This layer is a tech support trick box that is designed to help diagnose causes of functional and performance issues in applications. It works by letting you quickly test your application with a set of API behavior overrides applied, which can help to identify likely problem areas in the application if an override causes an issue to disappear.
This layer requires you to provide a configuration file that specifies the overrides to apply. Details of the configuration options in each override group are document in the Behavior overrides section below.
The layer_config.json
file in this directory is a
complete template configuration file you can start from. However, note that it
does not enable any overrides by default, so running the layer with this
configuration used "as is" will not do anything useful. Take a copy and
modify it to enable the options you want to try.
For Android you can run this layer with the following command line:
python3 lgl_android_install.py --layer layer_gpu_support --config <your.json>
This layer is device agnostic and should support any GPU.
The current override groups are supported:
- Serialization: control serialization of GPU workload scheduling to diagnose issues caused by missing queue or command stream synchronization.
The serialization overrides allow forceful serialization of submitted workloads, ensuring that they run in queue submit order. The synchronization can be configured per workload type, allowing control over where serialization is added to command buffers and queues.
The configuration file allows control over specific options, as well as convenience options to force disable or enable all serialization.
- If the
none
option istrue
then no serialization is applied, irrespective of other settings. - Else, if the
all
option istrue
then all serialization is applied, irrespective of other settings. - Else, the individual options are applied as specified.
"serialize": {
"none": false, // Enable no serialization options
"all": false, // Enable all serialization options
"queue": false, // Enable cross-queue serialization of submits
"commandstream": {
"compute": {
"pre": false, // Insert full barrier before dispatches
"post": false // Insert full barrier after dispatches
},
"renderpass": {
"pre": false, // Insert full barrier before render passes
"post": false // Insert full barrier after render passes
},
"tracerays": {
"pre": false, // Insert full barrier before trace rays
"post": false // Insert full barrier after trace rays
},
"transfer": {
"pre": false, // Insert full barrier before transfers
"post": false // Insert full barrier after transfers
}
}
}
Copyright © 2025, Arm Limited and contributors.