qkeras.qtools.qenergy.qenergy

Calculate energy consumption of a given quantized model.

Functions

`energy_estimate`(model, layer_map, ...)	estimate energy.
`get_op_type`(quantizer)
`memory_read_energy`(is_input_layer, ...[, ...])	compute energy to bring tensors from DRAM to SRAM.
`memory_write_energy`(is_output_layer, ...)	compute energy to bring tensors from SRAM to DRAM.
`parameter_read_energy`(layer, layer_item, ...)	read weights/bias from memory.
`to_scalar`(x)

qkeras.qtools.qenergy.qenergy.energy_estimate(model, layer_map, weights_on_memory, activations_on_memory, min_sram_size, rd_wr_on_io)[source]: estimate energy.

qkeras.qtools.qenergy.qenergy.memory_read_energy(is_input_layer, tensor_shape, mode, min_sram_size, rd_wr_on_io, quantizer_bits, is_tensor=True)[source]: compute energy to bring tensors from DRAM to SRAM.

qkeras.qtools.qenergy.qenergy.memory_write_energy(is_output_layer, tensor_shape, mode, min_sram_size, rd_wr_on_io, quantizer_bits)[source]: compute energy to bring tensors from SRAM to DRAM.

qkeras.qtools.qenergy.qenergy.parameter_read_energy(layer, layer_item, weights_on_memory, min_sram_size, rd_wr_on_io)[source]: read weights/bias from memory.