Apple has already done something similar when they joined two chips in the Ultra processors. Many people claimed this was no different than what others had done, and they were wrong about that also.
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.
M5 Chip Achieves Impressive Feat in 14-Inch MacBook Pro Speed Test
- Thread starter MacRumors
- Start date
- Sort by reaction score
Apple has already done something similar when they joined two chips in the Ultra processors. Many people claimed this was no different than what others had done, and they were wrong about that also.
...but TSMC are working in close collaboration with Apple, and only making Apple Silicon chips for Apple, so Apple should have more control over planning and scheduling than when Intel, Motorola or IBM were making x86 or PPC chips for a whole bunch customers with different priorities.
Anyway - it even happens between models with the same processor class, so we now have the Air, Mini and iMac still on M4 and the 14" MBP & iPad on M5... It's not like Intel where "Nth generation i5" covered a whole bunch of different chips with various TDPs, clock speeds and GPU options depending on whether it was an utraportable, regular laptop or desktop & some of the combinations needed by particular Macs didn't actually arrive until Intel were advertising the next generation...
What are people using all this horsepower for in these machines that get better every year? I know it's nice to have the latest, but for someone who does video editing and photography, anything after M1 should be sufficient correct? No gaming or anything else, just Davinci Resolve and Lightroom. A good deal on an M4 Pro should be more than enough correct?
For Video and Photo work, i would suggest M3 or higher. Due to 2 reasons: Hardware based AV1 Decoding and Ray Tracing support. Both are missing on M1 and M2 class Chips.
- AV1 Decoding is useful for high resolution video editing and smooth editing playback
- Ray Tracing is useful for graphic editing work.
So yes M4 or M4 Pro should be sufficient.
It depends on the software. Here are some "actual performance in real life" in Blender. Apple's top GPUs are not only close to RTX 40 series, they're as fast as some desktop and laptop GPUs from RTX 50 series.
Unless the current generation was supposed to be the previous generation that was purposefully handicapped for financial reasons.
Last edited:
LOL, Blender is CPU intensive software and you literally admitted that Apple GPU sucks by comparing M3 Ultra to RTX 4070 and 5070. Remember, Apple compared M1 Max to 90 series.
My main computer is a 2021 14 MBP with M1 Pro chip, but i have access to a 2025 MBA with M4 chip…the air m4 is just as powerful and does 99% everything the m1 MBP can do. Just as you said, it’s more than sufficient. Vast majority of people who say they want a pro, don’t actually NEED a pro. They just want the latest and greatest. If I could get an Air with 120hz screen, that would be the ultimate for me
If OpenCL increase from M4 Max -> M5 Max is the same % as M4 > M5, M5 Max will score 147.000. That's on par with the following:
AMD Radeon RX 6800 XT (scores a bit higher)
NVIDIA GeForce RTX 5070 Ti Laptop GPU
NVIDIA GeForce RTX 2080 Ti
AMD Radeon RX 7800 XT (scores a bit lower)
And far above M3 Ultra, which scores 130.000.
Pretty cool for a SoC.
As for Metal score, same rules, M5 Max would claim the lead, yielding around 250.000 points. Number 2 would be the AMD Radeon Pro 6950XT at 238.000 points and M3 Ultra at 229.000.
FYI the power draw from a M5 Max under full load, including everything else such as CPU etc, should be half that of an AMD Radeon RX 6800 XT under full load (excluding extreme torture tests etc).
This effectively means saying ”macOS is not for gaming” is like saying a 6800 XT doesn’t cut it, which is obviously untrue. Perhaps more games will find macOS support now.
AMD Radeon RX 6800 XT (scores a bit higher)
NVIDIA GeForce RTX 5070 Ti Laptop GPU
NVIDIA GeForce RTX 2080 Ti
AMD Radeon RX 7800 XT (scores a bit lower)
And far above M3 Ultra, which scores 130.000.
Pretty cool for a SoC.
As for Metal score, same rules, M5 Max would claim the lead, yielding around 250.000 points. Number 2 would be the AMD Radeon Pro 6950XT at 238.000 points and M3 Ultra at 229.000.
FYI the power draw from a M5 Max under full load, including everything else such as CPU etc, should be half that of an AMD Radeon RX 6800 XT under full load (excluding extreme torture tests etc).
This effectively means saying ”macOS is not for gaming” is like saying a 6800 XT doesn’t cut it, which is obviously untrue. Perhaps more games will find macOS support now.
Last edited:
I was thinking the same thing. Also, what's the point of having a "Pro" iPad when there are very few pro apps that are available? I would love to ditch my MBP for an loaded iPad Pro.
You just can't take anyone serious that bad mouths the M series chips. The performance you get for the power draw is like magic.
It is already on paper.
Have m1 air and m1 pro 14”.
M2 and m3 was meh updates more great for first time arm buyers.
M4 showed great results, but upon looking at numbers, 50% better single core was a wash since m1 works fine. Gpu at 56k lags behind m1 pro at 67k, and stepping down didn’t make sense since i drive 2 4k screens and regular m1 gpu struggles while m1 pro flies.
Now M5. Long story short this thing is 1.9x faster than m1 series in single core (browsing, clicking and opening apps, working in apps before export happens).
It is 50% faster in multi core. It finally surpassed gpu power of m1 pro base model.
Tldr: m5 air will kick mbp 14 m1 pro’s bottom across the board while being a slimmer machine with no fan.
Yet my m1 pro works just fine, so i will update to whatever Mx will happen the year this mac will struggle software wise.
Comparing chip that consume 50W with chip that can draw 700W and complain that it is slow just show that you don't really understand were Apple currently is. I think we finally have some competition to Nvidia dominance. Not really for games, but for AI.
It is kind of funny because Nvidia started with games and ended with AI. Nvidia is not 3T company because of games, it is because of AI chips.
Apple with chiplets can make bigger GPUs than we have seen in M4 family.
Yep same with my M1 Max. Apple will have to do something awfully compelling to get me to upgrade. I mean I definitely overbought, but, by doing so I get outstanding performance for many years. It hasn't dropped off at all. My last Intel Mac was slowing down badly after like 3 years and a total dog at 5 years.
I'll take a good hard look at the M6 MBPs when released, assuming that is in fact the gen with the big hardware redesign as rumored.
I kinda know what will get me according to past examples with Intel Macs.
I sit on latest macos but refuse to do macos26 because nerfing is real. So since i sit on a previous macos, there will be day when it will be an OS issued many years ago. Simply put my old intel mac stopped opening web pages in safari, my software updates started requiring newer os or new installs were totally for newer os.
I hacked it by using chrome, but two years later that thing needed to update to open my banking apps but the update required newer os.
That is when i got a new laptop, even though the last one never glitched.
(edit: This won't matter to most users, probably, but for heavy data operations...)
The ~30% improvement in memory bandwidth (153GB/s compared to 120) is more impressive to me, honestly. Hopefully the better bins will see the same or better percentage improvement from the M4 Pro (273GB/s), etc. The Ryzen AI 300 series only does 128 GB/s, and that's the closest comparison with another desktop-class unified memory model I could think of.*
*My own old pc's GPU does 360 GB/s, but its CPU does under 50, so this is where the unified memory model really shines. Speaking of, I just looked up the top speed for CAMM2 modules (the new hotness for some DDR5 based pcs) and it's something like only 10GB/s? Thus explaining a little why Apple charges so much for the upgrades. Still wish they'd drop the price, but suddenly, buying an M4 Mini (or Pro) with memory upgrades when they go on closeout, after the M5 Mini comes out, makes a lot more sense than just buying a regular Mini. Now if we could only get them to officially support Vulkan or OpenGL...
The ~30% improvement in memory bandwidth (153GB/s compared to 120) is more impressive to me, honestly. Hopefully the better bins will see the same or better percentage improvement from the M4 Pro (273GB/s), etc. The Ryzen AI 300 series only does 128 GB/s, and that's the closest comparison with another desktop-class unified memory model I could think of.*
*My own old pc's GPU does 360 GB/s, but its CPU does under 50, so this is where the unified memory model really shines. Speaking of, I just looked up the top speed for CAMM2 modules (the new hotness for some DDR5 based pcs) and it's something like only 10GB/s? Thus explaining a little why Apple charges so much for the upgrades. Still wish they'd drop the price, but suddenly, buying an M4 Mini (or Pro) with memory upgrades when they go on closeout, after the M5 Mini comes out, makes a lot more sense than just buying a regular Mini. Now if we could only get them to officially support Vulkan or OpenGL...
I threw linux on my 2006 Macbook when it got obsoleted, and kept using it as my only laptop until I finally decided it was too slow in about 2016. If that old Mac of yours is quite a bit newer, you might try doing that and keep it around for web browsing. For Apple Silicon it won't be nearly that easy; I'd personally try Asahi Linux but I don't mind doing a lot of stuff that normal people wouldn't want to bother with.
Good advice! It was a 2013 retina MBP 13 that i sold to student in 2021 to move to M1 Air. I would have kept using it, but kids dropped it and spilled water on it, so just thought it was time to move on.
Though i kept my dell xps 15z (bought on june 2011) with broken hinges (in 2025) to play call of duty from time to time.
In that case you also literally just admitted that Nvidia’s latest 5070 Ti, 5080 M and 5070 Ti M ”suck” in GPU intensive Blender Render because their performance is on par with M3 Ultra and M4 Max.
You also showed that you don’t know much about Blender. Blender is not only CPU intensive. It uses both the CPU and the GPU for different tasks. You use the GPU for fastest rendering in Blender, not the CPU. The benchmark results I posted are for how quickly Cycles can render path tracing samples on the GPU using Optix (Nvidia), Metal (Apple), HIP (AMD) and ONEAPI (Intel).
I never talked about M1 Max and 3090 either. Who cares about a 4-year-old GPU and what Apple said back then? I’m talking about now comparing real-world GPU benchmarks.
You said ”Tell me if M4 series are even close to RTX 40 series” and I showed that M4 Max is not only close but faster than desktop 4070 and laptop 5070 Ti. How does that suddenly ”suck”? You’re not making any sense.
You said ”Apple cant even make a high-end GPU” and I showed that M3 Ultra is faster than desktop 4070 Ti and 4070 Ti Super and almost as fast as desktop 5070 Ti. Those three Nvidia GPUs are all considered as high-end, not mid-range with a price of $750—$850. That makes M3 Ultra a high-end GPU too.
The base M5 is already 60% faster than M4 in the same Blender GPU test. If M5 Max 40c performs as well it would score around 8443. That would make it faster than 5090 M, desktop 4080 and 4080 Super. If it will have more GPU cores, like 48c it would score around 10 000. That would make it faster than desktop 5080 and close to 4090. M5 Ultra would score over 12 000 which would make it as fast as 4090 D or 5090 D. Yeah, that would really "suck".
Last edited:
70 series are not even the best GPU. Where is 5080 or 5090? Ultra series are the only best option for Apple which is only 70 series level. Clearly, you admitted yourself that Apple GPU sucks. If you cant compare with 90 series, then it only proves me right.