M4 Pro Chip Benchmark Results Reveal an Extremely Impressive Performance Feat

[chucker23n1]

That's the problem, there's no such thing as "longevity" on these recent Macs when Apple practically bashing the last gen M chip by a huge margin, year over year. Just keep buying the newest baseline Mac because it'll be faster than last gen M Pro chip, which may only got out 6 months earlier (i,e M3 Pro chip vs. baseline M4). I hope you get the idea.

“Products have poor longevity when their successors are significantly better” is an absurd take.

 

[chucker23n1]

A friendly note that Geekbench6 is not designed for big CPUs

Yes it is.

and the multicore performance scaling drops significantly after a certain core amount due to its design to reflect common consumer tasks, which means it might not reflect the workstation tasks performance that scales well with core count, like compiling a lot of source code using all cores.

Even on high-end workstations, very few tasks actually scale like that. Hence Geekbench 6’s design.

 

[Newgoblin49]

Welocome to tech. Y'all need to get over your fear of next year being better, because it always will be. And next year it also will want more RAM...

you’re not getting the point.

We’re not talking about next year being better , that is obvious . We’re talking about the value of these Pro/Max devices erratically losing its value because of this.

 

[chucker23n1]

starting when Mac users care about the benchmarks and score and specs so much? another sign of just another IBM

Back in the late 90s, Steve had Phil do performance comparisons between Intel PCs and Power Macs.

So, for quite a while.

 

[chucker23n1]

you’re not getting the point.

We’re not talking about next year being better , that is obvious . We’re talking about the value of these Pro/Max devices erratically losing its value because of this.

But they’re only losing value if you intend to sell them. They’re not actually losing value to you at all. They’ll still compute the same way they did on day one.

 

[Newgoblin49]

The improvements are just no where near as significant as you are making out even if we accepted your premise.

I’m still confused as to whether you want Apple to innovate slower or not.

Or whether you just think that the top model should be avoided by most people. Which is far less controversial.

He’s just pointing out that its value drops very fast . It’s not that hard to understand

 

[Malus120]

Great result although honestly about what I would’ve expected given it’s a 10P/4E setup on a new arch.

Much more interested in how the M4 Pro’s GPU stacks up to the older Mx Max chips, but I guess we’ll know soon enough…

 

[Will Co]

“Products have poor longevity when their successors are significantly better” is an absurd take.

Fully agree. Let's not bother with progress at all, eh?

 

[wyrdness]

This comment with more than a dozen likes complaining about tech advancing too fast amazes me. You folks need to come to grips with tech always advancing, and not "feel much gutted when the next gen chip is launched." Sheesh.

But that comment doesn't have a dozen likes. At time of writing this it has 13 reactions. Only 3 of those are 'likes', the 10 others are mostly dislikes or laughter at his ridiculous 'reasoning'.

 

[Gnattu]

Even on high-end workstations, very few tasks actually scale like that.

I compile code every day and I can tell you code complication DOES scale like that. It does not matter to your workload does not mean it does not matter for everyone.

 

[Will Co]

I don’t understand these complaints about M series chips getting “too fast too quick. This has happened before. In the early days of x86 you could expect that if you bought top of the line CPU next year it’s gonna be totally eclipsed by whatever comes out, that was guaranteed.

We are at that point again with ASi.

Agreed. At a given point in time, you purchase the best piece of kit that your budget allows and that you can justify from a technical point of view. Then you use it. You're already out of date the moment the kit arrives: the company that made it probably already has the next generation of silicon on the bench with their development team. Just get on and enjoy your shiny new bit of kit. It's a tool to get your job done.

 

[Fuzzball84]

Right, or anything else it can do that the Mini can’t.



You need to have a very narrow view of what a computer is and can do to think CPU speed is the only determining factor in what gives a computer value. No one with a MacPro is going to give it up for a Mini. People that need what a MacPro does won’t be buying a Mini.

For example, a Mini might suit my needs, but I’m waiting on GPU performance numbers against tasks I do. I don’t give much of a crap about the CPU speed, as those have been fast enough for a few generations.

This is missed every time. You’re right, there is a reason Apple has different categories. Mini, Studio, Pro. Each caters to a particular set of requirements and although there’s some overlap, they target different workloads and connectivity requirements.

The Mac mini however is still a superb ultra small form factor PC for most people and especially people new to Mac.

 

[Fuzzball84]

Agreed. At a given point in time, you purchase the best piece of kit that your budget allows and that you can justify from a technical point of view. Then you use it. You're already out of date the moment the kit arrives: the company that made it probably already has the next generation of silicon on the bench with their development team. Just get on and enjoy your shiny new bit of kit. It's a tool to get your job done.

This is exactly right. Buy nice and then forget that a newer model is coming out or that if you’d waited you’d get another feature. Unless you’ve got lots of money to burn, it’s financial sense to just buy the best you can and then extract value from what you bought.

Despite buying a base MacBook Air M2 a couple years ago and seeing the RAM doubling now, I don’t really care. It does what I need from it, it’s a fantastic quality device and I’ve probably got at least 5 more years from it.

I even know people who still use 10 year old Macs and MacBooks. A testament to that they are still useful and the build quality of the product. They have well and truly extracted value from an otherwise relatively expensive consumer device.

 

[chucker23n1]

I compile code every day and I can tell you code complication DOES scale like that.

In what ecosystem?

For example, with Swift, an M2 Ultra is only 75% faster than an M2 Max, when it should be 100%. And an M1 Ultra is only 38% faster than an M1 Max.

Amdahl's law continues to be a thing.

It scales even less when you're in ecosystems like .NET and Java. (I'm actually unsure and curious how the TypeScript compiler scales.)

It does not matter to your workload does not mean it does not matter for everyone.

Of course more cores matter to my workload, but even in such a "workstation" scenario, Geekbench 6 is correct to limit the impact of scaling to many cores.

 

[MayaUser]

“Products have poor longevity when their successors are significantly better” is an absurd take.

For normal user, yes. For prof that significantly better means more projects more money or same money but more free time..its absolutely not.

 

[iSayuSay]

“Products have poor longevity when their successors are significantly better” is an absurd take.

I said what I said, because you can't afford to buy a wrong Mac? Everything is built in and unified, so you're stuck with whatever you buy.

My PC rig has some of degree of freedom even if I mess up at first, I can add/switch a few things later on if I feel something is too slow. Add a bigger power supply, switch to bigger RAM, switch my GPU, switch my CPU (I love AMD for their prolonged socket support).

Not much to do when I felt a Mac Mini is becoming too old and slow, or replaced by something smaller yet faster like the M4, I'd be jealous but can't do nothing about it. While I love and own many of Apple mobile devices, I cannot say the same for Mac development.

 

[Kylo83]

But 99% of people will find M1 Max is fast enough, so you save 23 seconds on an export

 

[Geoffb9]

Because it creates confusion as when should I buy a Mac? Imagine spending $4000 for the most expensive, fully maxxed out dream setup Mac and just a year later Apple can squeeze the same performance on a base line Mac Mini?

Have you consider the fact that you’re the confused party here

starting when Mac users care about the benchmarks and score and specs so much? another sign of just another IBM

its called science. Benchmarking is part of that. Deal with it.

 

[Geoffb9]

I said what I said, because you can't afford to buy a wrong Mac? Everything is built in and unified, so you're stuck with whatever you buy.

My PC rig has some of degree of freedom even if I mess up at first, I can add/switch a few things later on if I feel something is too slow. Add a bigger power supply, switch to bigger RAM, switch my GPU, switch my CPU (I love AMD for their prolonged socket support).

Not much to do when I felt a Mac Mini is becoming too old and slow, or replaced by something smaller yet faster like the M4, I'd be jealous but can't do nothing about it. While I love and own many of Apple mobile devices, I cannot say the same for Mac development.

You seem you’re the one who’s happy with using Windows XP with DVD and USB 1.1 for 15 years.
If it was for you, people would still be using horses for traveling.

 

[Cervisia]

Yeez, people writing about hardware really should read the GB6 docs. M4 Pro got a higher score because it's higher single core performance and because the benchmark intentionally doesn't scale well with core number. In any task people usually buy an ultra chip - besides running GB6 - the M2 Ultra will be definitely faster.

 

[Plutonius]

Some people need the absolute fastest but, for many people, is a few seconds quicker worth spending a lot of money for an upgrade ?

 

[Cervisia]

That's an impressive feat to be sure, but the more impressive achievement if you ask me is comparing those numbers to non-M-series CPUs.

If those M3 Pro benchmarks are indeed real (which they probably are, based on core count and M4 performance in the iPad), then that means Apple's current midrange laptop CPU is slightly faster than the i9-14900K, Intel's current top-of-line consumer desktop CPU--a $600, 125W base power, 253W max power, beast of a chip.

Intel has made significant gains in the past 3 generations, but having your top-of-line enthusiast desktop CPU still bested by a low-pro laptop CPU speaks to how far ahead of the game Apple was for a while there.

The Ryzen 9 9950X, AMD's consumer desktop top-of-line (with a 170W base, 230W peak power draw), is I believe a bit faster than the i9-14900K, but I believe still comes in a hair behind these M4 Pro results.


...Then there's the M4 Max; with two more performance cores and double the memory bandwidth, it should turn in scores significantly higher unless there's something really wrong with it. The difference between the 10-core and 12-core M3 Max is not proportional to the core speed, with the two cores only adding about 10% to the multicore results. If the M4 Max is the same, it will presumably score around 25,000 on Geekbench, if the extra media engine processors don't boost that. If Apple has managed to do better with its thermal management (I assume that's the bottleneck on the M3 Max), and/or the beefier media unit and extra memory bandwidth have a disproportionate impact, it wouldn't be shocking to see it manage above 27,000.

Looking past the gamer end of Intel and AMD's chart and into workstation CPUs, there's the AMD Ryzen Threadripper PRO 7995WX, with 96 cores and a monstrous 350W power draw. I think that's the current record-holder for CPU performance on Geekbench (it's not on their CPUs leaderboard officially yet), coming in somewhere around 28,000 or so; it'll be really interesting to see if the M4 Max can get up that high.

Intel's competing high-end-workstation W-series seems to be harder to find on the Geekbench browser, but as far as I can tell doesn't perform that much higher than the i9-14900K for the kind of tasks that Geekbench tests. The Threadripper is being marketed as a Workstation CPU, not a server CPU like the Xeon or EPYC, so it arguably should be in rough competition with Apple's Ultra series CPUs (even if those don't have ECC and presumably some of the other "workstation"-specific CPU features apart from the extra rendering engines). The M4 Max equaling it would be quite a feat.

If Apple ships an M4 Ultra that manages even the relatively poor core-increase-to-performance-increase of the M2 Max vs. M2 Ultra, that hypothetical M4 Ultra would be up in the 35,000 range and as far as I know completely without equal.

It won't be there, for the same reason the lower chips can beat top end Amd and Intel chips: Geekbench 6 doesn't scale well. It's intentional, and was a change made with good intentions, if only texhnwriters would understand and point out this when they talk about geekbench results.

 

[nieks]

buying an expensive tool, like a fully maxxed out Mac computer, only to get nerfed by a baseline, rapidly changing Mac a year later, is a good investment? 😊

Yes. You buy a computer when you need a computer. You buy that computer with specs that fulfill the needs you have.
If the next year a better computer comes out, does that make your buy a bad investment? No it does not, because your computer still serves you in the way it was intended.

That's the problem, there's no such thing as "longevity" on these recent Macs when Apple practically bashing the last gen M chip by a huge margin, year over year. Just keep buying the newest baseline Mac because it'll be faster than last gen M Pro chip, which may only got out 6 months earlier (i,e M3 Pro chip vs. baseline M4). I hope you get the idea.

Again: buy what you need, when you need it. In that case, it is always a good investment. Also: don't look at your wallet alone, also look at your ecological footprint. It is better for our planet to not buy a new computer every year, just to be the fastest.

Buy what you need, do not buy something because of the fear of missing out. All your replies scream "I've got FOMO and Apple is bad because they make too much progress". PSA: Apple will continue to progress, sometimes big steps, sometimes small steps. Sometimes they will take longer, sometimes they will take shorter. For instance: the step from M2 to M3 was not as impactful as the step from M3 to M4.

Well I'm not used to that concept. If I buy a top-of-the-line machine, I'd expect it to stay that way at least for a while. Not for bragging rights, but knowing that I can enjoy my purchase and use it for my edits and renders faster than most people.

Again: you buy a computer for what you need. If you think that what you need is 'to always be the fastest', you've got your priorities wrong. Buy a computer that succesfully exports your projects, but like it is with ALL technology: be prepared, there will come faster successors and faster competition. Does that make your device redundant? Or a bad buy? Or complete crap? Of course not. You've still got something that's top of the line, that still does your work, that still produces the results you want. It's just not the best anymore. It was, but other (newer) devices are better. Thats just how progress works. You're just disgruntled that Apple makes an effort and progresses faster than you like.

Hence, I said this is a bad time to get an M series chip.

I get the feeling that you're not gonna buy a Mac when the development stalls, because by then you'll think the M-series is a stalling platform and will go the way of Intel: never ever going forward anymore. You will probablly think it is a bad time to buy an M-series chip by then.

We don't have to agree with each other

But you sure go out of your way to counter every reply that does not agree with you. Your disclaimers (i've read more than one) don't make it okay to keep going on and continuously repeat your arguments. We get it: you're not going to upgrade, please stop telling us why. By now, everybody knows.

(PS: you don't have to agree with me )

 

[cocoua]

This no sense price disparity was public knowledge since the May rumor that until 2025 spring wouldnt be released the MS/MacPro M4 max/ultra.

Buying the so called “MacPro” right now means paying 5000k just for the Pci lanes.

In 6 months from here, you must be a fool buying a brand new MSMax M2

 

[MayaUser]

Yeez, people writing about hardware really should read the GB6 docs. M4 Pro got a higher score because it's higher single core performance and because the benchmark intentionally doesn't scale well with core number. In any task people usually buy an ultra chip - besides running GB6 - the M2 Ultra will be definitely faster.

Definitely not in single core perf. For multicre we have to wait for real work applications. In the gpu department m2 ultra will definitely be faster expect where ray tracing is fully supported