SW Acceleration: Embedded vs non-embedded performance
I'm interested in the Zynq UltraScale+ MPSoc (or better). My goal is to accelerate a piece of c++ software that I have. Have seen the Alveo accelerator card, but don't really understand how the acceleration is achieved, dont know which product to choose (with shortest time too market and best acceleration)
I have a background in SW, but know absolutely nothing of hardware.
My main question is, before I embark into this world of embedded platforms:
Q1. Will my software run faster and more deterministically when embedded or is it best to just leave it running on my linux server?
Q2. Which of these two platform is best suited for the acceleration job (without the usual fpga vhdl/verilog dev stuff)?
apologies for this very newbie question, but couldn't find an answer elsewhere.