I have been testing the new M4Pro and M4Max chip on a Mac Mini(12 core , 24Gb ram) and a Macbook Pro (16 cores, 64Gb ram), and the results where quite good until I start testing some flip simulation, the M4Max was almost twice faster than my PC (Ryzen 9 9300x, 128Gb Ram, RTX3090) , but when I pushed the particle separation, something weird start happening, as soon as the simulation has more than 3Million particles, it sarts to slow down exponentially making the simulaton super slow. while the PC was simulating around 12 millions particles (2min per frame) , the Mac was taking more than 10 min per frame for just 4Million particles.

I did the same test on Houdini 20.0 and Houdini 20.5, I also tested the same file on a Macbook pro m2Max (12 cores, 32Gb ram) and had the same weird results.

Anyone wiht an Apple Silicon had experience the same issues when doing flips?

I attach the file if anyone with a Apple Silicon wants to do a test.. thing start to slow down frome frame 80, aound frame 95 , it's almost stuck.

I can reproduce this on an M1 Ultra. Can you please submit a bug to support so that this doesn't get lost?

Has there been an update to this? I'm having the same issues on an m4 max(16 cores, 128GB ram) Houdini 20.5.526.

It looks like we are running into an issue within the macOS kernel. Apple is investigating.

I am trying to find a workaround in the meantime.

Thank you for the quick reply.

Can you confirm this is a software issue and not hardware?

It's a software issue.

Has there been any update to this? I'm running into the same issue with an M3 ultra (32 cores, 256gb ram)

Hi! Did you manage to solve the problem? i'm just about to buy mbp on m4 max and may not buy it if there is still no solution.

FWIW, I just bought a M4 MacBook Air and have been doing basic procedural modeling successfully for a few days.
HOWEVER, I just ran into an issue with POPs (Apprentice 20.5.613, Sequoia 15.5). Even a super simple network with only Gravity and Wind starts to misbehave in a way I've never seen (been using Houdini for about 9 years). It starts when I set the constant activation to "$F<=1". After that particles start disappearing on most frames, flashing in on occasion. Makes POPs unusable. I haven't tested any further to see what else sets off the issue.
EDIT: Restarting Houdini seems to have fixed the issue. The POPs networks are behaving again. Don't know what that was about.

I've been informed by our contacts at Apple that this issue has been fixed in macOS Tahoe Beta 5, and has also been backported to macOS Sequoia in 15.6.