I just got a RME Fireface UCX and will test it on linux soon. it has a USB class compliant mode, so it should run. The documentation says that the latency in the class compliant mode is not as good as with the proprietary driver. It would be interesting to have numbers here how much better exactly the rme driver is over the class compliant. I guess i'd simply connect output to input and measure the delay? Someone must have done this before, Is there a standard method? A patch?
I must say that I am a bit disappointed about the drivers for os x. I installed the firewire and USB drivers and updated the firmware, with the result that the settings app now crashes and the firmware update also quits unsuccessfully now. in the regular mode now all the led on the device are on and the sound is chpped up. USB class compliant mode however still works. i mailed the support. Disappointing for a device i just spend >1000 eur on. also the rme fireface driver is the only process which isn't 64bit that is loaded by the os.
Am 18.10.2012 um 14:24 schrieb Peter Venus news@petervenus.de:
a quick look on ffado.org brought up, that RMEs fireface 400 and 800 have now the "full support" status. i haven´t tested this yet, but if this works, this would be a perfect solution i think. anyone checked this yet? cheers, peter
Gesendet: Mittwoch, 17. Oktober 2012 um 23:33 Uhr Von: "Ed Kelly" morph_2016@yahoo.co.uk An: "Rob Bothof" rbothof@xs4all.nl, "pd-list@iem.at" pd-list@iem.at Betreff: Re: [PD] firewire is dead? Jack + fa101? I have never, ever got this to give me anything other than glitchy, unusable audio!
actually it was still called smartcard.. either how i use it still with edirol fa101 never failed me, rocksolid performance with jack
On 10/16/2012 6:11 PM, yvan volochine wrote:
On 10/16/2012 02:45 PM, patrick wrote:
good, i was a little bit worried about using a pci-e firewire adapter for audio in linux / jack.
about latency, yesterday i used jdelay to measure the round-trip of my latency, at 44100 with 128, 3 = 17 ms (jack reports 8 ms). at 96000 i get 7 ms, BUT my cpu seems to be working hard and i get glitches. so the question is:
with a faster cpu, can i lower the latency?
yes =)
i.e. I have to use huge buffer sizes (so more latency) because of my old Intel Core Duo 1.6Ghz (great thinkpad x60 with expresscard slot for my FW card though)...
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
- Max abonnements@revolwear.com [2012-11-19 09:05]:
I just got a RME Fireface UCX and will test it on linux soon. it has a USB class compliant mode, so it should run. The documentation says that the latency in the class compliant mode is not as good as with the proprietary driver. It would be interesting to have numbers here how much better exactly the rme driver is over the class compliant. I guess i'd simply connect output to input and measure the delay? Someone must have done this before, Is there a standard method? A patch?
Use jdelay by Fons Adriaensen. http://kokkinizita.linuxaudio.org/linuxaudio/
There is also a patch: /usr/local/lib/pd/doc/7.stuff/tools/latency.pd
best, P
I must say that I am a bit disappointed about the drivers for os x. I installed the firewire and USB drivers and updated the firmware, with the result that the settings app now crashes and the firmware update also quits unsuccessfully now. in the regular mode now all the led on the device are on and the sound is chpped up. USB class compliant mode however still works. i mailed the support. Disappointing for a device i just spend >1000 eur on. also the rme fireface driver is the only process which isn't 64bit that is loaded by the os.
Am 18.10.2012 um 14:24 schrieb Peter Venus news@petervenus.de:
a quick look on ffado.org brought up, that RMEs fireface 400 and 800 have now the "full support" status. i haven´t tested this yet, but if this works, this would be a perfect solution i think. anyone checked this yet? cheers, peter
Gesendet: Mittwoch, 17. Oktober 2012 um 23:33 Uhr Von: "Ed Kelly" morph_2016@yahoo.co.uk An: "Rob Bothof" rbothof@xs4all.nl, "pd-list@iem.at" pd-list@iem.at Betreff: Re: [PD] firewire is dead? Jack + fa101? I have never, ever got this to give me anything other than glitchy, unusable audio!
actually it was still called smartcard.. either how i use it still with edirol fa101 never failed me, rocksolid performance with jack
On 10/16/2012 6:11 PM, yvan volochine wrote:
On 10/16/2012 02:45 PM, patrick wrote:
good, i was a little bit worried about using a pci-e firewire adapter for audio in linux / jack.
about latency, yesterday i used jdelay to measure the round-trip of my latency, at 44100 with 128, 3 = 17 ms (jack reports 8 ms). at 96000 i get 7 ms, BUT my cpu seems to be working hard and i get glitches. so the question is:
with a faster cpu, can i lower the latency?
yes =)
i.e. I have to use huge buffer sizes (so more latency) because of my old Intel Core Duo 1.6Ghz (great thinkpad x60 with expresscard slot for my FW card though)...
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
hi,
finally i got my new computer with an external firewire PCI-E TI chipset card (25$/CAD). it is working. my latency is now 8ms (round-trip measure with jack_iodelay). i am using ubuntu-studio that use the low-latency kernel by default.
à +
Am 19.11.2012 um 09:45 schrieb Peter P. p8rpp@aol.com:
- Max abonnements@revolwear.com [2012-11-19 09:05]:
I just got a RME Fireface UCX and will test it on linux soon. it has a USB class compliant mode, so it should run. The documentation says that the latency in the class compliant mode is not as good as with the proprietary driver. It would be interesting to have numbers here how much better exactly the rme driver is over the class compliant. I guess i'd simply connect output to input and measure the delay? Someone must have done this before, Is there a standard method? A patch?
Use jdelay by Fons Adriaensen. http://kokkinizita.linuxaudio.org/linuxaudio/
There is also a patch: /usr/local/lib/pd/doc/7.stuff/tools/latency.pd
Had no chance to test the Fireface UCX on linux yet, but here the results for OS X 10.8.2:
Tested with Pure Data 0.43.4-extended-20121101 /doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64, No callbacks
latency via USB, RME Driver print: 0.00193273 print: 0 print: 0 print: 0.000630273 print: 0 print: 0.000632601 print: 0.00253273 print: 0.000634696 print: 0.00190642 print: 0.000636792 print: 0.00191317 print: 0 print: 0.00191946 print: 0 print: 0.00192621 print: 0 print: 0.00193273 print: 0 print: 0 print: 0.000630273
latency via USB, Class Compliant Mode print: 24.3173 print: 24.3199 print: 24.3224 print: 24.3249 print: 24.3275 print: 24.33 print: 24.3326 print: 24.3352 print: 24.3377 print: 24.3403 print: 24.3429 print: 24.3454 print: 24.348 print: 24.3098 print: 24.3123
In the USB Class Compliant mode, when I change the delay in the Pd Audio properties to 40, the patch will output around 44.3. When set to 80 it will be 84.3 and so forth. Using the RME driver the latency shown in the patch always stays in the number range as seen above, no matter what delay is set in the Pd's Audio Properties. What's the explanation for this?
I have to send the Fireface back because the Firewire port is not working for me.
I got a new Fireface UCX, updated drivers and did a fresh test (replacement for the first one which didn't work). It has three modes: Firewire, USB and USB in Class Compliant (Plug and Play) mode. Interesting is, that Firewire is the worst when it comes to latency and Class Compliant is less than two milliseconds worse than the proprietary RME USB Driver. Bandwith might still be better in Firewire for multichannel setups, but I didn't test this. In my previous post on the topic the USB latency was wrong, it was a driver problem.
Tested with Pure Data 0.43.4-extended-20121101 /doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64, No callbacks
OS X 10.8.2 Mac Book Pro 2 Ghz i7
latency of internal laptop soundcard ("Intel High Definition Audio") for reference: print: 24.348 print: 24.3098 print: 24.3123 print: 24.3148 print: 24.3173 print: 24.3199 print: 24.3224 print: 24.3249 print: 24.3275 print: 24.33 print: 24.3326
latency Fireface UCX via Firewire, RME Driver print: 27.8751 print: 27.8777 print: 27.8803 print: 27.8829 print: 27.8854 print: 27.8472 print: 27.8497 print: 27.8522 print: 27.8548 print: 27.8573 print: 27.8598 print: 27.8624 print: 27.8649 print: 27.8675 print: 27.87 print: 27.8726
latency of Fireface UCX via USB, RME Driver print: 23.0024 print: 23.005 print: 23.0076 print: 23.0101 print: 22.9719 print: 22.9744 print: 22.9769 print: 22.9795 print: 22.982 print: 22.9845 print: 22.9871 print: 22.9896 print: 22.9922 print: 22.9947 print: 22.9973 print: 22.9998 print: 23.0024 print: 23.005
latency Fireface UCX via USB, Class Compliant Mode print: 24.3173 print: 24.3199 print: 24.3224 print: 24.3249 print: 24.3275 print: 24.33 print: 24.3326 print: 24.3352 print: 24.3377 print: 24.3403 print: 24.3429 print: 24.3454 print: 24.348 print: 24.3098 print: 24.3123
Hi Max,
Is it possible for you to test this same card in linux?
J
On Nov 26, 2012, at 9:25 AM, Max wrote:
I got a new Fireface UCX, updated drivers and did a fresh test (replacement for the first one which didn't work). It has three modes: Firewire, USB and USB in Class Compliant (Plug and Play) mode. Interesting is, that Firewire is the worst when it comes to latency and Class Compliant is less than two milliseconds worse than the proprietary RME USB Driver. Bandwith might still be better in Firewire for multichannel setups, but I didn't test this. In my previous post on the topic the USB latency was wrong, it was a driver problem.
Tested with Pure Data 0.43.4-extended-20121101 /doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64, No callbacks
OS X 10.8.2 Mac Book Pro 2 Ghz i7
latency of internal laptop soundcard ("Intel High Definition Audio") for reference: print: 24.348 print: 24.3098 print: 24.3123 print: 24.3148 print: 24.3173 print: 24.3199 print: 24.3224 print: 24.3249 print: 24.3275 print: 24.33 print: 24.3326
latency Fireface UCX via Firewire, RME Driver print: 27.8751 print: 27.8777 print: 27.8803 print: 27.8829 print: 27.8854 print: 27.8472 print: 27.8497 print: 27.8522 print: 27.8548 print: 27.8573 print: 27.8598 print: 27.8624 print: 27.8649 print: 27.8675 print: 27.87 print: 27.8726
latency of Fireface UCX via USB, RME Driver print: 23.0024 print: 23.005 print: 23.0076 print: 23.0101 print: 22.9719 print: 22.9744 print: 22.9769 print: 22.9795 print: 22.982 print: 22.9845 print: 22.9871 print: 22.9896 print: 22.9922 print: 22.9947 print: 22.9973 print: 22.9998 print: 23.0024 print: 23.005
latency Fireface UCX via USB, Class Compliant Mode print: 24.3173 print: 24.3199 print: 24.3224 print: 24.3249 print: 24.3275 print: 24.33 print: 24.3326 print: 24.3352 print: 24.3377 print: 24.3403 print: 24.3429 print: 24.3454 print: 24.348 print: 24.3098 print: 24.3123
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Am 26.11.2012 um 16:18 schrieb J Oliver jaime.oliver2@gmail.com:
Hi Max,
Is it possible for you to test this same card in linux?
J
I wanted to do this, but I don't have a linux machine with firewire, so i can't test the ffado driver for the card. I wanted to try out the class compliant mode, but there is a bug in ALSA which prevents it to work. see http://www.mail-archive.com/alsa-user@lists.sourceforge.net/msg28935.html I didn't wanted to get started recompiling the kernel for this. sorry. So in theory it should work, but only if you remove the bug in the ALSA part.
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037
Hmm - I'm getting latencies between 6 and 7 (Fedora 17, Core 7200, INTEL "HDA", latest Pd from git, but I think Pd 0.43 should do similar).
cheers Miller
On Mon, Nov 26, 2012 at 06:53:31PM +0100, Max wrote:
Am 26.11.2012 um 16:18 schrieb J Oliver jaime.oliver2@gmail.com:
Hi Max,
Is it possible for you to test this same card in linux?
J
I wanted to do this, but I don't have a linux machine with firewire, so i can't test the ffado driver for the card. I wanted to try out the class compliant mode, but there is a bug in ALSA which prevents it to work. see http://www.mail-archive.com/alsa-user@lists.sourceforge.net/msg28935.html I didn't wanted to get started recompiling the kernel for this. sorry. So in theory it should work, but only if you remove the bug in the ALSA part.
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037 _______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hi Miller,
Is this with the test tone patch?
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
I have also gotten the HDA way lower than 20 ms in ubuntu 12.04...!
best,
J
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037 _______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Am 27.11.2012 um 17:02 schrieb J Oliver jaime.oliver2@gmail.com:
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
I have also gotten the HDA way lower than 20 ms in ubuntu 12.04...!
for comparability I've set the delay time in Pd to 20. So that's a constant you should subtract from the latency measured by the latency.pd patch. The rest of ~1.9 ms is a little hard to believe.
you are right - those numbers seem to be to good to be true. On the other hand I followed the instructions in the latency patch and don't know what could have been wrong.
Am 26.11.2012 um 19:04 schrieb Miller Puckette msp@ucsd.edu:
Hmm - I'm getting latencies between 6 and 7 (Fedora 17, Core 7200, INTEL "HDA", latest Pd from git, but I think Pd 0.43 should do similar).
cheers Miller
On Mon, Nov 26, 2012 at 06:53:31PM +0100, Max wrote:
Am 26.11.2012 um 16:18 schrieb J Oliver jaime.oliver2@gmail.com:
Hi Max,
Is it possible for you to test this same card in linux?
J
I wanted to do this, but I don't have a linux machine with firewire, so i can't test the ffado driver for the card. I wanted to try out the class compliant mode, but there is a bug in ALSA which prevents it to work. see http://www.mail-archive.com/alsa-user@lists.sourceforge.net/msg28935.html I didn't wanted to get started recompiling the kernel for this. sorry. So in theory it should work, but only if you remove the bug in the ALSA part.
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037 _______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
If I'm reading rour post right, you're specifying 20 msec latency and getting about 22, which is OK, but I think you should be able to get lower latencies (i.e., I don't see that numbers like 21.9012 are too good to be true - those are typical Macintosh latencies but I think in linux you should be able to do much better.)
cheers Miller
On Tue, Nov 27, 2012 at 05:15:35PM +0100, Max wrote:
you are right - those numbers seem to be to good to be true. On the other hand I followed the instructions in the latency patch and don't know what could have been wrong.
Am 26.11.2012 um 19:04 schrieb Miller Puckette msp@ucsd.edu:
Hmm - I'm getting latencies between 6 and 7 (Fedora 17, Core 7200, INTEL "HDA", latest Pd from git, but I think Pd 0.43 should do similar).
cheers Miller
On Mon, Nov 26, 2012 at 06:53:31PM +0100, Max wrote:
Am 26.11.2012 um 16:18 schrieb J Oliver jaime.oliver2@gmail.com:
Hi Max,
Is it possible for you to test this same card in linux?
J
I wanted to do this, but I don't have a linux machine with firewire, so i can't test the ffado driver for the card. I wanted to try out the class compliant mode, but there is a bug in ALSA which prevents it to work. see http://www.mail-archive.com/alsa-user@lists.sourceforge.net/msg28935.html I didn't wanted to get started recompiling the kernel for this. sorry. So in theory it should work, but only if you remove the bug in the ALSA part.
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037 _______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
oh really? that's interesting. The later numbers come from a Ubuntu machine which isn't set up for sound at all, so when I compare those with a os x machine and the "famous" rme drivers the numbers are still impressive.
Am 27.11.2012 um 17:47 schrieb Miller Puckette msp@ucsd.edu:
If I'm reading rour post right, you're specifying 20 msec latency and getting about 22, which is OK, but I think you should be able to get lower latencies (i.e., I don't see that numbers like 21.9012 are too good to be true - those are typical Macintosh latencies but I think in linux you should be able to do much better.)
cheers Miller
On Tue, Nov 27, 2012 at 05:15:35PM +0100, Max wrote:
you are right - those numbers seem to be to good to be true. On the other hand I followed the instructions in the latency patch and don't know what could have been wrong.
Am 26.11.2012 um 19:04 schrieb Miller Puckette msp@ucsd.edu:
Hmm - I'm getting latencies between 6 and 7 (Fedora 17, Core 7200, INTEL "HDA", latest Pd from git, but I think Pd 0.43 should do similar).
cheers Miller
On Mon, Nov 26, 2012 at 06:53:31PM +0100, Max wrote:
Am 26.11.2012 um 16:18 schrieb J Oliver jaime.oliver2@gmail.com:
Hi Max,
Is it possible for you to test this same card in linux?
J
I wanted to do this, but I don't have a linux machine with firewire, so i can't test the ffado driver for the card. I wanted to try out the class compliant mode, but there is a bug in ALSA which prevents it to work. see http://www.mail-archive.com/alsa-user@lists.sourceforge.net/msg28935.html I didn't wanted to get started recompiling the kernel for this. sorry. So in theory it should work, but only if you remove the bug in the ALSA part.
for the record i compared the on board intel card of the linux machine, the result is quite impressive, i don't think it can get any better:
/doc/7.stuff/tools/latency.pd Sampling rate 44100 Hz, delay 20 ms Blocksize 64 Ubuntu 12.04 Intel® Core™ i3 CPU 550 @ 3.20GHz × 4 Latency HDA Intel (hardware)
print: 21.9012 print: 21.9037 print: 21.9063 print: 21.9088 print: 21.9114 print: 21.914 print: 21.9165 print: 21.9191 print: 21.9217 print: 21.8834 print: 21.886 print: 21.8885 print: 21.891 print: 21.8936 print: 21.8961 print: 21.8986 print: 21.9012 print: 21.9037 _______________________________________________ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-
J Oliver -
Max -
Miller Puckette -
patrick -
Peter P. -
Peter Venus