M4+ Chip Generation - Speculation Megathread [MERGED]

[Populus]

I just learned a few ways to use my iPad as the main screen for a Mac mini…

So… yeah, now I’m waiting for the M5 Mac mini, definitely.

 

[DrWojtek]

The GPU improvement from M4 -> M5 seems bigger than M4 Pro -> M5 Pro? Or did I interpret that wrongly?

The CPU improvement in multicore M4 Pro -> M5 Pro is massive though.

 

[Pressure]

The GPU improvement from M4 -> M5 seems bigger than M4 Pro -> M5 Pro? Or did I interpret that wrongly?

The CPU improvement in multicore M4 Pro -> M5 Pro is massive though.

That was also accompanied by a big memory bandwidth increase.

It will depend on GPU workload though.

 

[innerproduct]

In my own tests on custom SW (develped on MBP M3 MAX) mem bandwidth often is saturated quite early with few threads so in that perspective in makes sense to have mid-cores making up the bulk of the cores. So for my specfic sw I would not be surprised to see a massive uplift on common tasks. (Think composition SW like nuke but simpler and for another purpose) On the other hand, for true embarrassingly parallel tasks like CPU rendering the improvement might not be that impressive. I wonder if an ultra chip has to be a 12p+24m core (I will not user their marketing terms!) machine or if there could be other arrangements?

 

[chars1ub0w]

In my own tests on custom SW (develped on MBP M3 MAX) mem bandwidth often is saturated quite early with few threads so in that perspective in makes sense to have mid-cores making up the bulk of the cores. So for my specfic sw I would not be surprised to see a massive uplift on common tasks. (Think composition SW like nuke but simpler and for another purpose) On the other hand, for true embarrassingly parallel tasks like CPU rendering the improvement might not be that impressive. I wonder if an ultra chip has to be a 12p+24m core (I will not user their marketing terms!) machine or if there could be other arrangements?

In another thread, someone pointed out that memory bandwidth is not saturated by CPU tasks. So the Pro and Max, when having the same P+E or C+P cores, have the same performance on CPU tasks despite the Max having *twice* the memory controllers (and memory bandwidth). And old benchmarks bear this out. Only the GPU cores takes advantage of the extra bandwidth. If one's code is CPU, like one I'm currently running, one can save $900 with NO loss of performance by opting for a M5 Pro over the corresponding M5 Max.

 

[innerproduct]

for CPU the actual mem BW usable for CPU is about 180 GB/s On the M4 that was actually about 300GB/s. this new m5 I have no idea about but I would not be surprised if it was around 400. Fun fact is that for the m3 ultra actual CPU mem BW is just slightly higher than the m4 max. GPU is another thing that I have less experience on

 

[crazy dave]

for CPU the actual mem BW usable for CPU is about 180 GB/s On the M4 that was actually about 300GB/s. this new m5 I have no idea about but I would not be surprised if it was around 400. Fun fact is that for the m3 ultra actual CPU mem BW is just slightly higher than the m4 max. GPU is another thing that I have less experience on

Is that M4 Max CPU that was 300GB/s? And your prediction of 400GB/s is for the M5 Max CPU BW?

That was also accompanied by a big memory bandwidth increase.

It will depend on GPU workload though.

Are the bandwidth increases not the same? I haven't checked but they should be, right?

 

[crazy dave]

See, my internal struggle between the mini and the air, is the following: for my academic tasks, my iPad Pro is enough 95% of the times, and that’s what I carry with me. For commute: iPad and iPhone. For working from the countryside , having a mini isn’t a problem either: I could grab it, put it on the suitcase and when I arrive there, I have a monitor and a TV, plus another set of keyboard and mouse.

However, some people are encouraging me to go abroad to complete my formation or even work… I still haven’t decided it yet, but if I do it, it’s one of this two things: either go with a MacBook Air, Which is the most convenient solution, or I grab an 80€ HDMI Chinese screen from Amazon just to be able to boot into my Mac mini and then pair it via sidecar with my iPad Pro. Those are my two options.

I cannot play the card of buying and selling in a year, because I really want the top RAM configuration, and that makes it more complicated to sell it without losing a lot of money…

Anyway, I don’t want to extend the off-topic much more.

Notebookcheck just published their 13" M5 MBA review (no energy measurement for CB R24 yet, sadly for me). The M5 in the 13" MBA is indeed slower than the M5 in the MBP for gaming and productivity, but it might still work for you as it's still pretty good:

Insane performance and efficiency without fans - Apple MacBook Air 13 M5 Entry Review

Notebookcheck review of the Apple MacBook Air 13 M5 with 16 GB RAM, 512 GB SSD and Wi-Fi 7.

www.notebookcheck.net

vs.

Apple MacBook Pro M5 (2025) review - The fastest single-core performance in the world

Notebookcheck has tested the Apple MacBook Pro 14 M5 with the new M5 SoC, 32 GB RAM and a nano-texture glass mini-LED display.

www.notebookcheck.net

I take the entry MBP is a no-go?

 

[Populus]

Notebookcheck just published their 13" M5 MBA review (no energy measurement for CB R24 yet, sadly for me). The M5 in the 13" MBA is indeed slower than the M5 in the MBP for gaming and productivity, but it might still work for you as it's still pretty good:

Insane performance and efficiency without fans - Apple MacBook Air 13 M5 Entry Review

Notebookcheck review of the Apple MacBook Air 13 M5 with 16 GB RAM, 512 GB SSD and Wi-Fi 7.

www.notebookcheck.net

vs.

Apple MacBook Pro M5 (2025) review - The fastest single-core performance in the world

Notebookcheck has tested the Apple MacBook Pro 14 M5 with the new M5 SoC, 32 GB RAM and a nano-texture glass mini-LED display.

www.notebookcheck.net

I take the entry MBP is a no-go?

It’s a no go because the lighter the better, that’s why I was hoping the MacBook Neo to be the new 12” MacBook, but it’s not: only 8GB of RAM and same weight.

Anyway, maybe I’ll just grab a base MBA and use it for a year, then sell it and get what I think is the best, either the M6 MacBook Air, or a mini.

 

[yuda398]

Ah, I think I understand what's confusing people.
Consider the Pro class of machine. That has four of let's call them memory controllers. (A has 1, M has 2.)
If we take the base memory bandwidth of the available DRAM as, say, 50 GB/s, that gives our Pro 200 GB/s.
All easily understood.

BUT
when it comes to segmentation, starting with the M3, Apple added a wrinkle. The lower Pro is deficient in THREE ways
- you lose CPU cores
- you lose GPU cores AND
- you lose a memory controller.
(If you understand the issue, this is not quite trivial. You now need to hash addresses across three slices, not four, so you can't just hash by two physical address bits fairly high in the address. Apple actually have a patent for a circuit to do the relevant hashing.)

I THINK at the M3 generation they didn't do with the Max.
But with the M4 they did. Look at the M4 Pro and Max (if the Apple pages are still up.)
In both cases the lesser version
- loses CPU cores
- loses GPU cores AND
- has 3/4 memory bandwidth because one of the four (or two of the eight) memory controllers are disabled.

Does this extend to also losing a quarter of SLC capacity? I believe it does – in terms of HW (and the memory hashing) that's really the only option that makes sense.
To the extent that this is driven by yield binning (rather than mostly driven by market segmentation) that also makes sense. The area of an SLC slice is reasonably larger, larger than the memory controller, so if something's going to fail because of yield bad luck, it's more likely to be the SLC slice than the associated memory controller.

M5 Pro and Max are just following this tradition.

I think the binned M4/M3 Pro’s haven’t lose memory controllers only the Max’s have. All the M3 Pro variants have 3 controllers and all the M4 have 4

 

[DrWojtek]

Why do you think they use this staggered release of Macs? It's mildly annoying. Like now: If the M5 mini was available, it would be a cheap way to get M5 even though I won't need it. But when the mini comes, in a few months probably, M6 is not far away. I'll hold off and won't get it. Wth apple?

 

[Basic75]

Why do you think they use this staggered release of Macs? It's mildly annoying. Like now: If the M5 mini was available, it would be a cheap way to get M5 even though I won't need it. But when the mini comes, in a few months probably, M6 is not far away. I'll hold off and won't get it. Wth apple?

You can't do pipelining if everything is at the same stage. And as for the customers, most are interested in buying something that does what they need when they need it, without bothering about better tools being available now or sometime in the future. When your hobby consists of buying the greatest even when whatever you have is enough you're playing a different game. If you need a Mac mini now, buy it now.

 

[innerproduct]

So m5 max reviews out now. mem bw for cpu about 350GB/s for best case stream like triad ops (basically "image compose"). MT uplift about as expected it seems. neural accelerators showing that 3X+ uplift where it can be applied like prefill for dense models.
All in all a solid upgrade. I do really hope to see this doubled for the ultra class, that would be a really solid desktop.

 

[diamond.g]

Notebookchecks CP2077 1440P results seem off, no way the M5 Max is only 1 FPS faster than the M4 Max.

 

[Populus]

Hey! Do we already have reviews of the M5 Pro and M5 Max SoC? I’m curious to see the chip exposed on a teardown…

 

[name99]

I think the binned M4/M3 Pro’s haven’t lose memory controllers only the Max’s have. All the M3 Pro variants have 3 controllers and all the M4 have 4

Maybe? I lose track of which precise detail goes with which precise model and only remember the big picture.
Very LLM hallucination...

 

[Basic75]

I think the binned M4/M3 Pro’s haven’t lose memory controllers only the Max’s have. All the M3 Pro variants have 3 controllers and all the M4 have 4

• All M1, M2, M3, M4, M5 bins have 8 memory controllers (128 bits).
• All M1 Pro and M2 Pro bins have 16 memory controllers (256 bits).
• All M3 Pro bins have 12 memory controllers (192 bits).
• All M4 Pro and M5 Pro bins have 16 memory controllers (256 bits).
• All M1 Max and M2 Max bins have 32 memory controllers (512 bits).
• M3 Max, M4 Max and M5 Max "lower" bin have 24 memory controllers (384 bits).
• M3 Max, M4 Max and M5 Max "upper" bin have 32 memory controllers (512 bits).

 

[wmy5]

When can we see a die shot of M5 Pro and M5 Max? Can someone volunteer to teardown their newly delivered MacBook Pro, please 🙏🙏🙏🙏🙏🙏🙏

 

[chars1ub0w]

And as for the customers, most are interested in buying something that does what they need when they need it, without bothering about better tools being available now or sometime in the future.

Well, if a new generation is to be released within a few weeks or months with no price increase, it makes really good sense to hang onto the existing computer for just a little while longer.

(Of course, if you lost a computer (or phone) yesterday and have to replace it right away, you have to take whatever configuration happens to be in stock at the Apple store right now, even next week might be too far away. Been there, done that while traveling.)

 

[Pressure]

Blender performance/watt for M5 Max is looking pretty solid.

 

[Populus]

The M5 Pro and M5 Max have separate dies for both the CPU and the GPU, right? Do you think they will be encapsulated together?

Do you think they will engineer two heatipes or dissipators, one for the CPU and a bigger one for the GPU? Maybe that’s not possible for the MacBook Pro, but for the Mac mini and Mac Studio…?

I mean, with those neural accelerators, the 20/40 cores GPU will get very hot in machine learning tasks… and if they share the heatpipe, the heat could be transferred to the CPU, which is usually not as hot, right?

 

[innerproduct]

Blender performance/watt for M5 Max is looking pretty solid.

View attachment 2612615

Indeed, the results align well with expectations and makes this a very tempting chip. Sadly only available in a laptop form factor atm making sustained workloads less viable due to cooling and noise.
What possible combos will we see for ultra level chips and will the new way of producing reduce time until it gets to market? WWDC is what I am hoping for but I guess they will wait until October.
what are the current rumors pointing at? anyone know?

 

[Populus]

WWDC is what I am hoping for but I guess they will wait until October.
what are the current rumors pointing at? anyone know?

Honestly, no idea, but I really hope Apple releases the Mac mini and the Mac Studio before WWDC. October sounds so distant…

 

[diamond.g]

The M5 Pro and M5 Max have separate dies for both the CPU and the GPU, right? Do you think they will be encapsulated together?

Do you think they will engineer two heatipes or dissipators, one for the CPU and a bigger one for the GPU? Maybe that’s not possible for the MacBook Pro, but for the Mac mini and Mac Studio…?

I mean, with those neural accelerators, the 20/40 cores GPU will get very hot in machine learning tasks… and if they share the heatpipe, the heat could be transferred to the CPU, which is usually not as hot, right?

If they share the same heatspreader/TIM does that really matter? AFAIK they are not direct die cooling anything right?

 

[Tdude96]

WWDC is what I am hoping for but I guess they will wait until October.
what are the current rumors pointing at? anyone know?

I'm definitely hoping for WWDC and that Apple doesn't pull another dirty trick with the Ultra by sticking it with an obsolete chip, but MacRumors has been hedging on that over the last couple of weeks, no longer saying M5 Ultra and instead phrasing it "M4 or M5 Ultra" which is disturbing. I know people are saying M4 Max doesn't have the proper interconnect, but M3 Max didn't either when it released.

I think WWDC is our best chance of seeing one this year, but timing is tight if they stick to what they've done in the past.

MacBook Pro M1 Max -> Mac Studio M1 Max/Ultra = 4 months, 23 days
MacBook Pro M2 Max -> Mac Studio M2 Max/Ultra = 4 months, 20 days
MacBook Pro M4 Max -> Mac Studio M4 Max/M3 Ultra = 4 months, 4 days

Anytime they've released a Max chip in a MBP and followed it up with a refresh for the Studio with that same Max chip, it's been about 4-5 months. WWDC would be closer to 3 months. I suppose there could be a late June/early July press release since it's just supposed to be a chip upgrade, no design changes.

Of course, Apple's free to change their timing on things whenever they please, but I just don't see an October release for a Studio. I think if we don't see one by Summer, it'll be pushed to 2027, but both October and 2027 seem crazy far away when it seems like inventory of the Studio is drying up now. So to me, WWDC is looking really good.