Required Info:
- Software Version
DRIVE OS 6.0.6 - Target OS
Linux - SDK Manager Version
1.9.2.10884 - Host Machine Version
native Ubuntu Linux 20.04 Host installed with DRIVE OS DOCKER Containers
Describe the bug
In the docs of dwcgf, I can see a lot of configurations which implies that there would be more than 1 SoC that running with dwcgf-app.
- the schedule of cgf can use global resources that with more than 1 tegra devices. see TegraB Compute Graph Framework SDK Reference: Demo Pipeline Introduction (nvidia.com)
{
"resources": {
"TegraA": {
"CPU": ["CPU0", "CPU1", "CPU2", "CPU3", "CPU4", "CPU5"],
"GPU": ["dGPU"]
},
"TegraB": {
}
},
...
}
- the process layer configuration of dwcgf-app can be more than 1 tegra devices, see machine0, machine0,
driverorks-5.10/src/cgf/graphs/descriptions/systems/CGFDemo.app.json
"processes": {
"ssm": {
"runOn": "machine0",
"executable": "SSM/vanillassm",
"logSpec": "file:{{logpath}}/{{appname}}.log"
},
"schedule_manager": {
"runOn": "machine0",
"executable": "ScheduleManager",
"argv": {
"--enableScheduleSwitching": "true",
"--scheduleManagerHostIP": "127.0.0.1",
"--scheduleManagerHostPort": "40100",
"--scheduleManagerNumClients": "5"
}
},
"sensor_sync_server": {
"runOn": "machine0",
"executable": "sensor_sync_server",
"logSpec": "file:{{logpath}}/{{appname}}.log",
"argv": {
"--syncServerHostPort": "40001",
"--cameraGroup": "0",
"--radarGroup": "0",
"--imuGpsCanGroup": "0"
}
},
"stm_master": {
"runOn": "machine0",
"executable": "stm_master",
"logSpec": "file:{{logpath}}/{{appname}}.log",
"argv": {
"--allow-unregistered-runnables": true,
"--timeout-us": "600000000",
"--log": "stm_sreport",
"--soc": "TegraA",
"-m": true,
"-v": true
}
},
"camera_pipeline0": {
"runOn": "machine0",
"executable": "LoaderLite",
"subcomponents": [
"cgfDemo.cameraPipelineFront0",
"cgfDemo.arender"
],
"argv": {
"--nodeInitThreadPoolSize": "7",
"--channelBindingThreadPoolSize": "7"
}
},
- the sample_cgf_dwchannel supports C2C communication Compute Graph Framework SDK Reference: CGF Channel Sample
To run C2C nvscistream PCIE broadcast:
On Dual Firespray Tegra A:
sudo ./sample_cgf_dwchannel --type=NVSCI --prod-stream-names=nvscic2c_pcie_s0_c6_1:nvscic2c_pcie_s0_c6_2 --prod-reaches=chip:chip
On Dual Firespray Tegra B:
sudo ./sample_cgf_dwchannel --type=NVSCI --cons-stream-names=nvscic2c_pcie_s0_c5_1:nvscic2c_pcie_s0_c5_2 --cons-reaches=chip:chip
qst1
What are these configs mean in the dwcgf app.json?
qst2
Can dwcgf run over 2 OrinDevkits in DriveSDK6.0.6?
There is an answer NO in Creating a custom hello world application and integration - Driveworks CGF - #21 by SivaRamaKrishnaNV
But I would like the double confirmation.
qst3
Will this feature be available in the future release of dwcgf?
qst4
When will this feature be availble, in which release of sdk?