Some tactics do not have sufficient workspace memory to run

[JsBlueCat]

Some tactics do not have sufficient workspace memory to run

when using yolov4 in jetson NX1

already have set the workspace-size = 4000

[marcoslucianops]

Edit /opt/nvidia/deepstream/deepstream-5.0/sources/libs/nvdsinfer/nvdsinfer_model_builder.h file

static const size_t kWorkSpaceSize = 450 * 1024 * 1024; // 450MB

To high value based on the available GPU memory.

See intructions in README file (in folder) to compile it again.

If you get Cuda Error in allocate: 2 (out of memory) error, try decreasing the value until it works

Set workspace-size (in config_infer_primary.txt file) to same value you defined in nvdsinfer_model_builder file

In my tests, these steps don't increase performance

[JsBlueCat]

Edit /opt/nvidia/deepstream/deepstream-5.0/sources/libs/nvdsinfer/nvdsinfer_model_builder.h file

static const size_t kWorkSpaceSize = 450 * 1024 * 1024; // 450MB

To high value based on the available GPU memory.

See intructions in README file (in folder) to compile it again.

If you get Cuda Error in allocate: 2 (out of memory) error, try decreasing the value until it works

Set workspace-size (in config_infer_primary.txt file) to same value you defined in nvdsinfer_model_builder file

In my tests, these steps don't increase performance

thanks , solved!