MBP M1 Max vs. Core i9

[throAU]

Sometimes I'm not so impressed with my M1 Max MBP 16" (64GB RAM). Sometimes, it seems barely faster than the Core i9 15" (2018) MBP (32 GB RAM) that I also have. I leave one at work and keep one at home, so I rarely compare them side by side. Sure, somethings are perceptively faster, e.g. some Photoshop tasks, but for general tasks, I'm not so sure. So I ran some code that I use for teaching non-determinism. And I'm surprised. The M1 Max and the MBP ran neck and neck on this Perl one-liner. I checked. The Perl executable for the Max is arm64. Perl for the i9 is x86. You can run the same code easily. View attachment 2101246

You’re running perl interpreter code that has had 20 plus years of optimisation for x86.

The magic of M1 is in power efficiency (i.e., doing the same work with less power and getting better battery life and less fan noise) and drastically improving performance in tasks that can make use of the accelerators for GPU and ML.

Your perl one liner is basically worst case for M1.

Also bear in mind that the only difference between your M1 Max and an M1 Pro is the GPU which has zero bearing on this test.

Additionally this task will very likely not be multi-threaded, so the fact that the M1 Max is keeping up at much lower clock speed (Intel CPUs boost hard when only a single core is active (like 4Ghz plus vs. the M1 sitting around 3.2) is pretty impressive. You’d probably get almost exactly the same performance on this test on a base model M1 Air.

Try the same test running 8-10 or more copies of your script in different processes concurrently.

 

[leman]

You’re running perl interpreter code that has had 20 plus years of optimisation for x86.

The magic of M1 is in power efficiency (i.e., doing the same work with less power and getting better battery life and less fan noise) and drastically improving performance in tasks that can make use of the accelerators for GPU and ML.

Your perl one liner is basically worst case for M1.

Please, not you too. It’s exactly this kind of posts that various internet trolls use to talk about “Apple apologists”.

No, Perl is not specifically optimized for x86, it’s regex engine it’s just plain C that should work well enough on anything that dares call itself a “CPU”. And no, M1 is not just about power efficiency, it’s also a really fast, capable general purpose CPU. And how is that one liner worst case for M1? It’s not Cinebench.

I mean, come on guys. Let’s keep our feet on the ground. The methodology of this “test“ is completely flawed and the results are utterly meaningless (not to mention that we have a poster whose results contradict what OP published). But let’s deal with facts and computer science instead of inventing random justifications.

 

[throAU]

Please, not you too. It’s exactly this kind of posts that various internet trolls use to talk about “Apple apologists”.

No, Perl is not specifically optimized for x86, it’s regex engine it’s just plain C that should work well enough on anything that dares call itself a “CPU”. And no, M1 is not just about power efficiency, it’s also a really fast, capable general purpose CPU. And how is that one liner worst case for M1? It’s not Cinebench.

I mean, come on guys. Let’s keep our feet on the ground. The methodology of this “test“ is completely flawed and the results are utterly meaningless (not to mention that we have a poster whose results contradict what OP published). But let’s deal with facts and computer science instead of inventing random justifications.

I updated my post.

But i stand by the fact that the biggest win for M1 is efficiency, not outright performance. Additionally plain string processing is not what it was design for, as that runs fast enough on a potato in 2022.

 

[sam_dean]

I concede your point. I looked at the cost of electricity in the EU and the cost difference between 2020 and 2022 is horrifically high and I can see it being a major incentive having a laptop sipping electricity.

If it wasnt so tedious and I could have it automated I'd love to replace all my lighting, appliances and other things that uses electricity with the most power efficient models available at that time.

I'd cut my power consumption by half if I did.

 

[leman]

For what is worth, I have now run a simple perl benchmark on both my M1 Max and the 2019 i9 machine. This is matching 100k emails agains a RFC822 regex (taken from here https://www.ex-parrot.com/pdw/Mail-RFC822-Address.html). I preload the data into RAM and only time the regex matching section to equalise the SSD performance (hint: it doesn't matter, the launch + data load overhead is around 40-50ms on both machines). Obviously this is single-threaded benchmark. Code is attached. Pardon my crappy perl, it's the first time ever I tried to use this language.

Results:

M1 - 100ms
Intel i9 - 175ms

Hence, M1 can process 70% more real-world regex matches within the same time unit.

I hope this will put all the silly speculations to rest. Can't believe I had to waste half an hour for this nonsense.

But i stand by the fact that the biggest win for M1 is efficiency, not outright performance. Additionally plain string processing is not what it was design for, as that runs fast enough on a potato in 2022.

And you would be wrong on both counts. M1 has more cache, almost double the integer processing units and superior branch prediction. It excels in generic integer code like basic regex engines. The only chance for x86 to catch up is to use SIMD-optimized string processing where it has a chance of winning due to longer vectors and higher clocks.

 

[thinkinginventor]

I was a fan of the Apple M chip, How fast is was. I bought a M1 Mac book pro last year. But Mac Studio gives a preview how Apple or "Tim Cook" is going to use the M chip.

Un-upgradability
Planned obsolescence

The rumors of the new Mac Pro having one PCI-E slot is extremely disappointing. Will it even work with anything other than a apple card.

This is not the PowerPC re-run I was excited for but a completely close un-upgradable system.

Now intel and AMD are catching up in speed. M chip saves power "Great"

I have a Trash Can Mac Pro I bought new in 2018. Dropped it with Ventura which is fine. Kinda crappy but whatever. I also have a Mac Pro 2009 Both are 12 core 64gb ram. 2009 has a Vega 64. PCIE storage. Runs faster than the 2013. 13 years old.

A laptop thats not upgradable is fine, But a Mac Pro or Studio is a fail if you can't upgrade the ram at the very least.

I will not pay 600+ for extra ram from apple thats BS.

If i'm dumb enough I will buy a 2019 Mac pro when prices fall.

I want a M chip Mac pro in a 2010 configuration at the very least or Powermac G5 configuration.

If Apple gives there Pro Market the finger then I know where to put my shoe.

I hate intel, I hated the intel switch in 2006. I rocked a G5 and G4 until 2013. Thats when I got my Mac Pro 2009.

I would rather have intel or amd with upgradability vs a really good iPhone chip in a little desktop box costs 4k-100k lol

 

[thinkinginventor]

For what is worth, I have now run a simple perl benchmark on both my M1 Max and the 2019 i9 machine. This is matching 100k emails agains a RFC822 regex (taken from here https://www.ex-parrot.com/pdw/Mail-RFC822-Address.html). I preload the data into RAM and only time the regex matching section to equalise the SSD performance (hint: it doesn't matter, the launch + data load overhead is around 40-50ms on both machines). Obviously this is single-threaded benchmark. Code is attached. Pardon my crappy perl, it's the first time ever I tried to use this language.

Results:

M1 - 100ms
Intel i9 - 175ms

Hence, M1 can process 70% more real-world regex matches within the same time unit.

I hope this will put all the silly speculations to rest. Can't believe I had to waste half an hour for this nonsense.



And you would be wrong on both counts. M1 has more cache, almost double the integer processing units and superior branch prediction. It excels in generic integer code like basic regex engines. The only chance for x86 to catch up is to use SIMD-optimized string processing where it has a chance of winning due to longer vectors and higher clocks.

Until i see a ram slot or replaceable HD or at least 3 PCIE it's dead in the water for me. Last Desktop i owned that you couldn't upgrade was original Macintosh lol. Came out with 512 and Mac plus real quick.

 

[senttoschool]

If you run a task that is too simple, the megahertz might matter more. M1 Max runs at ~3ghz. Intel CPUs often boost 4 Ghz+.

IE. If you do a simple string log of "hello world" 1,000,000 times, the Intel is probably going to beat any Apple Silicon because it runs at a higher frequence.

To measure real performance, consider testing something that actually does more than just a one liner.

 

[Bodhitree]

My question would be, does it make a practical difference? For my use cases I rarely max out all the cores on my base M1, and I suspect for a large market segment thats true also.

 

[sam_dean]

If you're not sufficiently impressed by Apple Silicon's performance, please watch this!
And remember, Apple wasn't even the chip business 15 years ago.

I should have kept to

2012	Macbook Air 13" 22nm
2015	iMac 27" 14nm

I regret buying these laptops below

2017	Dell Inspiron 7567 15" 14nm
2018	Macbook Air 13" 14nm
2019	Macbook Pro 16" 14nm

Then replacing the 2012 & 2015 models with

- 2020 Macbook Air 13" M1 5nm
- 2022 Mac Studio M1 Pro 5nm

Then replace the 2020 & 2022 with

- 2030 Macbook Air sub-1nm
- 2032 iMac sub-1nm

 

[leman]

If you run a task that is too simple, the megahertz might matter more. M1 Max runs at ~3ghz. Intel CPUs often boost 4 Ghz+.

IE. If you do a simple string log of "hello world" 1,000,000 times, the Intel is probably going to beat any Apple Silicon because it runs at a higher frequence.

To measure real performance, consider testing something that actually does more than just a one liner.

This is not how this works. Apple Silicon has lower clock frequency, but it has more execution units. Every M1 CPU core for example contains seven integer execution units versus Intels five. Which means that Apple can do 20-30% more integer operations per cycle. And there are many more nuances.

 

[sam_dean]

Now intel and AMD are catching up in speed. M chip saves power "Great"

Apple is the most cash rich company in the world.

It can out buy anyone when it comes to future die shrink node manufacturing capacity.

So long as no idiot in the company becomes greedy Apple will always have 1st option to buy future tech.

As such Intel & AMD will always buy a node 1 generation older than Apple. This isn't a bad thing as it forces them and other companies to offer cheaper goods than Apple.

People who cannot afford Apple can buy the cheaper option.

As for all Apple Silicon Macs not having any internal upgrade options this is unfortunate but very understandable especially when it comes to memory. Making it not built-in will low its efficency and make it slower.

Although I wish they allowed for 1 or 2 NVMe M.2 slots to allow for 1 o 2 4TB SSDs.

 

[senttoschool]

This is not how this works. Apple Silicon has lower clock frequency, but it has more execution units. Every M1 CPU core for example contains seven integer execution units versus Intels five. Which means that Apple can do 20-30% more integer operations per cycle. And there are many more nuances.

When something is so simple, it does not matter how many integer execution units there are. All it matters is clock cycles.

Again, it all depends on how complex the code is. I gave an example of printing "hello world" one million times. Try it. If you have an Intel CPU that boosts to 4ghz+, it should beat any Apple Silicon handily.

 

[Gudi]

A regret I bought these laptops below

2017	Dell Inspiron 7567 15" 14nm
2018	Macbook Air 13" 14nm
2019	Macbook Pro 16" 14nm

Instead of waiting for the

- 2020 MBA 13" M1 5m
- 2022 Mac Studio M1 Pro 5nm

Understandable. Not only because of the big silicon transition, people in general (including myself) buy too many computers in quick succession. One should never get too excited about current offerings and always wait at least three years before buying the next device. Otherwise you end up with a bunch of computers from the same generation, who are all limited by the same technology. So instead of M1 and M1 Pro, maybe an M1 and M3 Pro? Then an M5 iPad and so on.

 

[MayaUser]

iPhone made it possible. If Apple stuck to Macs as their main profit driver then they'd be a die shrink or 2 behind Intel until now.

not the iphone, but the people behind the SoC...because iPhone could get the SD level of performance, because of the Intel chip guy that came to Apple and he is head in hardware technologies

Always the people behind it are the main reason, not the device per say...even without iphone, mac was always going for perf/watt, thats why they also swap to Intel back then, and now to arm
But i think the best reliability test was the iphone since they shipped billions over the decade

 

[leman]

When something is so simple, it does not matter how many integer execution units there are. All it matters is clock cycles.

Again, it all depends on how complex the code is. I gave an example of printing "hello world" one million times. Try it. If you have an Intel CPU that boosts to 4ghz+, it should beat any Apple Silicon handily.

If you are printing, your performance is limited by the ability of your console and OS to display messages. It’s no longer CPU-bottlenecked in most cases.

A better example would be incrementing a single value many times. There is no instruction level parallelism there and multiple execution units cannot be utilized. But no real code works like that.

 

[sam_dean]

not the iphone, but the people behind the SoC...because iPhone could get the SD level of performance, because of the Intel chip guy that came to Apple and he is head in hardware technologies

Always the people behind it are the main reason, not the device per say...even without iphone, mac was always going for perf/watt, thats why they also swap to Intel back then, and now to arm
But i think the best reliability test was the iphone since they shipped billions over the decade

Using your logic we should give difference to the suicidal Chinese factory workers that many westeners keep insisting want to jump off tall buildings

If iPhones did not have an annual volume of quarter billion units shipped then Apple would have outsourced chip R&D to an outside company.

 

[sam_dean]

Understandable. Not only because of the big silicon transition, people in general (including myself) buy too many computers in quick succession. One should never get too excited about current offerings and always wait at least three years before buying the next device. Otherwise you end up with a bunch of computers from the same generation, who are all limited by the same technology. So instead of M1 and M1 Pro, maybe an M1 and M3 Pro? Then an M5 iPad and so on.

This is why I advocate mandatory replacing devices to the next year model after last macOS/iOS/iPadOS Security Update is released if no pending paid project will pay for the upgrade sooner.

Intel CEO stated that typical laptop/desktop replacement cycle occurs every 5-6 years while Apple says 4 years.

 

[sam_dean]

When something is so simple, it does not matter how many integer execution units there are. All it matters is clock cycles.

Again, it all depends on how complex the code is. I gave an example of printing "hello world" one million times. Try it. If you have an Intel CPU that boosts to 4ghz+, it should beat any Apple Silicon handily.

Most people do not do that application.

Like how would say a 2015 Canon EOS 5Ds R 51 megapixel RAW image impact the speediness of a 14nm Intel chip vs 5nm M1 or M2 derived chip?

 

[thinkinginventor]

My question would be, does it make a practical difference? For my use cases I rarely max out all the cores on my base M1, and I suspect for a large market segment thats true also.

M1 Macbook pro runs great, fast! Lack of ram holds it back. Battery life etc runs great. Desktop's I'm concerned. Mac Pro 2013 is about on par with it, except 64 gb ram. and 2009 mac pro has vega rx 64 graphics is much better.

 

[thinkinginventor]

Understandable. Not only because of the big silicon transition, people in general (including myself) buy too many computers in quick succession. One should never get too excited about current offerings and always wait at least three years before buying the next device. Otherwise you end up with a bunch of computers from the same generation, who are all limited by the same technology. So instead of M1 and M1 Pro, maybe an M1 and M3 Pro? Then an M5 iPad and so on.

Yup! Thats what i've been doing. 13 years out of my 2009 mac pro. Many years out of trashcan. Tested the water with the M1 macbook pro. Waiting for Mac Pro M3? They should change the name Mac pro kinda boring. PowerMac had a better ring to it. ehh I'm getting older stuck in my ways! hahah

 

[lcubed]

If you are printing, your performance is limited by the ability of your console and OS to display messages. It’s no longer CPU-bottlenecked in most cases.

A better example would be incrementing a single value many times. There is no instruction level parallelism there and multiple execution units cannot be utilized. But no real code works like that.

this is absolutely true. i remember speeding up a bit of matlab code by three orders of magnitude by getting rid of a print statement.

 

[leman]

My question would be, does it make a practical difference? For my use cases I rarely max out all the cores on my base M1, and I suspect for a large market segment thats true also.

If you do this kind of work, of course it makes a practical difference. Apple Silicon is much much faster on these workloads.

 

[thinkinginventor]

If you run a task that is too simple, the megahertz might matter more. M1 Max runs at ~3ghz. Intel CPUs often boost 4 Ghz+.

IE. If you do a simple string log of "hello world" 1,000,000 times, the Intel is probably going to beat any Apple Silicon because it runs at a higher frequence.

To measure real performance, consider testing something that actually does more than just a one liner.

All Risc based chips are good at that. Powermac g4 half clock speed of P3 or P4.

 

[thinkinginventor]

Apple is the most cash rich company in the world.

It can out buy anyone when it comes to future die shrink node manufacturing capacity.

So long as no idiot in the company becomes greedy Apple will always have 1st option to buy future tech.

As such Intel & AMD will always buy a node 1 generation older than Apple. This isn't a bad thing as it forces them and other companies to offer cheaper goods than Apple.

People who cannot afford Apple can buy the cheaper option.

As for all Apple Silicon Macs not having any internal upgrade options this is unfortunate but very understandable especially when it comes to memory. Making it not built-in will low its efficency and make it slower.

Although I wish they allowed for 1 or 2 NVMe M.2 slots to allow for 1 o 2 4TB SSDs.

I don't think its about not being able to afford it....

I can run out get 8k machine. What can it do for me?

how can it help my work flow?

If I make that type of investment. Macs last a long time. Pro machines are supposed to last longer. I don't need apple to limit the life of my investment.

Chip vs Chip is just a sliver of the pie. As a machine ages, software size grows. Fast ram is good. More ram is better. I've seen more speed improvements with Ram and HD and Video Card upgrades then the CPU.

Apple Silicon replaces, GPU, RAM upgrades with locked in pure performance. But If I buy a fast machine now, it needs more ram later im stuck. Virtual Memory on the HD is going to slow it down where, just going to have to buy a new machine.

Mac Studio, they choose to use proprietary HD's.

Now I've been a Mac user for 30+ years. I been though the 68k to PPC change PPC change to intel. All of them had upgrade paths. They had upgrades turn a 68k mac into a PPC. My three longest lasting macs Powermac 8500. dual g4 upgrade. Powermac g4 with a dual 1.4ghz upgrade. & Mac pro 2009 with a 12 core 3.46 chips.

My worse purchase was mac pro 2013 I bought in 2018. I needed that machine for software for work. Apple killed support for Ventura. I went to do a trade in for a Mac studio.

They offered to Recycle it. What a nice company lol. I should of waited a year and half for the 2019 mac pro. I couldn't.

Apple needs to be pressured to offer upgrade paths for pro machines. Every-time i go into the Apple store, I remind them why entertainment industry is moving away from apple. Mac Pro 2019 was a comeback, I'm waiting for new Mac Pro see what they do.