de
en
frMaximum Delay over Buffersize
Mach flott den Schrott:
- oldjack: Intel, 333 MHz, 260 MB RAM, 10 GB HD (noch weitere 10 GB HD frei)
- jack: AMD, 750 MHz, 400 MB RAM
- Messrechner1: Intel, 450 MHz, 130 MB RAM, 5 GB HD
- Messrechner2: Intel, 450 MHz, 130 MB RAM, 10 GB HD
- Messrechner3: Intel, 450 MHz, 130 MB RAM, 10 GB HD
avc_jack_benchmark
Das Programm avc_jack_benchmark ist ein Programm zum Test der Rechnerleistung unter JACK-Anwendungen. Es simuliert Rechnenarbeit und erhöht diese stetig.
| Rechner | System | Chipsatz | Taktfreq. | RAM | Soundcard | maxSampleDelay |
|---|---|---|---|---|---|---|
| brutejack | lenny, Xorg | Intel | 2 x 3 GHz | 4 GB | snd_virtuoso | 26.7 µs |
| brutejack | lenny, Xorg | Intel | 2 x 3 GHz | 4 GB | snd_hda_intel | 11.8 µs |
| heike | lenny, Xorg | Intel | 450 MHz | 256 MB | ad1816a | 9.8 µs |
| jack | etch, RT-lsm, Xorg | AMD | 750 MHz | 400 MB | ens1371 | 22.7 µs |
| marie | lenny, Xorg | Intel | 630 MHz | 512 MB | snd_hda_intel | 0,16 µs |
| masech | etch, Xorg | Intel | 2,8 GHz | 2 GB | snd_intel | 11,8 µs |
| tycho | etch, RT-lsm, Xorg | Intel | 1,73 GHz | 1 GB | snd_hda_intel | 4,5 µs |
Messprotokolle
- brutejack (Xonar):
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -n JackServer0 -dhw:1 -r32000 -p32 -n3: 29.6 µs, 24.2 µs, 25.8 µs, 27.7 µs, 26.3 µs; 01.08.2008 - brutejack (Intel):
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -n JackServer1 -dhw:0 -r32000 -p128 -n3: 5.4 µs, 6.9 µs, 9.6 µs, 4.4 µs, 32.5 µs; 01.08.2008 - jack:
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r32000 -p32 -n2: Ergebnisse: 22.75 µs, 22.5 µs, 22.72µs; 14.07.2008 - johannes (Messrechner 1): oom-Error; 14.07.2008
- heike (Messrechner 3): –low:
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r22050 -p64 -n2: Ergebnisse: 8.5 µs, 10.35 µs, 6.92 µs, 10.92 µs, 12.5 µs; 23.07.2008 - marie (Asus eeePC):
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r44100 -p1024 -n3; 0.13 µs, 0.15 µs, 0.19 µs, 0.17 µs, 0.17; 23.07.2008 - masech:
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r32000 -p32 -n2: Ergebnisse: 13.0 µs, 10.6 µs, 12.9 µs, 10.5 µs; 09.10.2008 - tycho:
/usr/bin/chrt 99 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r32000 -p128 -n3: Ergebnisse: 4.3 µs, 3.8 µs, 5.4 µs; 14.07.2008
Abhänigkeit der maximalen Verzögerung von der Puffergröße
Der jackd startet auf tycho mit /usr/bin/chrt 89 /usr/bin/jackd -R -P89 -t1000 -dalsa -dhw:0 -r32000 -p128 -n3.
Messdaten
| FrameSize | MaxDelay pro Sample | Jack-Latenz |
|---|---|---|
| 32 |  | 3 ms |
| 64 | | 6 ms |
| 128 | 1,67 µs | 12 ms |
| 256 | 1,71 µs | 24 ms |
| 512 | 1,7 µs | 48 ms |
| 1024 | 1,53 µs | 96 ms |
| 2048 | 0,62 µs | 192 ms |
Auswertung
Diagramm
set terminal dumb set title 'max Delay [µs] over Buffersize' unset key set ylabel 'maxDelay pro Sample [µs]'; set xlabel 'BufferSize'; plot '-' with linespoints; 128 1.67 256 1.71 512 1.7 1024 1.53 2048 0.62 e
max Delay [µs] over Buffersize
1.8 +-----------------------------------------+
| *A****A****+ + + + + + + |
| A* **** |
1.6 |-+ * +-|
| A** |
| * |
1.4 |-+ ** +-|
| * |
| ** |
1.2 |-+ * +-|
| * |
1 |-+ ** +-|
| * |
| ** |
0.8 |-+ * +-|
| ** |
| + + + + + + + + + + |
0.6 +-----------------------------------------+
0 200 40 600 800 100 120 140 16 180 200 2200
BufferSize
halblog. Diagramm
Verwendung von set logscale x;
set terminal dumb set title 'max Delay [µs] over Buffersize' unset key set logscale x; set ylabel 'maxDelay pro Sample [µs]'; set xlabel 'BufferSize'; plot '-' with linespoints; 128 1.67 256 1.71 512 1.7 1024 1.53 2048 0.62 e
max Delay [µs] over Buffersize
1.8 +-----------------------------------------+
| *****A*********A***** + |
|***** **** |
1.6 |-+ * +-|
| A |
| * |
1.4 |-+ * +-|
| * |
| * |
1.2 |-+ * +-|
| * |
1 |-+ * +-|
| * |
| * |
0.8 |-+ *-|
| * |
| + *|
0.6 +-----------------------------------------+
1000
BufferSize