Revise results pretty-printing#143
Conversation
|
dynamic sizing in action: ❯ zig build examples && ./zig-out/bin/shuffling_allocator
benchmark runs total time time/run (avg ± σ) (min ... max) p75 p99 p995
-------------------------------------------------------------------------------------------------------------------------------------------------
My Benchmark 1 64 3.69s 57.668ms ± 16.766ms (26.449ms ... 93.597ms) 70.557ms 93.597ms 93.597ms
My Benchmark 2 (tracking) 64 3.483s 54.43ms ± 16.342ms (28.76ms ... 91.848ms) 65.804ms 91.848ms 91.848ms
My Benchmark 2 (tracking) [MEMORY] 512B ± 0B (512B ... 512B) 512B 512B 512B
My Benchmark 3 (shuffling) 64 4.151s 64.864ms ± 16.543ms (33.672ms ... 107.314ms) 74.66ms 107.314ms 107.314ms
My Benchmark 4 (shuffling + track) 64 3.775s 58.995ms ± 19.485ms (31.037ms ... 114.873ms) 68.632ms 114.873ms 114.873ms
My Benchmark 4 (shuffling + track) [MEMORY] 512B ± 0B (512B ... 512B) 512B 512B 512B
zBench on revise-prettyprinting [!] via ↯ v0.16.0-dev.2694+74f361a5c took 16s
❯ zig build examples && ./zig-out/bin/basic
benchmark runs total time time/run (avg ± σ) (min ... max) p75 p99 p995
---------------------------------------------------------------------------------------------------------------------------
My Benchmark 229 2.017s 8.808ms ± 3.216ms (3.162ms ... 15.862ms) 11.16ms 15.824ms 15.842ms
zBench on revise-prettyprinting [!] via ↯ v0.16.0-dev.2694+74f361a5c took 4s
❯ zig build examples && ./zig-out/bin/sleep
benchmark runs total time time/run (avg ± σ) (min ... max) p75 p99 p995
------------------------------------------------------------------------------------------------------------------------------
Sleep Benchmark 19 1.901s 100.091ms ± 8.106us (100.084ms ... 100.118ms) 100.094ms 100.118ms 100.118ms |
|
@hendriknielaender after some minor modification, this code still works with the latest Zig-master, do you think we should merge it? I noted a new issue with the shuffling allocator example and updated #141 accordingly. |
hendriknielaender
left a comment
hendriknielaender
left a comment
There was a problem hiding this comment.
Cool stuff 👍 yeah totally agree here. Lets merge this one.
@FObersteiner can you open seperate issues for it (e.g. the failing test for the shuffling allocator.)
OK I have made bug report issues for the examples that currently fail. Memory tracking seems to work fine in principle; I made a small change to the example file to clarify what should be expected here. |
2 participants
'D'for integer nanoseconds to duration is gone from Zig masterstd.Io.Durationhas a formatter method, but that does not support alignment + padding (yet)I re-built padding + alignment as it is done for other types in std. And since I was at it, I added dynamic adjustment of the pretty printing to the name length of the benchmarks. Overall there might be a more elegant solution but at least it makes the output a bit more flexible.
Resolves #130 - although it is not exactly configurable...