diff options
| author | Ben Noordhuis <info@bnoordhuis.nl> | 2021-01-02 11:27:16 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2021-01-02 11:27:16 +0100 |
| commit | 88855b5d95be04747ec0ff67d8b33dfde3189641 (patch) | |
| tree | a44fcb6e3b49e21770c11cd9d4fe0a6edc1c48f6 /std/encoding | |
| parent | 012f99bd9a0463813653bf8438e09b3247395984 (diff) | |
fix(test_util): make hyper server single-threaded (#8951)
Use Tokio's single-threaded scheduler. The hyper server is used as a
point of comparison for the (single-threaded!) benchmarks in cli/bench.
We're not comparing apples to apples if we use the default
multi-threaded scheduler.
This drops the requests/sec from 284k to 130k on my 12 core Ryzen 5
system. That still leaves a 50k gap for us to close. Working on it!
Diffstat (limited to 'std/encoding')
0 files changed, 0 insertions, 0 deletions