567: The M1 Honeymoon Continues

3 Likes

Thank you David Sparks for confirming that 8GB with an ARM chip ≠ 16GB with an Intel chip. That was just wishful thinking and nothing that Apple ever claimed.

2 Likes

How would a “ARM”-Byte be and different than a “Intel”-Byte? I ordered my M1 pretty soon after release and I immediately went for 16GB.

Ah, but this one goes to eleven…

1 Like

I think there was some thought that the unified memory of the M1 Macs would somehow cut the memory usage down significantly. Also the lack of availability of >16GB must mean that less memory was needed. :smile:

1 Like

so what you are really saying is…I should not have returned my maxed out Intel Mac Mini ($2,800) for the sake of the M1 Mac Mini ($1,200ish) :joy:

More that the immediate availability of 8GB helped a lot of people re-evaluate their memory needs, especially in light of the more acceptable performance during periods of high memory use.

As I made my way through roadworks to Woolies supermarket listening to MPU here in Western Australia yesterday I heard David was talking about a ‘clever’ listener who sent his Kindle highlights to his email address which was set up to tag the email. Hey, that’s me! Then there is Obsidian. I naturally have thoughts about those highlights and I have learned that if I put them straight into Obsidian I build my note collection. I preserve and interlink my thoughts as I go along. All these years I had been letting those thoughts rattle around in my head and for the most part lost them or struggled to recreate them. I’m one of those people who cannot write an outline until after I’ve wrestled through the writing process, but have have usually been able to see that a particular idea comes before another when they are on notecards. So I’m following David and Stephen’s Obsidian adventures with keen attention. It is breakthrough time once again in computing with a new class of software like Obsidian and the M1. I’m like wow all over again!

1 Like

I’m typing this on my fairly maxed out i5 2018 Mini which I bought in 2018 only to discover that was not noticeably better than my 2011 Mini at video editing evidently because the intel graphics were not that much better - at least in practice. My mistake. Now I can buy the New M1 Mini for about what an egpu set up would cost so I’m taking my time deciding among the three M1 machines and exactly how I’ll set it up.

After watching endless videos of various video editors seamlessly handle up to 4K footage in the M1s regardless of 8 or 16 GB of ram it is pretty clear that the M1 graphics cores are getting access to ram in a revolutionary way - even on the 8 GB models. Both the 8 and 16 GB models will use the SSD too, but the testers have had to really push the machines to find video files that begin to slow down either machine. Of course the testers are not running any other software, so it is not immediately obvious that 16 GB is vastly superior. But what David brought out in this episode is that power users normally use many programs, often memory hungry, at once and will probably find themselves having to shut some of them down more often if they only get 8 GB.

A 4K 10 bit frame needs 47,2MB of RAM. Very fast access to RAM and SSD will make the effects of reaching memory limits less noticeable, if at all. Linear rendering of a video file, even 4K won’t even push RAM to it’s limits. And since access to the SSD is blazing fast, reloading a clip/sequence won’t be even noticeable. Also remember, in the editor, proxies are used. Yes, I also tested video editing on the M1. It’s amazing. But does not, for me, led to the conclusion that less memory is OK. Sure, the effects of reaching memory limits are barely noticeable.
I am working with several datasets that are several GB big. If I start a multivariate analysis or optimization model, it’s pretty easy to grind the computer to a standstill. Because the swapping/offloading events start to run into the thousands, if not millions, if the whole dataset is not in RAM. I was considering a 16"MBP with max memory because of this, but went the M1 route because…new and shiny. Right now, I am using my PC for some of those tasks. It’s GPU+memory just blow away the M1 with those tasks.
We have seen this effect before, when we moved from HDD to SSD. Memory swapping suddenly became way faster and you could get away with less memory. With HDDs is even made sense to avoid/turn off swapping, because…HDD.
I am really looking forward to the next generation(s) of Silicon Macs and I wish for one with 64+GB.

Thanks for those insights. Rest assured, I’m not about to buy an 8Gig M1! Also, I am not going to throw the large scale tasks you do at the M1 or I would wait for more powerful Apple silicon. I think I am going to get all I want or need from the M1 for my level of usage which is using Resolve for simple editing of HD and short 4K clips - typically a minute or two.

Brenda, show me your bookmarks.

I found it really odd to think that anyone would buy a modern machine with only 8GB ram! Even my 2013 iMac has 16GB on it and so does my MacBook Air I got in 2019.

But I guess I’m one of those folks that buys the largest most powerful system when I buy a new computer. I’ve never been happy starting out with something on the low end and I STILL run off the end of the machine abilities on occasion.

1 Like

I bought my 2012 Retina MacBook Pro and 2017 5k iMac with 16gb (the iMac now has 48gb in it). I certainly wouldn’t go less than 16gb in a general-purpose machine.

However, I did get a 2018 Mac mini with just 8gb. It was intended as a home server and I figured that it didn’t really need a lot of RAM for that role. It’s also user-upgradable (albeit with some difficulty) in case I decided that 8 gigs wasn’t enough.