Update on Taskflow's benchmarks

[tsung-wei-huang]

Hi @ConorWilliams , thank you for the great library! It's nice to see the comparison between libfork and Taskflow. In the recent release, Taskflow has suggested the implementation of recursive parallelism, such as fibonacci patterns, using tf::Runtime. The benchmark folder also includes the implementation of fibonacci, integrate, nqueen, primes, and skynet using tf::Runtime. Unless you need a subgraph with complex dependencies, choosing tf::Runtime is much better than tf::Subflow.

Please don't hesitate to let me know if there is anything I can help :)

[ConorWilliams]

Hi @tsung-wei-huang much love to you and Taskflow as well ❤ Great to hear that Taskflow has adopted a more efficient implementation for recursive parallelism. I'm planning a refresh of Libfork in V4 (pending wider C++23 module support) and I will update all the benchmarks then, I don't have the bandwidth (or compute) to do this sooner but I'll leave this issue up in the mean-time so people can see that things have improved.