I feel like i'm alone to dislike M1/what is it good for (Merged Mega Thread)

[ric22]

We eventually ended up with a “no upgrade” policy at work when we first started moving from HDDs to SSDs and got about 50 OCZ drives. Nearly all of them failed due to corrupted firmware about s month or so after use.

Plus, from a Capital Expense perspective, getting a random drive off Newegg a year later throws off the books. It’s just not worth it. These are business systems, not for your grandma. You do capital expenses and have the product last for 3-5 years and spread the cost out. There’s more to CapEx but I could write an entire book.

It’s fine for users or grandmas to get Mac Studios, but that’s not the target audience.

The last big firm that I worked for had stacks of SSD's waiting to be cycled in to the computers (or used as replacements, but failures were rare). I didn't realise replacing old storage with new was considered a grandma thing...

The desktops in the offices all had high end specs, in various configurations depending on workflow... and the directors had iMac Pros because they made them feel special. 😅 Not everything would get replaced or improved, that would be silly, but staff could request a newer/better GPU if they needed it, for example. No need to bin a desktop and replace the whole thing to give them better graphics performance, is there? I'm pretty sure no one worried about "throwing off the books"

 

[turbineseaplane]

Sorry, completely disagree. Intel CPUs run (ran?) fine in non-Apple laptops or computers.

Absolutely true

My i7 8700 based Hack is whisper quiet and performs like a champ in macOS and Windows 11 for gaming.

 

[darngooddesign]

People say Macs are designed for creativity rather than gaming. But I wonder what creativity task is M1 Max suitable for?

Let’s look at a few examples:
1. Photo Editing - I think base M1 works fine unless you’re editing images with a thousand layers, which most people apparently don’t...

Photo editing can def make use of the Max, especially if you want 64GB RAM. You don't need a thousand layers either, just running denoise software, focus stacking, HDR stacking, or panorama stitching are very demanding. Not just professions either, many hobbiest photographers can make use of the Max.

Just seems like you aren't very familiar with the industries you say don't need the M1-Max.

 

[Ethosik]

The other factor is that progress is slow enough in CPU and memory capacity demand increases now that by the time you need to upgrade you're likely also wanting things that aren't on your old board.

Like, new USB standard, new storage IO standard, etc. We're not living in the 90s any more where 50-100% generation on generation performance improvements for CPUs were common. You're more likely wanting to upgrade every 3, 5 or more years.

Upgrading the CPU or memory on an old machine won't get you those things.

As an example: lets say I was to upgrade a Skylake desktop (such as the one I have here at work) today:

• doesn't have any M.2 slots
• DDR4
• no USB4 / type C ports
• no PCIe 4

Its stuck with USB 3, no type C, SATA drives, etc. Upgrading the CPU in it, it's still going to be massively bottlenecked by everything else getting data in and out of it.

Heck my 2010 Mac Pro is still very usable today! I only upgrade systems every 5 years at best. 10 at worse. Things have slowed WAY down. But now with Apple Silicon, things have sped WAY up to where I’m going to have 5 different M1 * products.

 

[AlphaCentauri]

Music sequencing and processing with lots of virtual instruments and effects can bring ANY processor to its knees, including M1 Max and Ultra.

Not audio recording though - this is easy peasy even for Intel I3 or I5, not even talking about M1…

 

[Ethosik]

A) For what purpose do we even need 400GB/s?
B) Can a Mac even utilise all that speed? No, it can't, not even in lab tests designed to utilise as much as possible. I posted a link with a lot of detail on this the other day.

If you even need to ask, that tells you right there you aren’t the target audience for these products. Cost is no concern when you make up $10,000 on one large job. Performance of RAM is very needed for large data analysis.

 

[MauiPa]

Photo editing - only you're an actual professional
Video editing - ok, maybe
video encoding - no actual benchmarks for software encoders so no idea, hardware encoding useless
Sound editing - don't see why
Coding - only if you target Apple/Web, otherwise just issues upon issues
compilation - only for small projects
modelling/cad - no way, needs good gpu with fp64, good ray tracing
gamedev - only if you target mobile ios
chess - terrible performance for price
web browsing - anything would do
streaming - what would you stream anyway?
machine learning on gpu - cuda is just miles ahead
machine learning on cpu - why?, neural engine comes with big usability issues anyway
gaming - not really, apple doesn't like gamers
high ram workloads - usually if you do need a lot of ram you need more than apple offers and an option to expand
accounting - windows + office
databases - not M1 issue but good luck with the performance of the ssd with full syncs


so basically you don't need it if you're not already fully converted to the apple land

hardware encoding useless? so funny

 

[MauiPa]

I'm not seeing the purpose in this post? What professional work do professionals do? That is your answer the work that professionals do. The fact that no one games (is that work?) on Macs who cares? Otherwise, virtually anything that people can do on a computer. sure some folks who are windows fanboys willl tell you certain software won't run - and they are right, but there are always other ways to do things. If your education and training are restricted to certain windows only things, then sure, go windows, go deep, run along, nothing to see here folks (Southpark credit).

Or maybe the fact that apple sales are increasing while windows is declining means no one is buying computers for any real work, because real work is done on Windows. Did I get that right, or am I rambling?

 

[MauiPa]

Normal business stuff, office apps. Mail clients.
Java applets (machine control/logging)
VM's (Windows and Linux)
Coding/monitoring.
remote to work location.

And then there the general web stuff one does at home.
Chess, and I don't care about the performance as I normally play with speed chess settings and I absolutely abhor benchmarking.

is someone really whining about that chess benchmark, the one that wasn't written to even run on Apple silicon? does that smell of desperation?

 

[ric22]

If you even need to ask, that tells you right there you aren’t the target audience for these products. Cost is no concern when you make up $10,000 on one large job. Performance of RAM is very needed for large data analysis.

Pity you failed to read the other posts and discovered the Mac can't even utilise all that lovely bandwidth, eh? Spend those 10s of thousands you earn on top spec Macs in forlorn hope of saturating the bandwidth...

 

[clevins]

Well, here's a high level look at the M1 for some basic data science benchmarks: https://towardsdatascience.com/are-...d-for-data-science-lets-find-out-e61a01e8cad1 nd one more specific to TensorFlow and which compares the base M1 to the Pro https://betterdatascience.com/macbook-m1-vs-m1-pro-for-data-science/.

Machine learning and other things often benefit from GPU power so perhaps machine learning and other DS tasks would be helped by the Max or Ultra. My larger point is that you're looking at a subset of things and wrongly saying "well the Max isn't worth it for these"

Also.. So what? Are you just here for attention? Or is there a point to this, i.e. youre considering a new M1 and want input? I mean, the simple answer is "The Max and Ultra will benefit workflows that make demands on the GPU and/or can benefit from a lot of RAM and which don't need some specific GPU functions that they don't do". If that's not you, that's fine. It's not me either.

 

[bobcomer]

is someone really whining about that chess benchmark, the one that wasn't written to even run on Apple silicon? does that smell of desperation?

Not me, but it certainly gets mention a lot around here.

I really do hate benchmarks of any kind and only use them as rough indicators when comparing 2 similar machines.

 

[leman]

M1 is amazing for data science (unless you are exclusively on the ML bandwagon). But general data processing/transformation, statistics, simulations and all that good stuff, it flies.

Coding - only if you target Apple/Web, otherwise just issues upon issues
compilation - only for small projects

I doubt you will find any other laptop thats overall faster at building software. M1 offers unprecedented performance when worked untethered, and it has the battery life to make it feasible.

 

[bobcomer]

the simple answer is "The Max and Ultra will benefit workflows that make demands on the GPU and/or can benefit from a lot of RAM and which don't need some specific GPU functions that they don't do". If that's not you, that's fine. It's not me either.

A lot of RAM, and more performance cores. Both those sum up my needs pretty well.

 

[Danfango]

Maximising single core performance is usually better for coding / programming. As much as all those cores are nice, compilers are nearly all single threaded or have minimal improvement running parallel compilation.

My corp laptop has 64GB to run 8 VMs and half our stack so the M1 Max would be required for that.

But I just connect to my Linux desktop (Ryzen 3700X+64GB) and run it there and save the battery and don’t burn my dick off.

 

[ahurst]

M1 is amazing for data science (unless you are exclusively on the ML bandwagon). But general data processing/transformation, statistics, simulations and all that good stuff, it flies.

I second this, my 14" Pro flies through R and Python pipelines. Important to note that to take full advantage of the M1 in Numpy, you need to compile it from source against Apple's Accelerate.framework: that way it can leverage the M1's internal AGX matrix math stuff. Unfortunately Accelerate still offers an ancient 2009 version of LAPACK that's no longer supported by SciPy, but once they get their act together on that front those workloads will also likely see a healthy benefit.

It looks like R can use Accelerate.framework too, though it seems to be cautioned against by the devs. Would be interesting to test out some benchmarks and see what the benefits are like.

EDIT: Wow, okay, found setup instructions and some benchmarks of Accelerate-based vecLib on R vs the default R vecLib. Looks like performance results are mixed, but a lot of the benefits are on the order of magnitude of 1000% to 2000%. I wonder whether RSTAN can be linked against Accelerate's vecLib and what kinds of benefits you'd see there...

 

[alanvitek]

I’m an indie game developer and use an M1 Max. Published my first game on the App Store and working on another now. Max allows me to run Unity, Blender, Photoshop and xcode simultaneously, and compiling is super fast. No, I don’t do AAA graphics and things like ray tracing, but there’s more to games and 3D work than those aspects. A windows PC could certainly do some of those things, but can’t run Xcode.

 

[vddobrev]

4. Coding - You don’t need GPUs for these. M1 Pro or even base M1 is usually adequate. And people usually use CI for big projects so performance is not that important in general.

Exactly my reasoning why I did not get a Mac Studio, M1 Mac mini 16GB is perfectly fine for my software development needs. No need for GPU power. I also have a Mac mini 2018 Intel to cover Windows Bootcamp for Visual Basic coding when needed for active Excel sheet that interactively connects to Oracle database.

 

[darngooddesign]

Pity you failed to read the other posts and discovered the Mac can't even utilise all that lovely bandwidth, eh? Spend those 10s of thousands you earn on top spec Macs in forlorn hope of saturating the bandwidth...

The Max is faster than the Pro and the Ultra is faster than the Max. If you need 64GB you get the Max and if need 128GB you get the Ultra. Whether it's using all the bandwidth is irrelevant, as is the cost if you are earning good money with your computers. The faster M1 helps you earn more money than than the slower M1 and you deduct the purchase expenses on your taxes. It's a win-win for everyone except you.

 

[JouniS]

Apple didn’t build systems in a vacuum, they would consult well ahead of time, align with Intel’s release schedule and, if Intel says, “We’ll definitely be able to ship on time, in quantity and in a form factor that you’ve specified”, Apple can do nothing but build the enclosures/motherboards specified and trust Intel’s roadmap.

Intel more likely said something like "This is kind of what we are hoping to do, but we promise absolutely nothing." R&D is inherently risky, and you don't want to make promises you can't keep. You rather prepare for many possible outcomes, react quickly to new developments, design many different products, and bring the ones that work best to the market.

Apple kept selling thermally constrained Macs for years, and this was most likely a deliberate choice. (The other alternative was that they were grossly incompetent.) They almost certainly had many internal prototypes of thicker laptops better suited for Intel CPUs and AMD GPUs, but they chose not to sell them.

 

[stevec618]

Why do people still write these asinine posts and why are they usually written by people who obviously have no experience in the industries they're questioning?

 

[roach1245]

M1 Pro only allows for 2 external monitors and 32GB RAM max. Hence why I went for the M1 Max with 64GB RAM and to drive at least 3 external (4K) monitors, working in academia with large datasets.

Best laptop ever, flies through everything, haven’t rebooted in 3 weeks, and beautiful screen to look at when working away from my desk.

 

[vddobrev]

Why do people still write these asinine posts and why are they usually written by people who obviously have no experience in the industries they're questioning?

I think it is obvious why - the M1 package is take it or leave it CPU and GPU together, there is no separation. Comparing to Intel offerings, the GPU can be separate and not built in. The OP is questioning the benefit (or lack there of) of the included GPU with the M1 package for professionals that do not need GPU power, but still will benefit from CPU power.
Why is this so hard to understang.

I am a full stack developer for a multi billion international company (in the top 2 list) and I do not see any benefit for my work in the Mac Studio M1 Max or Ultra compared to the M1 - all it matters to me is code compile and SSD speed, GPU is not needed. So yes, the OP question is very valid.

 

[meanmean]

I second this, my 14" Pro flies through R and Python pipelines. Important to note that to take full advantage of the M1 in Numpy, you need to compile it from source against Apple's Accelerate.framework: that way it can leverage the M1's internal AGX matrix math stuff. Unfortunately Accelerate still offers an ancient 2009 version of LAPACK that's no longer supported by SciPy, but once they get their act together on that front those workloads will also likely see a healthy benefit.

It looks like R can use Accelerate.framework too, though it seems to be cautioned against by the devs. Would be interesting to test out some benchmarks and see what the benefits are like.

EDIT: Wow, okay, found setup instructions and some benchmarks of Accelerate-based vecLib on R vs the default R vecLib. Looks like performance results are mixed, but a lot of the benefits are on the order of magnitude of 1000% to 2000%. I wonder whether RSTAN can be linked against Accelerate's vecLib and what kinds of benefits you'd see there...

I'll also back this up.

I have 3 M1 Macs, Mini, Pro, and Max. They all do great both in Python and R; my primary domain is NLP (natural language, not non-linear). For laptops/desktops doing EDA and building PoC's they are awesome, even for basic Keras/Tensorflow work. Furthermore, most ML people do tends to be GBT under regularization where having 8+ cores is extremely helpful.

For big-data (whatever this means anymore) you're probably going to be using EC2 or an on-prem server to push through the data anyway. So the M1 allows you to still play locally while running big jobs in the background. It also helps that a GPU/TPU instance doesn't heat up my office.

 

[clevins]

I think it is obvious why - the M1 package is take it or leave it CPU and GPU together, there is no separation. Comparing to Intel offerings, the GPU can be separate and not built in. The OP is questioning the benefit (or lack there of) of the included GPU with the M1 package for professionals that do not need GPU power, but still will benefit from CPU power.
Why is this so hard to understang.

Because, like many of these posts, they don't seem to be asking in a spirit of understanding but in a spirit of asserting that they already know and their answer is the correct one. It's annoying because it's a waste of time and bandwidth and it's been going on for decades.

Look, if OP is genuinely curious along the lines of "I'm debating brtween a Pro and a Max and I don't see why the Max would be better since I do X and Y" that's one thing. But they're arguing that aside from video editing there IS NO advantage the Max which is both deeply silly and easily disprovable.

The OP is questioning the benefit (or lack there of) of the included GPU with the M1 package for professionals that do not need GPU power,

This is basically a tautology. Of course there's no advantage to having more GPUs if you don't need GPU power. The issue is what kinds of work do need GPU power and as we've already seen from the replies, that's not just video work.

I am a full stack developer for a multi billion international company (in the top 2 list) and I do not see any benefit for my work in the Mac Studio M1 Max or Ultra compared to the M1 - all it matters to me is code compile and SSD speed, GPU is not needed. So yes, the OP question is very valid.

And this is a perfect example of what I mean. You've outlined what you do and what affects your work and concluded that FOR YOU, the difference isn't important. That's very different from saying "there's no advantage for anyone except video editors" and then listing things that the Max isn't good for (erroneously in serveral cases).