daw_json_link VS Thrust

They might have good luck with https://github.com/beached/daw_json_link it has support for stuff like JSON lines and alike plus other ways that only use as much ram as their underlying data structures do as it parses directly to the user DS. Plus it has an iterator/range interface for things like arrays if needed.

We will add more benchmarks in the future, but for now you can see the comparison of daw_json_link with rapidjson. glaze is faster than daw_json_link, which is over twice as fast as rapidjson.

I am biased, but prefer https://github.com/beached/daw_json_link Super fast and you work with your native data structures without the overhead of DOM parsing/lookup

You can also look into https://github.com/beached/daw_json_link which claims to support nullable values in the readme.

The library I author does this. https://github.com/beached/daw_json_link . It's fast, GB/s too, and provides the mapping mechanism, iteration types, json lines support, event based parser, along with a non-owning json_value for when the mappings don't fit right or if one is querying. Pretty much everything but an owning JSON value as it's not something I've ever needed more than temporarily and brings a lot of complexity that is solved by using the actual C++ data structures one is eventually parsing into anyways.

So DAW JSON Link does have a DOM view, however it does not have a owning view. The json_value(even supports JSON Path in a limited form) type and json_raw mappings can help here. But there is no hard line between parsing view the json_value and the mappings to concrete data structures. One can mix and match.

Is this the repo? Never saw it linked in our convo, and I’d like to give it a whirl.

daw_json_link is what you're looking for : https://github.com/beached/daw_json_link

Check out https://github.com/beached/daw_json_link , it provides a non-typeerased way to parse JSON straight into user-defined data structures.

this is frankly starting to sound a lot like the ridiculous "blue bubbles" discourse.

AMD's products have generally failed to catch traction because their implementations are halfassed and buggy and incomplete (despite promising more features, these are often paper features or career-oriented development from now-departed developers). all of the same "developer B" stuff from openGL really applies to openCL as well.

http://richg42.blogspot.com/2014/05/the-truth-on-opengl-driv...

AMD has left a trail of abandoned code and disappointed developers in their wake. These two repos are the same thing for AMD's ecosystem and NVIDIA's ecosystem, how do you think the support story compares?

https://github.com/HSA-Libraries/Bolt

https://github.com/NVIDIA/thrust

in the last few years they have (once again) dumped everything and started over, ROCm supported essentially no consumer cards and rotated support rapidly even in the CDNA world. It offers no binary compatibility support story, it has to be compiled for specific chips within a generation, not even just "RDNA3" but "Navi 31 specifically". Etc etc. And nobody with consumer cards could access it until like, six months ago, and that still is only on windows, consumer cards are not even supported on linux (!).

https://geohot.github.io/blog/jekyll/update/2023/06/07/a-div...

This is on top of the actual problems that still remain, as geohot found out. Installing ROCm is a several-hour process that will involve debugging the platform just to get it to install, and then you will probably find that the actual code demos segfault when you run them.

AMD's development processes are not really open, and actual development is silo'd inside the company with quarterly code dumps outside. The current code is not guaranteed to run on the actual driver itself, they do not test it even in the supported configurations.

it hasn't got traction because it's a low-quality product and nobody can even access it and run it anyway.

For a higher level GPU interface, Thrust provides "standard library"-like functions that run in parallel on the GPU (Nvidia only)

For GPGPU, I like thrust. C++-idiomatic way of writing CUDA code, passing between host and device, etc.

Users should work with higher level primitives like tasks, parallel loops, asynchronous functions etc. Think TBB, Thrust, Taskflow, lparallel for CL, etc.