Manage the electronic circuit

With AutomationLabs you can use several processors, such as the central processing unit (CPU) or the graphics processing unit (GPU) for tuning models. Also, it is possible to use several cpu cores to tune the models in parallel during the hyperparameters optimization.

Central processing unit

The list of tuned models with several cores for hyperparameters optimization are:

Architecture	Multi-threads CPU	Multi-processes CPU
Linear	[ ]	[ ]
Fnn	[x]	[x]
Rbf	[x]	[x]
Icnn	[x]	[x]
ResNet	[x]	[x]
PolyNet	[x]	[x]
DenseNet	[x]	[x]
NeuralNetODE_type1	[x]	[x]
NeulraNetODE_type2	[x]	[x]
Rnn	[x]	[x]
Lstm	[x]	[x]
Gru	[x]	[x]

where, [ ] not several cores and [x] several cores.

In order to tune a model with multi-threads, the computation_processor parameter is modified, such as:

~> julia --threads 4

julia-terminal-threads

julia> model(:tune; 
          project_name = "QTP_test",
          model_name = "model_qtp_nonlinear",
          io = "io_qtp",
          computation_solver = "radam",
          computation_maximum_time = Dates.Minute(15),
          model_architecture = "fnn", 
          computation_processor = "cpu_threads",
          )

In order to tune a model with multi-processes, the computation_processor parameter is modified, such as:

~> julia -p 4

julia-terminal-processes

julia> model(:tune; 
          project_name = "QTP_test",
          model_name = "model_qtp_nonlinear",
          io = "io_qtp",
          computation_solver = "RADAM",
          computation_maximum_time = Dates.Minute(15),
          model_architecture = "Fnn", 
          computation_processor = "CPUProcesses",
          )

By default the AutomationLabs tool uses single thread computing.

Graphics processing unit

By default the AutomationLabs tool will use the graphics processing unit calculation if it is available and CUDA.jl is installed and running on the machine. However, the GPU capability depends upon the blackbox model and the algorithm chosen to tune the weights.The list of tuned models with GPU capabilities are:

Architecture	GPU
Linear	[ ]
Fnn	[x]
Rbf	[x]
Icnn	[x]
ResNet	[x]
PolyNet	[x]
DenseNet	[x]
NeuralNetODE_type1	[x]
NeulraNetODE_type2	[x]
Rnn	[x]
Lstm	[x]
Gru	[x]

where, [ ] not gpu and [x] gpu.