Using ap_fixed for convolution
sktalha68 opened this issue · comments
Can we use ap_fixed datatypes for weights, inputs and output streams? The function by default only accept ap_uint.
I assume you're planning to use the hlslib function standalone without relying on the FINN compiler?
Then you are free to use whatever data interpretation you might choose. The ap_uint
type is merely a transport container. The semantics are defined by the interpreting functor types specified as the template parameters TSrcI
, TDstI
and TWeightI
.