Never tried karma on the mac laptop before (mainly use mac for other things than rendering)
Just for fun opened up the stage, dropped a pighead and switched the viewport to karma… ops, does this thing work? Only the two efficiency cores work at all. Am i doing something wrong?

Should work fine. Attached a screenshot of it rendering away at a pig head on my M1 MacBook Air (not sure why your render's going to take 2 hours though).

As you can see it's hitting the efficiency cores, and I'm not sure why. It does appear to hit them first and then ramp up the performance cores - if you render something a bit more complex than a pig head they will start to chip in. This doesn't seem particularly efficient. Karma's IPR thread reservation has always been a bit weird; it never seemed reserve cores as I'd expect.

To me this looks like it only uses the efficiency cores at all. In the case of your M1 it has 4 e-cores and those are used fully. On my M1 Max I have 2 e-cores and those are used fully. On my main TR Pro 3975wx workstation, all 64 threads are used.

It does predominantly use the efficiency cores. With a bit more geometry the other cores pick up some of the load, but it seems the efficiency cores are at 100% and the performance cores peak at 20%. Possibly that 20% is it picking up system tasks as the efficiency cores are saturated.

Seems like this is an Apple Silicon issue, rather than a Mac one; on my iMac, Karma happily renders away using all 10 core / 20 threads.

Not sure if this is accurate or not but may be of interest

https://apple.stackexchange.com/questions/426874/how-does-macos-decide-when-to-use-m1s-performance-vs-efficiency-cores [apple.stackexchange.com]

i reported this as a bug from M1Pro, but got a (reasonable) response that it is not officially supported, and on a non-M machine it works as intended, so i got my bug report dismissed

PS - render to MPlay will utilize all cores. viewport render is stuck with a couple of cores and that is pretty much it, no matter what you tweak in sliders

Hi All,

Yup. This issue is on our radar and we are looking into a fix to have all cores used on M1. No ETA at the moment though but sooner rather than later.

Cheers,
Rob

rvinluan
Hi All,

Yup. This issue is on our radar and we are looking into a fix to have all cores used on M1. No ETA at the moment though but sooner rather than later.

Cheers,
Rob

we're still talking through rosetta, right?

chf
we're still talking through rosetta, right?

Yes. That is correct.

Cheers,
Rob

This has been fixed in 19.0.524.