Kleis Auke Wolthuizen

> Hopefully this should be fixed by #452 which will be included in https://github.com/rust-lang/rust/pull/93696. Please try it in the next nightly once that is merged. > _Originally posted by @Amanieu...

It may also be caused by https://github.com/libvips/libvips/discussions/2047. If you have issues with that, you could disable the vector path with the `VIPS_NOVECTOR=1` environment variable or you could use this in...

After applying these patches (to fix various runtime errors): Details ``` LoadError: /usr/lib64/gems/ruby/rmagick-2.16.0/RMagick2.so: undefined symbol: R__Bool_to_C__Bool - /usr/lib64/gems/ruby/rmagick-2.16.0/RMagick2.so ``` ```diff --- a/Gemfile +++ b/Gemfile @@ -1,7 +1,7 @@ source "https://rubygems.org"...

I tried to port sharp's dHash function to this library: https://github.com/lovell/sharp/blob/985e881e7a764d9cb317938b36ec1f118808787b/test/fixtures/index.js#L12-L36 https://github.com/lovell/sharp/blob/985e881e7a764d9cb317938b36ec1f118808787b/src/operations.cc#L57-L93 See for example this patch: Details ```diff --- a/lib/dhash-vips.rb +++ b/lib/dhash-vips.rb @@ -72,6 +72,51 @@ module DHashVips ((a...

Great! PR https://github.com/libvips/libvips/pull/1769 does break the 1.000 score of IDHash. ``` $ rake benchmark ruby 3.0.2p107 (2021-07-07 revision 0db68f0233) [x86_64-linux] vips-8.13.0 Version: ImageMagick 6.9.12-45 Q16 x86_64 17184 https://imagemagick.org model name...

Increasing the reducing gap here to `3.0`: https://github.com/kleisauke/libvips/blob/1a219366a0a667f48510afbd2d622ad4279268b4/libvips/resample/resize.c#L160-L163 (which would ensure that always the final 300% is done by `vips_reduce{h,v}`) Produces: ```bash $ rake benchmark ruby 3.0.2p107 (2021-07-07 revision 0db68f0233)...

Here's a list of completed functions with their corresponding commits: - [X] vaddhn_s32 (commit https://github.com/simd-everywhere/simde/commit/e9ee0666356a60f28f5be248cf4de37be24e4a95) - [X] vld1q_dup_s16 (commit https://github.com/simd-everywhere/simde/commit/650d5310baec682d9c5545d668554b8791b93a96) - [X] vld1q_lane_s16 (commit https://github.com/simd-everywhere/simde/commit/9051a51c20c077f9a76be1ddf3c217e9bb9ad845) - [X] vld2_u8 (commit https://github.com/simd-everywhere/simde/commit/85d2ed2449992c5897bb9c01977fc7f060bbcd7c)...

Great, thanks! I'll re-run the benchmark in [`test/bench`](https://github.com/kleisauke/wasm-vips/tree/master/test/bench) within Chrome/Firefox once this is done. For Node.js, this requires an update of V8 to 9.1 (https://github.com/nodejs/node/pull/38273) to match the renumbered/finalized WASM...

`vqshluq_n_s16` was implemented with commit https://github.com/simd-everywhere/simde/commit/77af9f12e93eacd5cf107faaf7c244d46c5c167f, which makes it possible to compile libjpeg-turbo for WebAssembly with SIMD support (by reusing the Arm Neon intrinsics, see commit https://github.com/kleisauke/wasm-vips/commit/acd4c8128bcb195fed8724e82c41f93014aea30d). :tada: I'll re-run...

First set of benchmarking/profiling results can be found here: [`test/bench/README-simde.md`](https://github.com/kleisauke/wasm-vips/blob/libjpeg-turbo-simd/test/bench/README-simde.md). It seems that reusing the Arm Neon intrinsics for WASM made it ~3.5x slower than its C implementation (on this...