Apple M1 Ultra released today

nick_papageorge_dayjob

Apple released the highest spec M1 chip thus far today, the M1 Ultra. Its basically two M1 Max's joined together. I ordered one with 64 GB of memory. My CAD top level product assemblies are maxing out my little 8GB M1 Mini right now (but work ok on a Windows CAD workstation with 32GB ram and 5700XT GPU). I'll have the machine in a few weeks and will report back how it does.

billy2

exciting!

Here's my old M1 for comparisons:

nick_papageorge_dayjob

I find that test is not meaningful for large assemblies that tax the ram on the computer. My M1 Mini scores similar to the PC I built 2 years ago to run Creo, but the PC blows it out of the water when I'm working on a design with many features and many parts.

wayne_sauder

@nick_papageorge073
 I would love to see the results of the test @MichaelPascoe wrote.

S1mon

nick_papageorge073 said:

I find that test is not meaningful for large assemblies that tax the ram on the computer. My M1 Mini scores similar to the PC I built 2 years ago to run Creo, but the PC blows it out of the water when I'm working on a design with many features and many parts.

Can you elaborate on the specs you attribute this difference to? (I'm guessing RAM would be #1, but interested to hear).

nick_papageorge_dayjob

S1mon said:

nick_papageorge073 said:

I find that test is not meaningful for large assemblies that tax the ram on the computer. My M1 Mini scores similar to the PC I built 2 years ago to run Creo, but the PC blows it out of the water when I'm working on a design with many features and many parts.

Can you elaborate on the specs you attribute this difference to? (I'm guessing RAM would be #1, but interested to hear).

Its still a strong hunch at this stage, but I do my work on 3 computers. A PC I built for Creo before switching to OS, a dirt cheap $600 Mac Mini with 8GB ram, and a dell precision laptop with a Quadro CAD card. The Mac outperforms the Dell laptop about double on the OS performance test. But the Dell laptop has 16GB ram, the Mac has 8GB ram. Working in larger assemblies, the Dell Laptop is noticeably snappier. I mean when you have to edit a mate, the time it takes for the mate edit to open and for your keystrokes to be recorded are noticeably faster on the Dell laptop.

On the other hand, the Mac can work inside of a single complicated 2d drawing a lot quicker than the Dell can. So I think what is happening is the Mac's better GPU can make simpler models faster, but when the models get too large, the motherboard ram becomes the limiting factor.

The PC desktop I built I put 32GB motherboard ram in it, and it has a 5700XT GPU that has 8 GB ram. That also performs about the same on the OS performance test as the Mac Mini. But it runs circles around it in big part studios and big assemblies in how fast the user interface responds when making edits. Things like spinning the model and zooming seems the same on all 3 systems. But editing features in Studio, or mates in Assembly, the more motherboard ram, its way faster.

nick_papageorge_dayjob

I've had the M1 Ultra for about one week now. I got the lowest spec M1 Ultra. It is fast, but I don't think its any faster in OS than the PC I built 2 years ago for Creo (third column in chart), or the M1 Max MacBook Pro (that I returned). I definitely feel the higher ram helps. I have not gone to a swap file in the whole week of using the computer, and I have had several big studios and assemblies open concurrently. I still have some jittering when spinning large assemblies or studios for a couple seconds until it smooths out. I also still get the annoying message that the Firefox browser window is slowing down my computer when I have a big studio or assembly loading.

If I were to do it again, I'd probably go with the M1 Max Studio instead of the Ultra, but upgrade it to the 64GB ram version. That would save $1400. That's a custom configuration, and is 12 weeks back ordered. The one I got is the standard configuration of the Ultra, and you can walk into an Apple store and pick it up same day.

I actually wanted to get that configuration and try them side by side, and return the more expensive one if there was no difference. But, I'd miss the return window of the Ultra because these are so far back ordered. Everyone wants them. Apple did well with this machine.

One other note, OS performance on Safari is horrible on the entire Mac ecosystem. OS should have a big banner that says "DON'T USE SAFARI". I went to the apple store and logged into my OS account using Safari, and it was just bad. Since it was a store machine, it didn't have Firefox on it for me to try. So it was pointless to try to compare a M1 Max to an M1 Ultra in the Apple store. The sales reps actually came up to me and asked why I was interested in this Studio machine. I tried opening my CAD to show them, and it was a stuttering and beachball mess. I told them on Firefox it opens fine.

Latest data below. M1 Ultra is all the way to the right.

steve_shubin

@nick_papageorge073

thanks for this info Nick !

that was interesting to hear how it compared to the MacBook Pro

billy2

@nick_papageorge073

Could you please post the OS speed test results? As bad as it is, it's a common denominator and somewhat useful.

It's looking like an upgrade is in the future when the new macbook arrives. In the past, a 3 times spin gain wasn't that noticeable and I was always looking for a 10x increase when upgrading video. Is the spin speed that noticeable in OS? Is it worth $2500 bucks?

MichaelPascoe

I would be interested to see the results of this with the Onshape benchmark (UNNOFICIAL)

nick_papageorge_dayjob

billy2 said:

@nick_papageorge073

Could you please post the OS speed test results? As bad as it is, it's a common denominator and somewhat useful.

It's looking like an upgrade is in the future when the new macbook arrives. In the past, a 3 times spin gain wasn't that noticeable and I was always looking for a 10x increase when upgrading video. Is the spin speed that noticeable in OS? Is it worth $2500 bucks?

849+ and 639+. It maxes out the test (as does any recent gaming PC), so the test becomes useless. See the right column on a couple posts up.

nick_papageorge_dayjob

MichaelPascoe said:

I would be interested to see the results of this with the Onshape benchmark (UNNOFICIAL)

I looked at that thread a few weeks ago and I was confused how to do it. I'll take another look soon and see if I can add results.

billy2

Thanks, it's better than my M1.