WebAssembly vs. Native Code: Performance Analysis (WASM 45% Slower)

(ar5iv.labs.arxiv.org)
20 points birdculture | 6 comments | 01 Nov 25 15:33 UTC | HN request time: 1.135s | source | bottom
1. seg_lol ◴[01 Nov 25 20:50 UTC] No.45785220[source]▶
(2019) https://arxiv.org/abs/1901.09056

2022, two comments https://news.ycombinator.com/item?id=33054931

2019, 172 comments https://news.ycombinator.com/item?id=19023413

Given that this paper is from so long ago, running new benchmarks on existing runtimes would be productive.

https://mastodon.world/@cfallin/109833747279454645 mentions https://github.com/bytecodealliance/sightglass

https://www.usenix.org/conference/atc19/presentation/jangda

https://www.usenix.org/sites/default/files/conference/protec...

2. mdhb ◴[01 Nov 25 22:12 UTC] No.45785900[source]▶
A lot has changed since this was published, you can’t trust these numbers at all.
replies(1): >>45788908 #
3. yellowapple ◴[02 Nov 25 04:15 UTC] No.45787811[source]▶
Being within the same order of magnitude is pretty impressive IMO, especially before 6 years' worth of performance improvements.
4. blablabla123 ◴[02 Nov 25 09:07 UTC] No.45788908[source]▶
True, that's from 2019. While benchmarks are rare, some actually claim near native speed
replies(1): >>45789179 #
5. solarkraft ◴[02 Nov 25 10:09 UTC] No.45789179{3}[source]▶
To be fair these (and similar) claims were already made when WASM was new.