Hello,
I'd like to design parallel multicore to assign 16x8 to one core as like the below.
I am using AWE6 engine and AWD_6.18.2_Standard.
The design is building(Generate Target Files..) error with message.
Could you review the error message?
1) Top level with two MulticoreSubsystem
2) MulticoreSubsystem1 -16x8 Mixer and Deinterleave
3) MulticoreSubsystem2 - 16x8 Mixer
Thanks,
Jackey
10:22am
Hi Jackey,
Can you successfully run the layout, or do you only get an error message when generating target files?
-Andrew
10:33am
Hi Andrew,
Thanks for your support.
1)Can you successfully run the layout,
[Jackey] The aws file is created but it is not working.
2)do you only get an error message when generating target files?
[Jackey] Yes, the error message is only as below. Both PC and target device have the same error.
Thanks,
Jackey
11:20am
Hi Jackey,
Looks like this is a known bug with multicore, but we have a workaround until the next release. The workaround is to use two multicore subsystems each with a single input pin. Pass all 16 channels to your first subsystem, deinterleave them and process the first 8 with the mixers. Re-interleave the 8 processed channels with the 8 additional NOT processed channels. You should now have 16 channels out of your first subsystem, where the first 8 are processed. Now, route all 16 channels from the first subsystem into the second subsystem. In the second subsystem, deinterleave the 16 channels and process the LAST 8 of them (the channels that were not processed by subsystem 1). Reinterleave them back together, and now you have 16 processed channels (the first 8 are processed in subsystem 1, and the last 8 in subsystem 2).
If this is unclear, I can email you an example .awd file.
Thanks!
11:41am
Hi Anderw,
Thanks for your information.
I will try it as your explanation.
Can you send me an example.awd file by email?
Also, I'd like to get Multicore document and design guide document.
Thanks,
Jackey
11:45am
Hi Jackey, I will email you the layout now.
12:02pm
Hi Anderw,
Thank for your support and email.
I will test your design and get back the result to you.
Thanks,
Jackey
12:04pm
No problem Jackey!
Let us know how it goes.
-Andrew
1:59am
Hi Adrew,
Thanks for your support.
The example.awd file is working well on linux target.
I have an question.
I'd like to know how to run an audio weaver script for multicore on linux side?
Currently I am using the following taskset -c2,3 command for imx6q multcore(cpu2, cpu3) according to the guide document.
taskset -c 2,3 /Audioweaver/AWE_command_line_vfp -nocmd -norate -script:xeno_core_21x16_mixer_rel_v01.aws -ml_port:15001 -cl_port:15003 -ut_port:15005
Is this command line right?
Thanks,
Jackey
11:51am
Hi Jackey,
The use of 'taskset' is unnecessary because our software is designed to handle assigning threads to multiple cores. Generally, it will assign threads to the least currently busy core, and so flexibly adapts to core usage in real time.
Try this,
./AWE_command_line_vfp -nocmd -norate -script:xeno_core_21x16_mixer_
Thanks,
Andrew
2:07am
Hi Andrew,
Thanks for your support .
I ran your command line with AWE6 engine on linux system(imx6q CPU + 3.14.38-rt36).
But your command line has an xrun as soon as AWE6 run.
Could you check how to run AWE6 on linux system?
Thanks,
Jackey
3:39pm
Hi Jackey,
Usually xruns are related to an issue with real time audio. If you use a simple, single core layout do you still get xruns?