The M1 Mac Mini 8GB RAM machine can only run 3-4 threads of MCM. It's an 8 core/thread CPU, so I'm getting at best 50% of the CPU power. Also, I'd like to add that Boinc/MCM is fairly often only running 3 threads (because the emulation gobbles up so much RAM). So I can really only count on 3 threads (again out of 8) running MCM.

To your point, the M1 threads are crunching at about the same rate as a several year old i9-9900 vPro threads, but again, I'm only getting 1/2 of my M1 CPU's threads tops.

The good side is that the M1 only uses 30 watts of power when tapped out and since I cannot tap out the chip, I'm guessing the Mac is around 20 watts. It also gives off very little heat (which I don't really care about). The M1 is also dead silent.

MCM is going to need to support the M1 real soon...all the Macs will be on M1 by end of summer and MCM is going to notice a massive drop in WUs being returned as more and more Mac people upgrade.

I would bet that if MCM was written natively to use the M1 natively, not only would we get all 8 threads, but the WUs would be crunched faster because there's no emulation software taking up CPU cycles.

As a comparison, my M1 Mac Mini is doing 3-4 threads and returning about 35 WUs a day and the machine cost me $589. 3 Years ago I bought Lenovo Thinkstation P330 machines with the i9 i9-9900 vPro which crunches 16 threads and returns about 180 WUs a day and that machine cost me $725. So I'm getting about 5x the results with the old Lenovo than the new M1. Comparatively, for the M1 to match the Lenovo, I would need to buy 5 Macs so thats just under $3000. $3000 vs. $725. Not good from a crunching point of view. Ideally, if the M1 was using all 8 threads, I would only need to buy 2-3 Minis so between $1200 and $1800. I'm not trying to compare an i9 to the M1 in overall performance. But for the $1200 or $1800 (if I could use all 8 threads of an M1 and hence buy 2-3 Macs) I get a super tiny Mac, using far less power, and giving off far less heat. There's value in that for me.

I don't really know a lot about the 4 performance cores vs. 4 efficiency. I've never found a website that has dived deep into it, provided examples and stats, and yet also summarized it in laymens terms.

What I meant about only getting 50% of the CPU power really was a typo...I'm only getting, at best, 50% of the cores.

If MCM were to run natively on M1, 1)MCM wouldn't take up all that RAM per thread, 2)it should/could use all 8 cores...albeit who knows what the performance is on the 4 efficiency cores, 3)always utilize all 4 performance cores, 4)likely the 4 performance cores will crunch at least a little faster than the cores operating now with the emulation. Summary: running natively on M1 should crunch significantly more WUs per day than currently through emulation. Heck, even just running all 4 performance cores 100% of the time would be a 33% increase in WUs per day over the 3-core processing it is doing today.

Wow, that M1 runs MCM in under 2 hours. My desktop AMD Ryzen 1400 runs them in under 2.5 hours

Bring out those native apps.

Hmmm, changing the BOINC/WCG memory setting was the main change that I made on my system. The only other thing I can think of is the version of the BOINC agent that you're running - did you download it from the Word Community Grid site or directly from the BOINC site? The version I'm running is the "Apple Mac (64-bit Intel and Apple Silicon)" 7.16.14 version downloaded directly from BOINC's site: https://boinc.berkeley.edu/dl/boinc_7.16.14_macOSX_universal.zip