Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I don’t think so.

I remember when M2 was out. A lot of people considered that all M1 machines were a better deal.

We expect the same with M3, but seems the 3nm process made those chips not as good as they should be. So, there is the M4.
 


Apple is planning to launch the first M4 Macs in early November, and new benchmarks for one of the upcoming models surfaced on Geekbench, giving us some insight into how the M4 Mac will perform compared to the M4 iPad Pro and prior-generation M3 Macs.

M4-Real-Feature-Red.jpg

The "Mac16,1" that was benchmarked features a 10-core CPU and it earned a single-core score of 3864 and a multi-core score of 15288.

Comparatively, the base M4 iPad Pro with 9-core CPU has an aggregate single-core score of 3647, and a multi-core score of 13135. The iPad Pro models have either a 9-core CPU or a 10-core CPU depending on storage capacity, and 10-core models have higher multi-core scores of around 14500 on Geekbench.

When pitting the M4 against the M3 chip with 8-core CPU and 10-core GPU that was used in last year's Macs, single-core performance is up 26.7 percent, and multi-core performance is up 30.6 percent. The M3 iMac, for example, has a single-core score of 3048, and a multi-core score of 11708.

In the iPad Pro, Apple said the M4 was up to 25 percent faster than the M3, and M4 performance in Macs will be the same or better. Note that this is just one benchmark, so there will be some variability in scores when the machines are actually released.

Metal scores for one of the new M4 Macs also surfaced today, and GPU performance is roughly on par with the M4 iPad Pro chip. Mac16,1, which has 16GB RAM, earned a Metal score of 57603. The M4 iPad Pro has a Metal score of 53374. The M4 Macs will likely perform a little bit better due to thermal differences, but again, there is variance in scores.

The M4 chip showing up on Geekbench is the standard M4, and it will be used in the Mac mini and 14-inch MacBook Pro models. We are also expecting to see M4 Pro and M4 Max chips for the higher-end MacBook Pro and Mac mini models, and benchmark results for those machines have not yet surfaced. The M3 chip had a Metal score of 47414, suggesting GPU performance will improve around 20 percent year-over-year.

Apple will likely hold a small event to announce the new M4 Macs toward the end of October, with a launch to follow on November 1.

Article Link: Mac M4 Chip Performance Unveiled in First Benchmarks
27% jump in single core on the M3 is huge, probably the biggest jump year on year since the M1.....strange that multicore is only 30% up, considering the chip has 2 extra cores....
Anyway, I'm happy with my 14" M3 MBP
 
Whether intentional or not, seems like Apple is emulating Intel's old 'Tick, Tock' cycle with the even-numbered M-series chips being the sweet spots: M2 and now the M4.

The reality is: there is a desire to ship new processors at a faster rate than fabs can cost effectively start new processes, so some chip releases aren't going to get a new fab.

Intel tried to regiment this: New Fab -> New Architecture -> New Fab -> etc... but that plan entirely fell apart a decade ago.

Apple not really emulating that, beyond just mitigating fab process node changes.

Apple's still in line for the M5 being about twice as powerful as the M1, which is fine progress.

Buy when you have to buy, wait if you can, be happy with what you get. :)
 
🌟 To Apples silicon design team for continual outstanding gains that are achieved with each iteration of the M processors. Considering both the M3 and M4 soc are based on the 3nm architecture. To have 25% gain is superlative to say the least!
Yes, but a different 3nm process and only the M4 completely redesigned for 3nm.....in fact the M4 is what the M3 should have been
 
in fact the M4 is what the M3 should have been
I don't understand this.

What does "should have been" mean?

Should the M3 not have been released and the people who were happy buying it not been given the option?

Should Apple have magically engineered the M4, on a process that didn't exist at the time, with designs that were still in the works?

Apple isn't pulling these things out of a hat.

This is hard development work that they seem to be doing a great, and speedy, job.

Apple doesn't expect people to buy every release; so comparisons of "upgrade value" on single generation steps just seems silly?
 
I don't understand this.

What does "should have been" mean?

Should the M3 not have been released and the people who were happy buying it not been given the option?

Should Apple have magically engineered the M4, on a process that didn't exist at the time, with designs that were still in the works?

Apple isn't pulling these things out of a hat.

This is hard development work that they seem to be doing a great, and speedy, job.

Apple doesn't expect people to buy every release; so comparisons of "upgrade value" on single generation steps just seems silly?


I've never think like you. It's nice to see things in your way.
Now, I'm encouraged to add: We never but the cutting edge technology. There is aways something better being tested in the lab.
 
Object detection 226% was the one I was looking for. That would look like where the majority of the improvement is.

Which is what I was asking.

I don’t think of one test out of sixteen as “a majority”, but sure. M4/A18 has improved performance across the board. An average of more than 20% is nothing to sneeze at, and it’s not like all the other workloads have improved less than 5%.
 
I don’t think of one test out of sixteen as “a majority”, but sure. M4/A18 has improved performance across the board. An average of more than 20% is nothing to sneeze at, and it’s not like all the other workloads have improved less than 5%.
Well yes it has, and it's also clocked 10% higher.

Multi-core scores are likely due to the M3 in that test having 11 cores vs 9 of the M4.
 
Not sure what you're doing on your machine, but my M2 Pro 16 GB is barely moving already. Eagerly looking forward to an update.
I "only" do RAW photo editing + 4k video editing (.MOV, 10bit 4:2:2) on my machine. Well, occasionally doing Stem Splitter to split audio sources on LogicPro. Plus 3 dozens tabs on Safari, WhatsApp and Telegram running in background simultaneously.

I think your workload and job desks are far heavier than me. Lots of creative professionals and programmers needs a beefier setup than most folks do.
 
I think your workload and job desks are far heavier than me. Lots of creative professionals and programmers needs a beefier setup than most folks do.
Yep, Android Studio alone with an emulator running can eat up all the available memory. And that's not mentioning Xcode and iOS simulators running at the same time. My laptop is literally on life support hitting 10 GB of swap more than regularly. So I'm looking forward to the memory bandwidth of the new M4 Pro chips and the max RAM they will allow. I might be better off with a refurbished M3 Max after all.
 
  • Like
Reactions: SamRyouji
25% supposed gain in 1 year is not so incredible alone! I think M4 will be a very nice update over M3 (MBP is the best laptop you can buy now, and I own a m3 max MBP with satisfaction ) , but honestly not enough to be "so surprised"... in more than 30 years INTC had similar (maybe a little less) year over year update... we need a little more updates in my opinion (screen, ...). ACTUALLY +22%... from official numbers, nice but not super
  • MacBook Pro with M4 Max (40-core GPU): 190,329 (2 results)
  • MacBook Pro with M3 Max (40-core GPU): 154,860 (More than 6,000 results)
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.