Re: [PD] spread Gem-computation over several dsp-cycles (?) (was: [Gem]: gem-pointer and [list] OR slow [repeat])

1 Mar 2007

      hello cyrille
thank you for the adjustments. i think i understand the difference
between measuring the gemhead loop and the time between 2 images. but
the other thing with the optimization still remains unclear to me and it
seems, that it doesn't work here. when i stop the first and start the
second gemhead, the gemwin becomes black. no primitives are drawn. there
is no error in the pd-window (there is only a message '[GEMglNewList]:
mode=4864' when i load the patch).
does the optimization need some flags enabled when compiling gem?
it seems, there is so much about gem, i don't know yet (like all these
objects [GEMgl*] and [GLdefine] and the like, or the message 'FSAA 4').
to understand them, is it needed to know opengl well? are these objects
documented somewhere in the Gem-documentation?
i am not insulted if you don't have the time to answer all these
questions...
cheers
roman
On Thu, 2007-03-01 at 11:18 +0100, cyrille henry wrote:
...
i made some change to this abstraction in order to compute only the time use for the gemhead loop and not the time between 2 images.
on my computer, it's about 11ms.
but with the display list optimisation, it fall to 6ms about.
cyrille
Roman Haefeli a Ã©crit :
...
as always: i forgot the attachment........
On Wed, 2007-02-28 at 23:24 +0100, Roman Haefeli wrote:
...
On Wed, 2007-02-28 at 07:14 -0600, chris clepper wrote:
...
On 2/28/07, Roman Haefeli reduzierer@yahoo.de wrote:
    i might be wrong but in my eyes it doesn't make sense to do
    all the work
    that could be done in 50ms in only 1.45ms. 

What?  GEM doesn't use the DSP clock.  It will take as much time as
needed to render.
oops. ok....
...
For example, the current work I have uses three or four 1080i clips, a
live feed and records to disk and there is no way that all runs in
1.45ms.  It takes about 12-15ms!
anyway, i get dropouts when doing gem-rendering, although 'top' tells me
that pd uses only 20% cpu-time. i don't care much about the audio (as
IOhannes mentioned, i could run two instances of pd). the problem is
that the timing is not nice. i'd like to run the patch with 20 frames
per second. but in praxis each cycle needs 70ms, which gives me a
framerate of 14.
is my gpu too slow? what happens, when the gpu is overloaded? can that
cause pd to stuck?
i attached a little gem-benchmark-test. although you say, gem doesn't
use the dsp-clock, it takes much longer to compute the first block after
a gem-rendering command. why is that? 
and: here on my system, the [realtime] measures up to 70ms, when i go
over [repeat 1400] (under 1400 it's 50ms). the funny thing is, that it
stays around 70ms, even if i set the [repeat] up to 3000 or more. why is
that? here on my system, cpu-time used by pd is always 20%.
sorry to ask you so much...... but i try to understand things a bit
better.......
roman
___________________________________________________________ 
Der frhe Vogel fngt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de

PD-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list

#N canvas 394 19 736 552 10;
#X obj 26 123 gemwin;
#X obj 26 42 sel 0 1;
#X msg 26 95 0 , destroy;
#X obj 26 19 tgl 15 0 empty empty empty 17 7 0 10 -262144 -1 -1 1 1
;
#X obj 216 102 gemhead;
#X obj 216 152 translateXYZ 1 1 0;
#X obj 23 152 gemhead;
#X obj 23 210 world_light;
#X msg 48 66 perspec -1 1 -1 1 2 2000 , lighting 1 , create , 1
;
#X obj 23 177 rotateXYZ 30 -54 0;
#N canvas 1115 145 249 311 measure_realtime 0;
#X obj 39 40 t b b;
#X obj 39 256 outlet;
#X obj 39 16 gemhead;
#X obj 39 141 f;
#X obj 39 186 +;
#X obj 39 163 * 9;
#X obj 39 209 * 0.1;
#X obj 39 118 t b f;
#X text 97 155 smooth it a bit;
#X obj 39 62 realtime;
#X connect 0 0 9 0;
#X connect 0 1 9 1;
#X connect 2 0 0 0;
#X connect 3 0 5 0;
#X connect 4 0 6 0;
#X connect 5 0 4 0;
#X connect 6 0 3 1;
#X connect 6 0 1 0;
#X connect 7 0 3 0;
#X connect 7 1 4 1;
#X connect 9 0 7 0;
#X restore 198 339 pd measure_realtime;
#X floatatom 198 368 5 0 0 0 - - -;
#X obj 217 236 rotateXYZ 1 0 0;
#X text 247 370 <- check if it goes higher than 50ms;
#X obj 217 298 cube 0.02;
#X obj 217 211 translateXYZ -0.001 0 0;
#X obj 217 263 translateXYZ 0 0 0.02;
#X obj 31 249 bang~;
#X obj 31 276 t b b;
#X obj 31 300 realtime;
#X obj 31 326 t b f;
#X obj 31 350 f;
#X obj 58 351 + 1;
#X obj 31 379 pack f f;
#X obj 99 245 gemhead;
#X obj 99 270 b;
#X obj 99 293 0;
#X floatatom 31 434 0 0 0 0 - - -;
#X floatatom 85 435 0 0 0 0 - - -;
#X floatatom 127 435 0 0 0 0 - - -;
#X floatatom 171 436 0 0 0 0 - - -;
#X floatatom 214 436 0 0 0 0 - - -;
#X floatatom 255 437 0 0 0 0 - - -;
#X text 37 457 1;
#X text 94 458 2;
#X text 135 458 3;
#X text 180 459 4;
#X text 225 459 5;
#X obj 31 403 route 0 1 2 3 4 5;
#X text 262 459 6;
#X text 28 477 realtime measured lenght of the nth dsp-cycle after
the [gemhead] starts rendering.;
#X obj 216 129 rotateXYZ 0 0 90;
#X obj 216 187 repeat 2000;
#X text 55 17 <- rendering on/off;
#X floatatom 380 146 5 0 0 0 - - -;
#X text 425 143 <- try different 'loads';
#X connect 1 0 2 0;
#X connect 1 1 8 0;
#X connect 2 0 0 0;
#X connect 3 0 1 0;
#X connect 4 0 41 0;
#X connect 5 0 42 0;
#X connect 6 0 9 0;
#X connect 8 0 0 0;
#X connect 9 0 7 0;
#X connect 10 0 11 0;
#X connect 12 0 16 0;
#X connect 15 0 12 0;
#X connect 16 0 14 0;
#X connect 17 0 18 0;
#X connect 18 0 19 0;
#X connect 18 1 19 1;
#X connect 19 0 20 0;
#X connect 20 0 21 0;
#X connect 20 1 23 1;
#X connect 21 0 22 0;
#X connect 21 0 23 0;
#X connect 22 0 21 1;
#X connect 23 0 38 0;
#X connect 24 0 25 0;
#X connect 25 0 26 0;
#X connect 26 0 21 1;
#X connect 38 0 27 0;
#X connect 38 1 28 0;
#X connect 38 2 29 0;
#X connect 38 3 30 0;
#X connect 38 4 31 0;
#X connect 38 5 32 0;
#X connect 41 0 5 0;
#X connect 42 0 15 0;
#X connect 44 0 42 1;

PD-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
plain text document attachment (gem_little_benchmark2.pd)
#N canvas 21 42 892 721 10;
#X obj 26 123 gemwin;
#X obj 26 42 sel 0 1;
#X msg 26 95 0 , destroy;
#X obj 26 19 tgl 15 0 empty empty empty 17 7 0 10 -262144 -1 -1 1 1
;
#X obj 216 102 gemhead;
#X obj 216 152 translateXYZ 1 1 0;
#X obj 23 152 gemhead;
#X obj 23 210 world_light;
#X obj 23 177 rotateXYZ 30 -54 0;
#X obj 216 236 rotateXYZ 1 0 0;
#X text 537 651 <- check if it goes higher than 50ms;
#X obj 216 298 cube 0.02;
#X obj 216 211 translateXYZ -0.001 0 0;
#X obj 216 263 translateXYZ 0 0 0.02;
#X obj 31 249 bang~;
#X obj 31 276 t b b;
#X obj 31 300 realtime;
#X obj 31 326 t b f;
#X obj 31 350 f;
#X obj 58 351 + 1;
#X obj 31 379 pack f f;
#X obj 99 245 gemhead;
#X obj 99 270 b;
#X obj 99 293 0;
#X floatatom 31 434 0 0 0 0 - - -;
#X floatatom 85 435 0 0 0 0 - - -;
#X floatatom 127 435 0 0 0 0 - - -;
#X floatatom 171 436 0 0 0 0 - - -;
#X floatatom 214 436 0 0 0 0 - - -;
#X floatatom 255 437 0 0 0 0 - - -;
#X text 37 457 1;
#X text 94 458 2;
#X text 135 458 3;
#X text 180 459 4;
#X text 225 459 5;
#X obj 31 403 route 0 1 2 3 4 5;
#X text 262 459 6;
#X text 28 477 realtime measured lenght of the nth dsp-cycle after
the [gemhead] starts rendering.;
#X obj 216 129 rotateXYZ 0 0 90;
#X obj 216 187 repeat 2000;
#X text 55 17 <- rendering on/off;
#X floatatom 380 146 5 0 0 0 - - -;
#X text 425 143 <- try different 'loads';
#X msg 48 66 perspec -1 1 -1 1 2 2000 , FSAA 4 , lighting 1 , create
, 1;
#X obj 486 558 f;
#X obj 486 603 +;
#X obj 486 580 * 9;
#X obj 486 626 * 0.1;
#X obj 486 535 t b f;
#X obj 486 510 realtime;
#X floatatom 486 651 5 0 0 0 - - -;
#X obj 537 444 gemhead 99;
#X obj 486 467 gemhead 1;
#X obj 486 489 b;
#X obj 537 489 b;
#N canvas 192 353 840 624 optimized 0;
#X obj 37 279 inlet gemlist;
#X obj 37 384 outlet gemlist;
#X obj 186 221 GEMglGenLists 1;
#X obj 156 398 GEMglNewList;
#X obj 37 322 GEMglCallList;
#N canvas 0 0 450 300 once 0;
#X obj 91 62 inlet;
#X obj 91 266 outlet;
#X obj 91 150 spigot 1;
#X obj 91 175 t a b;
#X msg 121 197 0;
#X obj 298 57 inlet open;
#X obj 298 92 t b;
#X msg 142 128 1;
#X connect 0 0 2 0;
#X connect 2 0 3 0;
#X connect 3 0 1 0;
#X connect 3 1 4 0;
#X connect 4 0 2 1;
#X connect 5 0 6 0;
#X connect 6 0 7 0;
#X connect 7 0 2 1;
#X restore 186 199 pd once;
#N canvas 0 0 450 300 once 0;
#X obj 91 62 inlet;
#X obj 91 266 outlet;
#X obj 91 150 spigot 1;
#X obj 91 175 t a b;
#X msg 121 197 0;
#X obj 298 57 inlet open;
#X obj 298 92 t b;
#X msg 142 128 1;
#X connect 0 0 2 0;
#X connect 2 0 3 0;
#X connect 3 0 1 0;
#X connect 3 1 4 0;
#X connect 4 0 2 1;
#X connect 5 0 6 0;
#X connect 6 0 7 0;
#X connect 7 0 2 1;
#X restore 156 372 pd once;
#X obj 235 354 loadbang;
#X obj 235 373 GLdefine GL_COMPILE;
#X msg 230 178 bang;
#X obj 156 488 GEMglEndList;
#X obj 156 425 t a a;
#X obj 156 113 gemhead 1;
#X obj 316 267 t b;
#X obj 156 138 t a a;
#X obj 286 241 t f b;
#X obj 165 268 t f f;
#X msg 305 343 bang;
#N canvas 0 0 360 251 primitive 0;
#X obj 37 9 inlet gemlist;
#X obj 34 68 rotateXYZ 1 0 0;
#X obj 34 130 cube 0.02;
#X obj 34 43 translateXYZ -0.001 0 0;
#X obj 34 95 translateXYZ 0 0 0.02;
#X connect 0 0 3 0;
#X connect 1 0 4 0;
#X connect 3 0 1 0;
#X connect 4 0 2 0;
#X restore 186 467 pd primitive;
#X connect 0 0 4 0;
#X connect 2 1 15 0;
#X connect 3 0 11 0;
#X connect 4 0 1 0;
#X connect 5 0 2 0;
#X connect 6 0 3 0;
#X connect 7 0 8 0;
#X connect 8 0 3 2;
#X connect 9 0 5 1;
#X connect 11 0 10 0;
#X connect 11 1 18 0;
#X connect 12 0 14 0;
#X connect 13 0 6 1;
#X connect 14 0 6 0;
#X connect 14 1 5 0;
#X connect 15 0 16 0;
#X connect 15 1 13 0;
#X connect 16 0 4 1;
#X connect 16 1 3 1;
#X connect 17 0 8 0;
#X restore 624 283 pd optimized primitive;
#X obj 624 168 gemhead;
#X obj 624 218 translateXYZ 1 1 0;
#X obj 624 195 rotateXYZ 0 0 90;
#X obj 624 253 repeat 2000;
#X floatatom 788 212 5 0 0 0 - - -;
#X obj 310 101 tgl 15 0 empty empty empty 17 7 0 10 -262144 -1 -1 0
1;
#X obj 624 147 tgl 15 0 empty empty empty 17 7 0 10 -262144 -1 -1 0
1;
#X msg 624 125 0;
#X obj 624 102 loadbang;
#X connect 1 0 2 0;
#X connect 1 1 43 0;
#X connect 2 0 0 0;
#X connect 3 0 1 0;
#X connect 4 0 38 0;
#X connect 5 0 39 0;
#X connect 6 0 8 0;
#X connect 8 0 7 0;
#X connect 9 0 13 0;
#X connect 12 0 9 0;
#X connect 13 0 11 0;
#X connect 14 0 15 0;
#X connect 15 0 16 0;
#X connect 15 1 16 1;
#X connect 16 0 17 0;
#X connect 17 0 18 0;
#X connect 17 1 20 1;
#X connect 18 0 19 0;
#X connect 18 0 20 0;
#X connect 19 0 18 1;
#X connect 20 0 35 0;
#X connect 21 0 22 0;
#X connect 22 0 23 0;
#X connect 23 0 18 1;
#X connect 35 0 24 0;
#X connect 35 1 25 0;
#X connect 35 2 26 0;
#X connect 35 3 27 0;
#X connect 35 4 28 0;
#X connect 35 5 29 0;
#X connect 38 0 5 0;
#X connect 39 0 12 0;
#X connect 41 0 39 1;
#X connect 43 0 0 0;
#X connect 44 0 46 0;
#X connect 45 0 47 0;
#X connect 46 0 45 0;
#X connect 47 0 44 1;
#X connect 47 0 50 0;
#X connect 48 0 44 0;
#X connect 48 1 45 1;
#X connect 49 0 48 0;
#X connect 51 0 54 0;
#X connect 52 0 53 0;
#X connect 53 0 49 0;
#X connect 54 0 49 1;
#X connect 56 0 58 0;
#X connect 57 0 59 0;
#X connect 58 0 57 0;
#X connect 59 0 55 0;
#X connect 60 0 59 1;
#X connect 61 0 4 0;
#X connect 62 0 56 0;
#X connect 63 0 62 0;
#X connect 64 0 63 0;
_______________________________________________
PD-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
___________________________________________________________ 
Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

Re: [PD] spread Gem-computation over several dsp-cycles (?) (was: [Gem]: gem-pointer and [list] OR slow [repeat])