@lemire @geofflangdale happy to report the Arm64 Neon port of #simdjson to Rust is nearly done! Tests are *almost* all green, just one edge case with carryless multiply (sign ext?) in computing the quote mask. Do you have any tips or ideas? Cc @heinz_gieshttps://github.com/Licenser/simdjson-rs/pull/35#issuecomment-521252269 …
-
-
Can you point specifically at the offending ARM64 code in Rust?
2 replies 0 retweets 1 like -
https://github.com/Licenser/simdjson-rs/pull/35 … is where we’re stuck at the moment. I think the current theory is that the detection of backslash in in quotes (so “\”” ignores the backslash and ends the string early).
@sunnygleason has made some nice visualisation of this a few comments up the issue”1 reply 0 retweets 1 like -
Well, you got the portable compute_quote_mask, and that’s definitively portable so it ought to work everywhere.
1 reply 0 retweets 1 like -
You’re right - the issue was from the add_overflow implementation only writing when there was a carry. The main test suite is totally clear, just a few more bugs to clean up in the property-based testing. Thank you so much for your help!https://github.com/Licenser/simdjson-rs/commit/42dcf83b3083c59bd176128da9c591e16b12c353 …
1 reply 0 retweets 2 likes -
You found it! :D
2 replies 0 retweets 1 like -
-
Replying to @darachennis @heinz_gies and
Not sure if it’d be interesting to anyone else, but I’ve been thinking about doing a WASM port next
2 replies 0 retweets 2 likes -
Replying to @sunnygleason @darachennis and
Shouldn’t the rust version compile to wasm just fine or am I missing something?
1 reply 0 retweets 1 like -
Replying to @heinz_gies @darachennis and
SIMD-wasm!
https://github.com/WebAssembly/simd …2 replies 0 retweets 2 likes
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

