[Hackathon] Stabilize oneTBB Example Benchmarks 2

[dnmokhov]

Summary

Several oneTBB examples serve as performance benchmarks for the library. It's recommended to check the relative error for these examples. If the relative error exceeds 5%, consider stabilizing the example to improve the reproducibility of results, while still maintaining a reasonable execution time.

Additional task
Conduct scalability analysis and see if the examples are scalable in the given HW. See modifying the input parameters like dataSize improve scalability.

Target Examples

seismic : https://github.com/uxlfoundation/oneTBB/tree/master/examples/parallel_for/seismic
can be considered as reference for the usage of a utility class for Relative Error compute.

oneTBB/examples/common/utility/utility.hpp

Line 362 in af1a128

class measurements {

primes
sudoku
tachyon

[golowanow]

oneTBB_team2 is working on this (Intel OSS Hackathon 2025)

[golowanow]

#1669 is ready for review with instrumentations added to 'primes', 'sudoku', and 'tachyon' examples.
Next step is to analyze performance statistics from these.