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.
Apple Unveils M5 Chip With Next-Generation GPU
- Thread starter MacRumors
- Start date
- Sort by reaction score
Yeh, my comment was more on the CPU side where I'm hoping they've just managed to lose some latency or reduce pipeline bubbles. I use my machine for work (genetics) so the graphics are already more than adequate. But the increase in memory bandwidth does look like a clock speed tweak as it seems too small to reflect an architectural change. [actually google says the clock speed is about the same and the performance gain is from cache size]
The M4 Pro maxes out at 48GB of RAM. The M4 Max minimizes at 32GB or memory.
The system is the "MBP 14" because it is the exact same chassis. Why would they call someone that was the exact same size, shape, and screen something different?
Apple has gotten to point now that the plain Mn version also has three USB-C ports. So it is literally the same encloure with no changes. Likely the MBA is going to stay kneecapped on just two ports.
The reason why is that sharing more parts across more skus lower the costs by building bigger economies of scale. Apple is trying to lower the prices on the MBP 14" family of products.
There is an extremely trivial solution if you need more memory and want exactly the 14" chassis body .... upgrade the internals as you buy it.
One of Apple'a benchmarks for the Pro/Max is DNA sequencing. It is showing similar uplifts as the gaming/rendering benchmarks. Some genertic is more parallel computations.
If just CPU cores only then the compiling benchmark is a reasonable 'rule of thumb' proxy for something that has multiple work threads.
The plain M4 just made it to Thunderbolt 4. How was it going to go to TBv5 in one cycle? The barrier on the M1-M3 was more so in the display controller limitations mandated by the relatively small chip. N5->N3 process shrinks don't impact I/O as well as more pure computational logic circuit. TBv5 isn't going to get magically smaller. And Apple's display controllers are abnormally large. The off-die , exterior signalizing isn't shrinking much at all.
The base M4 MBP just go to point can provision three USB-C style ports. ( the chassis was different before while the 14"/16" jumped to three ports. )
Besides this gives the Pro/Max product segmentation on ports.
Bigger dies with larger area budgets have more flexibility. That isn't going to change for immediate future.
It makes me wonder why Apple doesn't create a gaming studio like they did with Apple TV+. Though I guess Apple TV+ is device agnostic, whereas the available market for Apple only games is limited.
The Microsoft that has been laying off folks from their game studios? That Microsoft?
Apple doesn't "have to" do in house game development. The more Apple competes directly against its developments there is an increased chance of 'Sherlock'-ing their partners.
Microsoft was buying those more so for revenue growth. Now that the AI bubble has arrived that likel will completely stop because they can get the revenue (and stock price) growth elsewhere.
Buying Vision only games isn't going to help the Vision Pro. Primarily it not intended to being primarily a gaming 'thing'. So it isn't a bail out. That would just be digging a even deeper hole to get out of.
The is a deep mindset that VR== Gaming. Similar to the AI==chatbot . Neither one is true (outside the hype bubble.)
None of that is surprising. The release schedule has been widely discussed on this website and many others for at least a couple of months.
I thought it was an interesting choice to roll out the M5 in the 14-inch MBP.
The M5 is a 3nm processor manufactured by TSMC... they must need to increase the yields before they can roll out the usual M5 Pro, Max, and Ultra variants.
The MBP (and iPad Pro and Vision Pro) allow them to sell M5 products at premium prices and they can put the M5 in the lower-margin products later when volume production is at full speed (Mac Mini, iMac, MacBook Air, iPad Air, iPad Mini, etc.).
If they can hold the prices to reasonable levels, the M5 Mac Mini will become another value leader in 2026.
The M5 is a 3nm processor manufactured by TSMC... they must need to increase the yields before they can roll out the usual M5 Pro, Max, and Ultra variants.
The MBP (and iPad Pro and Vision Pro) allow them to sell M5 products at premium prices and they can put the M5 in the lower-margin products later when volume production is at full speed (Mac Mini, iMac, MacBook Air, iPad Air, iPad Mini, etc.).
If they can hold the prices to reasonable levels, the M5 Mac Mini will become another value leader in 2026.
Curious to see what these GPU improvements look like. Is it that much faster across the board or only for specific stuff like ray tracing that not many games use and would still be slower than rasterization? If I get a non-Pro-chip Pro to replace my 14” Pro M1 Pro, I’d at least want the GPU to surpass it by more than a little bit. The M4 looked like it came close, but maybe the M5 goes further.
I’ll probably wait for next year’s OLED model anyway, but the price would probably sting less if I can get by with the vanilla M6
I’ll probably wait for next year’s OLED model anyway, but the price would probably sting less if I can get by with the vanilla M6
New product day. And there's absolutely nothing today I want to buy, or more importantly, need to buy.
Roll on 2026 and new desktop Macs and monitors. That might make me move from "vaguely curious" to "possibly interested".
Roll on 2026 and new desktop Macs and monitors. That might make me move from "vaguely curious" to "possibly interested".
And that's more or less how it should be. It would be silly for most to upgrade yearly. Instead, most reach a point where finally their current hardware is too much a liability compared to newer machines and so they upgrade. Many computers these days are good enough that it could take a few or several years before that day comes. A lot like buying a car has been.
I still have a Mac as my primary computer, but I frankly don't use it much anymore. I do have a PC and consoles for gaming.
Frankly, despite the power, I'm not sure I would ever consider a Mac for gaming. If I want a hassle-free TV experience for the family I'll buy on PlayStation or Switch, which is a fixed hardware platform for developers to optimise for. Otherwise I'll buy on PC, which I can piecemeal upgrade every few years -- only recently put in a new GPU -- and which runs about every game under the sun, emulation and all.
A Mac for gaming is about as if not more expensive than a PC, while not really getting better overall performance (for now at least), and I can't upgrade it when it's getting slower.
Oh I know. I'm a fan of yearly incremental updates, and I'm not a fan of people complaining about yearly incremental updates being minor. What do you expect from an incremental update, do you expect companies to reinvent the wheel once and year., every year?
But I'm not in the market for a laptop or iPad, as I'm already happy with want I have. And I have no use for a Vision Pro. The only thing from Apple that would make me part with more money than planned would be desktop-base stuff.
I'm just slightly disappointed (though not overly surprised) there was absolutely no desktop stuff, so it'll be spring 2026 before I think about investing in new hardware.
Apple lags in bringing AI functionality to it's software service (Siri) but where it has a clear advantage is in its silicon.
Owning their own silicon Apple can react to changing demands quicker than rivals.
I've got an M4 Mac Mini and I can run large LLMs on it at a respectable speed. I imagine the M5 will be far better. Running LLMs locally will be the new differentiator between processors going forward.
Owning their own silicon Apple can react to changing demands quicker than rivals.
I've got an M4 Mac Mini and I can run large LLMs on it at a respectable speed. I imagine the M5 will be far better. Running LLMs locally will be the new differentiator between processors going forward.
I'm really not sure what anything you said has to do with my post.
If you think Apple's track record in Mac gaming is anything other abysmal then I don't know what else to tell you.
Buying a games studio and running it badly aren't the same thing.
The partners who have walked away from Mac development? Like Paradox? Those partners?
Who said anything about the Vision Pro? The OP was lamenting the loss of games developers for the Mac.
If you think Apple's track record in Mac gaming is anything other abysmal then I don't know what else to tell you.
Probably not a yield thing. Reportedly the M5. is on N3P . That is the next iteration after N3E and both years behind N3B. Historically, TSMC's iterative refinement processes have started out with relatively good yields because the vast majority of "lessons learned" from the previous iteration are folded into this refinement from the beginning. There is not a ton of "rediscovering the wheel" going on here. N3P is design compatible with N3E. It isn't a radical departure.
All the hand waving several years ago about TSMC N3 yields being 'bad' is largely just disconnected from reality at this point. Even N3B isn't in that categorization anymore.
Additionally, it is likely that there are other customers for N3P wafer starts. The Pro and especially Max are substantially larger dies so they don't produce as many working dies from a single wafer. Very good chance AMD will ship N2 product before Apple does next year. Next fall Apple won't have 'infinite' access to the next gen wafers either. Fall after that if AI bubble keeps going... even less so.
The M5 has more "hand me down" options than the Pro and Max do. Eventually the iPad Air will get the M5. Throwing away chips the size of the Max every 12 months does not make much economic sense at all. If they toss the Pro from the MBP and Mini every 12 months that has similar problems over the long term. Pointing at the MBP moving out on about a 12 month refresh cycle missing the point. If that refresh rate causess hiccups to Mini , Studio , and Mac Pro , then Apple is just sweeping the issue under the rug. That is only going to last so long.
The Vision Pro is a low volume product. This is more so to give Apple flexibility in juggling excess M5 inventory if VPv2 or iPad Pro don't hit their initial 3 month targets. If MBP doesn't also under sell they can soak up excess M5 allocation with that product. Vice versa also if iPad Pro did slightly better than MBP 14". The Vision Pro is just basically along for the ride. ( and M2 is likely wrapping up. The R1 never had volume so it creeps along probably in hope of getting to better return on investment. )
I wouldn't count on that. I suspect the tariff fog lead to Apple getting the production of the Mini 'wrong' in 2024 and they have been bleeding off inventory they didn't want . Maybe the flip-flop and random walk on tariffs continues into 2025 , but possibly not.
I assumed it would have Thunderbolt 5 ports (and Wi-Fi 7) but I guess that eats away at the profits.
Likewise on AV1 encode... I think it's going to be an important feature, and the M6 may be a larger jump next year.
Likewise on AV1 encode... I think it's going to be an important feature, and the M6 may be a larger jump next year.
Don't have to worry about running something badly if don't own it. Apple is not a do everything for everybody company.
Whenever Apple has bought something whose user base was 60+% Windows (or Unix) based and trying to dogmatically 'force' a macOS port on that userbase base then it has usually failed in the long term. (e.g, Shake).
Apple spun out Claris long ago when that software wanted to go down that track. Seen the Safari version on Windows lately? Nope.
If Apple bought it , then to be success it has be primarily Apple product focused. Siloed games isn't going to help Apple much here. [ there is some off-platform stuff they do like Apple Music, Apple TV app. but that is for web based services; not Apple hardware based hardware. ]
Apple has picked up about as many folks as they are lost with the GPU improvements they are rolled out over M3-M5. As you noted the iPhone is volume and is growing. The Metal API on the iPhone is the meta API on Macs also.
Choosing some specific vendors and saying 'Only of Fred and Barney " are there then there is a market is not what Apple does in general. There are some huge anchors like Office and Adobe . But none of these players typically thrown out in gaming are in that category.
I missed that part sorry. It was the previous post I made leaking through on something else.
Your premise started off with "if you build it they will come is lousy". It does work. Over time the iPhone has build up an ecosystem. It primarily just doesn't work 'fast'. That it doesn't do. Apple blowing up OpenCL and direct support for Vulkan only will make it slower. However, Apple buying up studios won't make it go faster. That isn't how Apple is structured nor is it what they are good at. ( "Beats" is far more the exception rather than norm as a result of Apple buying something. )
The volume track that Apple is on is to maximize synergy with the iOS ecosystem. By your own admission that has volume.
What Apple is not going to do with "monkey see, monkey do " Windows copying. Apple isn't bowing to Nvidia on GPUs. They aren't interested in endless sequence of graphics drivers updates to work around quirks with an endless set of gaming engines assumptions. ( biggest siee effect is solidifying duopoly of Nvidia and AMD. ) Games that are that brittle to drivers are an 'ocean away" from being portable. How games are constructed is also a contributing factor. It isn't unilaterally what Apple does (or doesn't do).
Apple has 9-15% of the market. As Jobs said last century the "PC wars are over". Apple is after the profitable subset of the market. Not market share just for market share bragging rights (or to soak in t he most developers possible.)
It’s so exciting though I love this timeline. I much rather have this than disappointing yearly upgrades.