hope that i will have patches for pd-0.42 that
will allow you to compile Pd with double-precision.
> 4) Is there now a Debian package for 64 bit Pd that I can apt-get?
Pd has been in debian/amd64 for some years now.
i don't know whether günther has backported all the 0.41 fixes for 64bit.
personally i always have the debian-packages installed and never use them.
> Thanks for any further clues.
More information about the Pd-list