On Apr 5, 2016, at 4:00 AM, pd-dev-request@lists.iem.at wrote:
I don't know whether externals can be used with libpd now, but I do know that iemnet[1] has threaded networking classes.
Most externals work fine with libpd.
-------- Dan Wilcox @danomatika https://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Hi,what are the reasons why a) DEFDACBLKSIZE is defined in multiple places extra/pd~/pd~.c:50:#define DEFDACBLKSIZE 64src/s_audio_esd.c:71:#define DEFDACBLKSIZE 64src/s_stuff.h:66:#define DEFDACBLKSIZE 64? b) DEFSENDVS is defined as src/d_global.c:10:#define DEFSENDVS 64 /* LATER get send to get this from canvas */ and not as #define DEFSENDVS DEFDACBLKSIZE ? c) the default value for DEFDACBLKSIZE is 64 ?On Bela I have DEFDACBLKSIZE==8 and DEFSENDVS==8 and I yet have to find troubles with this setup.
Thanks,Giulio
On 07/11/2016 06:56 AM, Giulio Moro via Pd-dev wrote:
On Bela I have DEFDACBLKSIZE==8 and DEFSENDVS==8 and I yet have to find troubles with this setup.
the joys of having not needing to support multiple operating systems?
also, the overhead with a blocksize of 8 is considerably higher than a blocksize of 64, so when using *heavy* Pd-patches you might hit performance problems sooner than later.
anyhow, i'm all for having a variable dac blocksize $ pd -dacblocksize 8
ain't this is a deja vu [1]?
gfdsamr IOhannes
[1] https://lists.puredata.info/pipermail/pd-dev/2004-11/003152.html
I've wanted variable dac block size as well, for embedding libpd in a plugin.. I'm guessing PD doesn't play well with jack when jack's block size is less than 64?
On Mon, Jul 11, 2016 at 2:44 PM, IOhannes m zmölnig zmoelnig@iem.at wrote:
On 07/11/2016 06:56 AM, Giulio Moro via Pd-dev wrote:
On Bela I have DEFDACBLKSIZE==8 and DEFSENDVS==8 and I yet have to find
troubles with this setup.
the joys of having not needing to support multiple operating systems?
also, the overhead with a blocksize of 8 is considerably higher than a blocksize of 64, so when using *heavy* Pd-patches you might hit performance problems sooner than later.
anyhow, i'm all for having a variable dac blocksize $ pd -dacblocksize 8
ain't this is a deja vu [1]?
gfdsamr IOhannes
[1] https://lists.puredata.info/pipermail/pd-dev/2004-11/003152.html
Pd-dev mailing list Pd-dev@lists.iem.at https://lists.puredata.info/listinfo/pd-dev
well if you are looking for sizes multiples of DEFDACBLKSIZE it is already feasible. My original question was also about why it is defined in multiple places and why DEFSENDVS is defined separately, yet setting them to two different values gives errors. Giulio
From: Alex x37v.alex@gmail.com To: IOhannes m zmölnig zmoelnig@iem.at Cc: pd-dev@lists.iem.at Sent: Monday, 11 July 2016, 23:33 Subject: Re: [PD-dev] DEFDACBLKSIZE and DEFSENDVS
I've wanted variable dac block size as well, for embedding libpd in a plugin.. I'm guessing PD doesn't play well with jack when jack's block size is less than 64?
On Mon, Jul 11, 2016 at 2:44 PM, IOhannes m zmölnig zmoelnig@iem.at wrote:
On 07/11/2016 06:56 AM, Giulio Moro via Pd-dev wrote:
On Bela I have DEFDACBLKSIZE==8 and DEFSENDVS==8 and I yet have to find troubles with this setup.
the joys of having not needing to support multiple operating systems?
also, the overhead with a blocksize of 8 is considerably higher than a blocksize of 64, so when using *heavy* Pd-patches you might hit performance problems sooner than later.
anyhow, i'm all for having a variable dac blocksize $ pd -dacblocksize 8
ain't this is a deja vu [1]?
gfdsamr IOhannes
[1] https://lists.puredata.info/pipermail/pd-dev/2004-11/003152.html
_______________________________________________ Pd-dev mailing list Pd-dev@lists.iem.at https://lists.puredata.info/listinfo/pd-dev
_______________________________________________ Pd-dev mailing list Pd-dev@lists.iem.at https://lists.puredata.info/listinfo/pd-dev
On 2016-07-12 04:51, Giulio Moro via Pd-dev wrote:
My original question was also about why it is defined in multiple places and why DEFSENDVS is defined separately, yet setting them to two different values gives errors.
was it? i don't see this question in your original mail. also: what is the error you get?
fgamsdr IOhannes
I see you found my question, thanks for your responses. Here is the error I get when DEFDACBLKSIZE is 8 and DEFSENDVS is 64 : error: throw~ out: vector size mismatch
I was getting similar errors when using [s~] [r~] or [catch~].
also, the overhead with a blocksize of 8 is considerably higher than a blocksize of 64, so when using *heavy* Pd-patches you might hit performance problems sooner than later.
Where is the performance hit for DEFDACBLKSIZE = 8 ?I see how there could be a performance hit when using a smaller audio driver block size, but would there be much difference between, e.g. DEFDACBLKSIZE = 8 and DEFDACBLKSIZE = 64 if the audio driver block size is set to 64 in both cases?
Thanks,Giulio
From: IOhannes m zmoelnig zmoelnig@iem.at To: pd-dev@lists.iem.at Sent: Tuesday, 12 July 2016, 8:46 Subject: Re: [PD-dev] DEFDACBLKSIZE and DEFSENDVS
On 2016-07-12 04:51, Giulio Moro via Pd-dev wrote:
My original question was also about why it is defined in multiple places and why DEFSENDVS is defined separately, yet setting them to two different values gives errors.
was it? i don't see this question in your original mail. also: what is the error you get?
fgamsdr IOhannes
_______________________________________________ Pd-dev mailing list Pd-dev@lists.iem.at https://lists.puredata.info/listinfo/pd-dev
On 2016-07-11 06:56, Giulio Moro via Pd-dev wrote:
Hi,what are the reasons why a) DEFDACBLKSIZE is defined in multiple places extra/pd~/pd~.c:50:#define DEFDACBLKSIZE 64src/s_audio_esd.c:71:#define DEFDACBLKSIZE 64src/s_stuff.h:66:#define DEFDACBLKSIZE 64?
well:
- extra/pd~/pd~.c only defines DEFDACBLKSIZE when compiling the external for max/msp - which most likely lacks this define, so it needs to be defined in order to use it. when compiling the external for Pd, it includes s_stuff.h and uses that define (as it should).
- src/s_audio_esd.c is unused (no build-system i know bothers to compile this file) - most likely this is just undeleted cruft lying around.
so for the sake of Pd, DEFDACBLKSIZE is only really defined once.
fgamsdr IOhannes
On 2016-07-11 06:56, Giulio Moro via Pd-dev wrote:
b) DEFSENDVS is defined as src/d_global.c:10:#define DEFSENDVS 64 /* LATER get send to get this from canvas */ and not as #define DEFSENDVS DEFDACBLKSIZE
because they are unrelated. DEFDACBLKSIZE is the blocksize enforced by [dac~]/[adc~]. DEFSENDVS is the blocksize enforced by [send~]/[receive~].
the plan obviously being that DEFSENDVS should be made dynamically "asap", whereas there is less urge to change DEFDACBLKSIZE.
-
Alex -
Dan Wilcox -
Giulio Moro -
IOhannes m zmoelnig -
IOhannes m zmölnig