20-30% is hardly detectable in normal use. And power users will use the Pro anyway...
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M4 Max Chip Up to 25% Faster Than M2 Ultra in First Benchmark Results
- Thread starter MacRumors
- Start date
- Sort by reaction score
This is why future-proofing your machine is a worse approach than simply buying the base model closest to your respective needs and then trading-in / upgrading every few years.
The gains you get from the new chips are going to quickly out-perform the thousands of dollars extra you spend in upgrades.
The gains you get from the new chips are going to quickly out-perform the thousands of dollars extra you spend in upgrades.
Yes. As of now, there is a distinction between the Pro and Max. The Max being the ultimate mobile SoC for Data analytics and Ai workflows. M4 Pro full 14-core will be the perfect medium.
Each generation of the "M" chips (each with four variations) have had nearly the same single core speeds. Most folks probably live in the single core world of booting up, reading email, spread sheets, look at photos etc. The first three generations had modest improvements in single core speeds at each generation. Now the M4 generation finally nearly doubles the M1 single core speeds.
I want a few thousand units to be built and actual numbers published before ordering my CTO 16" MacBook Pro MAX with 128GB of ram and 8TB SSD. With the proper hub, it could easily replace both my M1 MAX 14" MacBook Pro (64GBRam and 4TB SSD) and M1 Ultra Studio (128GB Ram and 8TB SSD) driving my two Studio Displays when in my office and all the accessories.
I want a few thousand units to be built and actual numbers published before ordering my CTO 16" MacBook Pro MAX with 128GB of ram and 8TB SSD. With the proper hub, it could easily replace both my M1 MAX 14" MacBook Pro (64GBRam and 4TB SSD) and M1 Ultra Studio (128GB Ram and 8TB SSD) driving my two Studio Displays when in my office and all the accessories.
For 3D rendering apps (and also games):
If you go to geekbech and look for M4 Max, you can also check the GPU score: A very nice 192532.
For comparision, this GPU score is:
7% slower than M2 Ultra
20% faster than M1 Ultra
35% faster than M3 Max
45% faster than M2 Max
75% faster than M1 Max
Let's wait for raytracing benchmarks, since Apple stated it doubled raytracing performance.
If you go to geekbech and look for M4 Max, you can also check the GPU score: A very nice 192532.
For comparision, this GPU score is:
7% slower than M2 Ultra
20% faster than M1 Ultra
35% faster than M3 Max
45% faster than M2 Max
75% faster than M1 Max
Let's wait for raytracing benchmarks, since Apple stated it doubled raytracing performance.
In have an M2 Ultra and honestly I dont find figures like 25% faster really that impressive even if it is a Max v an Ultra. These M-chip updates improvements are always incremental and often unremarkably incremental, but yes 25% is a better increment than we normally see. I wonder how much better the M4 Ultra will be? I'd like to a 100% or 2x improvement in a new gen M chip, then i'd really sit up and take notice.
I guess it depends on why you are buying an Ultra. If it is just for bragging rights on benchmarks, then yes, doesn't make much sense. But if it is because you earn money with the thing, well it might very well pay for itself in those six months.
Within the US channel, Mac Pro is around 3% of Mac sales and the Mac Studio is 1%. I imagine worldwide the percentages are similar or even lower. So even if half of Mac Studio buyers buy an Ultra configuration, that is 3.5% of all Mac sales. TSMC yields on initial batch M4 Max SoC runs may not be good enough to feed even such a small population considering a lot of folks buying MacBook Pros with the M4 Max are likely going to the higher-tier units that are also needed for M4 Ultra production.
MacBook Pro makes up over 50% of Mac sales in the US channel. Apple is not going to want to lose/defer two MBP sales (especially a high-config 16" Max model) because they do not have the SoCs available as they're parked in a Mac Studio Ultra or Mac Pro. Also, people replace desktops on a longer time frame than they do portables so the number of people jumping on a Mac Pro or Mac Studio Ultra at SoC launch would be far lower.
Right. But The Pro line means full house (ram, ssd, gpu etc) as well. Plus; at the price tag; it has to be needed for high tasks. I personally think, apple is selling not much more than a few thousands of that M Pro macs a year. Who needs and can use that tech at 100%... And the congruence does not sleep 😎
Honestly, I don't see the point of replacing him, especially if he plans to change chassis next year.
The Mac Mini price of $1,800 vs $2,400 will be more intriguing for buyers. Apple would most likely see more Mac Mini M4 Pro sales then MacBook Pro.
Seems like the more expensive computers should get the new chips first, then they work their way down to the less expensive models. Why not release the M4 Pro, Max, and Ultra first, then the normal M4?
Doing it this way it seems like the less expensive models are going to keep leapfrogging the more expensive ones. I know there are other things that make them more expensive, but it just seems like bad marketing to have your most expensive models with the oldest processors.
Doing it this way it seems like the less expensive models are going to keep leapfrogging the more expensive ones. I know there are other things that make them more expensive, but it just seems like bad marketing to have your most expensive models with the oldest processors.
I don't think that the aluminum can can cool down an M Pro. If so, you would have to screw it to the table so that it doesn't take off.😂
Note that this is the M2 ultra, and it is 2 generations of CPU (and almost 18 months) for the mid-tier to pass it. Also - the generation skip between M3/M4 was faster than normal. We can only speculate that it was because the M4 work was progressing so well and perhaps the architectural security flaw found in M1-M3, that they just decided to accelerate the M4 (along with the node process bump)
It just depends on what you need, today. If you can wait, there's always a better deal coming, whether that's today's high end discounted or tomorrow's mid-tier offering.
i'd actually be more interested in GPU Benchmarks in comparison to the M2 38 Core GPU ,
This exactly. I get the feeling Apple user culture has cultivated an intense feeling of FOMO (Fear Of Missing Out) in its user base over the years which I think has been a by product of the iPhone's yearly release cycle, and it has started to creep into their other product lines like their laptops... laptops are not throwaway items, nor are they an object to be upgraded every single year. I have an M1 Max MacBook Pro and it easily completes all the tasks I need it to do in the timeframes I need it to do them, and quite frankly it would be a poor lookout for a laptop if it didn't after such a relatively short period of time since its release.
Honestly, what a great problem to have. That in just 18-24 months or so, the more mainstream processors are meeting-or-exceeding that older top-of-the-line chip.
This doesn't instantly make those older top-of-the-line processors slower. Just means that if that power was adequate for the workload presented, when that person goes to upgrade, they now have more choices and don't have to grab another Ultra chip model...unless they want extra seconds off their compile, export, whatever.
Now, buying the top-of-the-line chip to only say you have the fastest, without actually using that horsepower, that's a bit pedantic. Again, new chips coming out do not make the existing chips slower relative to their own performance profile.
This doesn't instantly make those older top-of-the-line processors slower. Just means that if that power was adequate for the workload presented, when that person goes to upgrade, they now have more choices and don't have to grab another Ultra chip model...unless they want extra seconds off their compile, export, whatever.
Now, buying the top-of-the-line chip to only say you have the fastest, without actually using that horsepower, that's a bit pedantic. Again, new chips coming out do not make the existing chips slower relative to their own performance profile.
people complaining about last year's top of the line being too slow now. Thats not true. they have same speed as always. The only thing that changed is that these people now know there are better products out there that they cant afford (or just cant justify getting). The fast pace is a good thing. These people just need better income to be able to stay at the top tier
I believe the same!
But I wonder to what extent Apple will get future graphics power in the Mini? The small Mini design will eventually cause problems with cooling, right?
Of course, it's possible that in the future there will be a huge difference in the (graphics) performance of an M5 in the Studio and an M5 in the Mini.
How do the graphics of the M4 perform in the Mac Mini? Does anyone have a benchmark vs. AMD/nvidia?
In terms of CPU/GPU performance, maybe. The problem is that you can't upgrade RAM or SSD after purchase, so if those are what you need to future-proof then it makes sense to buy what you need now, then hold onto it for longer. If all you need is more speed, then it makes sense to buy the base models and upgrade frequently.
As pointed out earlier, Geekbench in particular intentionally doesn't scale very strongly with core count, to reflect most people's real-world use. There are still use cases where an M2 Ultra would be more performant CPU-wise than an M4 Max, in situations where all cores can be fully loaded.
However it does highlight a weakness with Apple's strategy of doubling up Max chips to make an Ultra, in terms of pure CPU performance. An M4 Max has 16 cores, vs an M2 Ultra's 24, but of those cores, 12 are performance cores on the M4 Max whereas 16 are performance cores on the M2 Ultra - in other words, the difference is not as large as it might initially appear.
I suspect Apple will move away from this double-chip setup for the Ultra chips because, by definition it forces an unnecessary and probably unwanted doubling in efficiency cores vs. the Max chips, skewing the efficiency/performance core balance in the opposite direction to the one you'd actually want for a high powered desktop chip. This will also keep the Ultra chips more clearly differentiated from the Max chips and prevent this sort of strange overlap we now have.
We'll know once we get clear die shots of the M4 Max - I'm guessing it won't have the UltraFusion interconnect on it.
However it does highlight a weakness with Apple's strategy of doubling up Max chips to make an Ultra, in terms of pure CPU performance. An M4 Max has 16 cores, vs an M2 Ultra's 24, but of those cores, 12 are performance cores on the M4 Max whereas 16 are performance cores on the M2 Ultra - in other words, the difference is not as large as it might initially appear.
I suspect Apple will move away from this double-chip setup for the Ultra chips because, by definition it forces an unnecessary and probably unwanted doubling in efficiency cores vs. the Max chips, skewing the efficiency/performance core balance in the opposite direction to the one you'd actually want for a high powered desktop chip. This will also keep the Ultra chips more clearly differentiated from the Max chips and prevent this sort of strange overlap we now have.
We'll know once we get clear die shots of the M4 Max - I'm guessing it won't have the UltraFusion interconnect on it.