| View previous topic :: View next topic |
| Author |
Message |
mbncp
Junior Jammer
Joined: 10 Jun 2006
Posts: 16
|
 Posted: Sat Nov 22, 2008 3:20 pm Post subject: Latency problems |
 |
|
Lately I added JS2 in my live setup (among other things) and since I have a bunch of crackles.
To see which plug could cause a problem I tried to debug with a tool I made some time ago, which records the time taken by a plug for each process.
When I test it with JS2 I found some strange result, once in a while JS2 seem to take a long time ( up to 10 msec), so I'm not suprised that I have pops in that case.
Here's an example of the results I get:
min:4812 max:10550378 count:10000 avg:14948
and some results (process ID, time in micro seconds):
6697 10550378
8135 133144
4632 77413
7116 69952
1601 69656
...
339 4864
833 4864
449 4841
417 4825
759 4812
Highest and lowest times out of 10000 consecutive buffer calls (64 samples @ 44.1 kHz).
I mainly tested in reaper but I had similar results in bidule standalone.
I tried pretty much anything, midi only, with and without GUI open, single core or multicore (4x), playing or stopped,...
Though I was having different results, in all situations I had big differences between the shortest and longest time.
I tried with some different plugs and didn't see that happen.
Anyway, when I disable JS2 I don't get the crackles anymore 
Any idea what this could be ? I tried today with 2.5.0, but I get the same problem. |
|
| Back to top |
|
 |
Ralph [RZ]
Site Admin
Joined: 02 Jan 2008
Posts: 13332
|
| Posted: Mon Nov 24, 2008 12:03 pm Post subject: |
|
|
When Jamstix is composing it will require more CPU, especially around the beginning of a bar. Raising your latency to 128 samples (3ms) should fix the crackles.
_________________
Ralph Zeuner
Rayzoon Technologies LLC
http://www.rayzoon.com |
|
| Back to top |
|
|
mbncp
Junior Jammer
Joined: 10 Jun 2006
Posts: 16
|
| Posted: Mon Nov 24, 2008 1:21 pm Post subject: |
|
|
| Ralph [RZ] wrote: | | When Jamstix is composing it will require more CPU, especially around the beginning of a bar. Raising your latency to 128 samples (3ms) should fix the crackles. |
I ploted a graph and indeed I can see a higher CPU usage at given intervals. But normally this is fine as it's around 0.1-0.2 msec, 10% of what a single core can do in my 64 samples buffer.
The problem is these 9-10 msec that I get once in a while, and there, even a 512 samples buffer wouldn't be enought.
Actually I increased my tool plug to 100'000 processes call, and then I get the 10 msec 1-2 times, and it doesn't seem to be at a regular interval.
Is JS2 making disk access once in a while, or huge ram allocation ,.. ? |
|
| Back to top |
|
|
Ralph [RZ]
Site Admin
Joined: 02 Jan 2008
Posts: 13332
|
| Posted: Mon Nov 24, 2008 2:00 pm Post subject: |
|
|
I have just sent you a test Build to track this down.
_________________
Ralph Zeuner
Rayzoon Technologies LLC
http://www.rayzoon.com |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
