[Question] Dhrystone benchmark -> 2 Core variant

[frdtgzu]

Hello!

I am (still (throwback to about a year ago when i posted the og question about dhrystone being added back)) testing softcore cpu's (now with a little bit more knowledge), and am just wondering how hard (if it would be at all possible) would it be to make a "port" of dhrystone/any benchmark utilising both cores in the 2 core variant of this CPU? :)
I think the data from this would be very useful in generalising improvements from multi-core vs single-core preformance.

Thanks for the anwsers in advance.

[stnolting]

The problem is that you have to divide the work load (evenly) between the two cores. CoreMark already has a mechanism for this, but Dhrystone does not as far as I know.

neorv32/sw/example/coremark/include/core_portme.h

Lines 148 to 170 in 1f17db9

	/* Configuration : MULTITHREAD
	Define for parallel execution
	Valid values :
	1 - only one context (default).
	N>1 - will execute N copies in parallel.
	Note :
	If this flag is defined to more then 1, an implementation for launching
	parallel contexts must be defined.
	Two sample implementations are provided. Use <USE_PTHREAD> or <USE_FORK>
	to enable them.
	It is valid to have a different implementation of <core_start_parallel>
	and <core_end_parallel> in <core_portme.c>, to fit a particular architecture.
	*/
	#ifndef MULTITHREAD
	#define MULTITHREAD 1
	#define USE_PTHREAD 0
	#define USE_FORK 0
	#define USE_SOCKET 0
	#endif

As a first approximation, you could simply run two instances of Dhrystone in parallel (one per core) and then calculate the average runtime. This would at least give a rough measure of parallel processing (seen from an architectural point of view).