search

HigherOrderCO / HVM

A massively parallel, optimal functional runtime in Rust
https://higherorderco.com
Apache License 2.0
10.52k stars 400 forks source link

Nicolas/wait threads #390

Closed developedby closed 5 months ago

HigherOrderBot commented 5 months ago

Perf run for dfba1b1:

compiled
========

file            runtime         main            (local)       
==============================================================
sort_bitonic    c                        4.40s           3.97s
                cuda                     0.23s           0.24s
--------------------------------------------------------------
sum_rec         c                        1.45s           2.06s
                cuda                     0.14s           0.15s
--------------------------------------------------------------
sum_tree        c                        0.12s           0.14s
                cuda                     0.10s           0.10s
--------------------------------------------------------------
tuples          c                        4.31s           3.97s
                cuda                   timeout         timeout
--------------------------------------------------------------

interpreted
===========

file            runtime         main            (local)       
==============================================================
sort_bitonic    c                        3.39s           3.60s
                cuda                     0.25s           0.24s
                rust                   timeout         timeout
--------------------------------------------------------------
sum_rec         c                        2.02s           2.03s
                cuda                     0.15s           0.13s
                rust                    13.96s          14.18s
--------------------------------------------------------------
sum_tree        c                        0.23s           0.22s
                cuda                     0.09s           0.08s
                rust                     0.84s           0.87s
--------------------------------------------------------------
tuples          c                        2.49s           2.92s
                cuda                   timeout         timeout
                rust                     3.66s           3.64s
--------------------------------------------------------------