11:09 AM - The problem with M1 Macs
I think a lot of folks aren't seeing the segmenting issues with M1 Macs, particularly the new pro/max variants. Let's talk about that for a minute.
First, the good. Apple did make an extremely fast laptop processor. They've got a fast ultrabook part in the M1 that's as fast as a ryzen 2700 desktop chip. They've also got this new CPU line that is some value faster with a nvidia mobile class GPU equivalent minus all the nvidia features. (well we should really compare these to amd parts)
Apple did add most ports back except ethernet.
Where's the problem?
First, there's the RAM issue. Now M1 fans will be quick to tell you the M1 laptops will hit swap so fast you barely notice it's happening. They love the swap speed. The problem of course is that swap use kills SSDs. Apple tends to ship fairly decent SSDs, but they're not enterprise grade. They will wear out eventually. Most consumer SSDs are rated for 5 years for the average person. This is often based on the size of the drive and being written multiple times. Many drives are sold with TBW values where we know when it should roughly wear out. Apple doesn't release this information. We have no idea when the drive will fail. On an M1 that is fatal due to the SoC design. Sure there are some EE folks who think they can possibly solder a new one on but most of us can't do that. It's effectively dead and when EOL apple won't replace/fix it either. The original M1 macs biggest problem aside from monitor support and ports was the lack of RAM. 1 6GB is fine for casual computer users who use facebook, office and even entry level youtube content types or front end web developers who only use node and vscode. It doesn't work for folks that need VMs or run a lot of docker containers and do backend development in java or other languages. They need a bit more.
Corporate developer specs are often now requiring 32GB or 64GB of RAM depending on the environment. My current employer requires 32GB for devs and my last job was 64GB. Many other folks I talk to are in the same boat. That means the original M1 was useless for most companies that needed dev systems. Enter the max/pro chips. They offer 32GB and 64GB variants that fit this problem. The issue there is the PRICE. When you really only need M1 performance plus 32GB of RAM, it hurts to go up to 2900 dollars to get it. Large companies will just suck it up and pay if they want macs of course. For smaller shops, you're better off buying an amd or intel laptop though.
Let's talk chip performance. From a laptop perspective, apple knocked it out of the park. We can all pretty much agree with that. In a desktop context, apple has failed so far. Granted they haven't release the higher tier iMac or Mac Pro yet. They are still working on this. The entry level iMac is useless for the same reasons mentioned above. It's a ryzen 2700 speed wise and no ram. It's fine for folks that don't do software development or other pro tasks. It's not an Intel or AMD killer though. It can't be at ryzen 2700 speed. The new chips are likely to get used in the iMac 27" or whatever replacement we're going to get. If they aren't overclocked or given extra cooling, they're likely to beat out the current intel parts shipped in iMacs but certainly not going to touch higher end alder lake parts coming from Intel or 5000 series ryzen. Apple would need to 3x the performance of the original M1 part to get to current AMD specs. 2x gets them past most of the 3000 series, with the exception of the 3900x and 3950x. in short, apple has likely beat amd's older 8 core parts but not their current parts. Now folks are thinking... but apple did kill it on performance per watt. That metric doesn't matter in this discussion because we're talking about actual performance. If you need X performance, a part that can't hit it doesn't matter even if it barely sips electricity. It failed to hit it. In a laptop context, battery life is helpful up to a point. Apple went way past what's needed. 4-8 hours is really the sweet spot for most people. 8 is basically "all day" for most workloads. Anything past that, and they should have overclocked the CPU a bit and used some extra juice. Back to desktop workloads. Apple should be competing with AMD, not Intel right now. That's who to beat. If Apple released a desktop faster than a 5950x, that would certainly be a huge deal. It's also needed to get past that performance for the mac pro anyway. That's the goal for apple, to scale up to xeon/threadripper territory for pro use. Apple is not hitting that goal right now. They may in the future, but right now it looks like their content making something as fast as a 10th gen intel chip. You might be thinking I'm being unfair because apple hasn't released that yet. Perhaps, but remember Tim Cook likes to reuse parts. The M1 iMac and MacBook Pro use the same chips. He'll likely do that with the new iMac 27/30 too.
Why is it a big deal if apple catches up to AMD? Well Apple has a process node ADVANTAGE right now. They're on 5nm. AMD is still kicking their butt on an older process node. Intel is still competitive on desktop with their older process node and have 10nm parts coming (well pre rebranding) with alder lake as well as big.little. If Intel and AMD can stay ahead of apple with a process node deficit, it means that either Apple is too fixated on mobile to ever make a performant desktop part or that they simply can't do it with their tech yet.
Apple has 5nm locked up. They should be destroying everyone with ARM + 5nm. Instead, we just get fast laptops. That's great for people who love laptops exclusively. it sucks for people who want the fastest SYSTEM possible regardless of form factor.