Reels · 2 of 55
1:04
Reel · 1:04
I measured Wild against mold again and mold won: the culprit was not the new versions, it was btrfs
Narrated: I reran September's linker bench and mold beat Wild. The same ripgrep link on four filesystems of the same disk shows why: on btrfs mold wins, on ext4 and tmpfs Wild wins. Forcing mmap on Wild on btrfs cuts 18 to 37 % of its time. And in debug, rust-lld takes 37 s on btrfs and 1.9 s on ext4 for the same bevy link.
- Length
- 1:04
- Published
This reel sums up I measured Wild against mold again and mold won: the culprit was not the new versions, it was btrfs, where the method, the tables and what did not work are.