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

Scalar implementation for BCHW, stores weights and biases, requires GROUP==1, ConvReluScalar<28,28,24,24,1,1,1,2,4,5,5,1,1> total = 368812 ConvReluScalar<24,24,10,10,2,2,1,2,4,5,5,1,1> total = 64413 ConvReluScalar<26,26,24,24,1,1,1,2,4,3,3,1,1> total = 159087.

#include <conv.h>

Public Member Functions
	ConvReluScalar (float(&w)[M *KH *KW *C], 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