If you can run your simulation (say, with VCS or NC-sim) on 100 CPUs, that's acceleration; If you run your system on a FPGA board, that's emulation. So I think the difference is how you speed up your simulation, by software means that's acceleration, by hardware means that's emulation.
I dont think using multiple cpus can only be called accleration. Porting the dut in fpga / acclerator will also save the same purpose. IMO the main difference between accleration and emulation is who is controlling the dut ,In case of accleration the dut is controlled software the clock_generation will happen in software , In case of emulation the clock generation will happen in hardware . With emulation we normally use signal genrators , logic analysers , or even silicon protos for application of stimuli and system building
Emulation system are specially designed hardware and software systems that typically contain reconfigurable logic, often FPGA. ...
Hardware acceleration maps some or all of the components in a software simulation into a hardware platform specifically designed to speed up certain simulation operations. ...