ConvHx4Relu< INP_H, INP_W, OUT_W, OUT_W_PAD, STEP_H, STEP_W, B, C, M, KH, KW, GROUP, IS_RELU > Class Template Reference

Vector implementation for 3x3 BCHW, stores weights and biases, requires KW<=4, INP_W%4==0, OUT_W_PAD%8=0, STEP_H==1, STEP_W==1, GROUP==1, assumes weights are padded to MxCx12, ConvHx4Relu<26,28,24,24,1,1,1,2,4,3,3,1,1> total = 10603.

#include <conv.h>

Public Member Functions
	ConvHx4Relu (float(&w)[M *CKK_ROW_SIZE], float(&b)[M])
void	filter (input_window< float > *in, output_window< float > *out)

Static Public Member Functions
static void	registerKernelClass ()

---

The documentation for this class was generated from the following files:

• design/aie_src/conv.h
• design/aie_src/conv.cc