Niall Horn

Would be cool, but you can just use flat 1D arrays and write a simple 2D/3D --> 1D indexing function to map the index to the flat array. This is done in a lot of libraries / codebases as it reduces the number of indirections needed when accessing the internal elements of the nD array.

Eg, map a 2D Index to a 1D flat array use : i * N + j.

I'm interested in this also, however as a former FX TD now in CS education, I think what people don't realize is switching to a completely different architecture is a big dev task for a relatively small subset of users. It's not as simple as just re-compiling the Houdini codebase for a different instruction set, ARM is a reduced instruction set and has different inner workings than x86/64, to get the best possible performance we know SideFX strive for, it would be a lot of work to tune and port existing code to run in a performant manner.

Not to mention the large amount of third party dependences some of which may not be open source and thus SideFX relies on those vendors porting and re-building their code to run on Arm64, the ones that are open source may need work to run as well on ARM as they could on x86-64, we as an industry have never really needed to think about building tools that run on ARM.

Then we have the issue of the M1 GPU, Apple deprecated OpenGL and OpenCL in favour of pushing their Metal API which is no doubt a good choice in the long run to maximize the throughput of their hardware, but for an industry whose tools for Hardware Rendering(eg Viewport) are currently heavily reliant on OpenGL (an older Graphics API) this would mean either writing a new back-end of the Houdini Viewport to run natively on Metal, or using translation frameworks such as MoltenVK with a Vulkan backend, that could run cross-platform, both big tasks in themselves and while there's no doubt the viewport utilizing the Vulkan API could have significant performance benefits, those benefits have a work to benefit ratio than is lower than it may seem. However I do belive current MacOS support does use an older OpenGL standard, but there is no guarantee how long this will be supported due to the deprecation.

With all that in mind, we should be grateful Houdini as is runs pretty well using Rosetta 2 !
Saying that, what Apple have done with the M1 Mac is really great and it set's up a exciting future, so in the long run it would be really cool to see a native M1 / Arm based Houdini binary from SideFX. Also some Linux distros (Ubuntu, Fedora etc.) are able to run on Arm and there is significant interest in reverse-engineering the M1 Chip to be able to run Linux, which may add more users and maybe even more interest in a native M1 Houdini binary.

Disclaimer - This is educated speculation and I do not work for SideFX (sadly !)

I just ran into this, thanks a bunch, Norton was causing this issue for me in H16.5, the firewall was the cause of the issue.