munrocket

> I vote for something much shorter, like dec128.mul or d128.mul, otherwise it may look too bulky. > why do you use Decimal.cmp not Decimal.compare if you have long names...

Also exist [rayon-hash](https://github.com/rayon-rs/rayon-hash) with different benchmarks of [concurrent hash tables](https://en.wikipedia.org/wiki/Concurrent_hash_table) that can be placed in SAB's theoretically.

According to 3dgs feedback subgroups not useful without barriers. What if we will leave HLSL broken for 2D/3D workgroups? At least will work for VR backends 🤔

Ben Clayton thank you for creating this issue. @jrprice it is the first thing that I tried. The only difference that I am passed this array into function. Sure that...

I can’t. This is a lazy function that was code-generated when you called it. :D

Not understand why we doesn't have invistigation with barriers ```js subgroupBarrier() subgroupMemoryBarrier() subgroupMemoryBarrierBuffer() subgroupMemoryBarrierShared() subgroupMemoryBarrierImage() ``` This topic rised 2 times in minutes. https://github.com/gpuweb/gpuweb/wiki/WGSL-2020-04-21 https://github.com/gpuweb/gpuweb/wiki/WGSL-2022-06-21-Minutes

I finded different package for goal, perhaps this will help you 1) Haxe https://github.com/HaxeFoundation/haxe 2) Pseudo https://github.com/alehander42/pseudo-python/

Here my example: without FMA error-free Dekker’s multiplication algorithm requires 17 floating-point operations. But if it will be available in WASM, it will be required only 2 FLOP. [1] This...

@yaffle also suggested the same syntax simplification ideas and even proposed further: the `D128` or `Dec128` class to the global context.

Awesome, even better.