Those pros have a disproportionate mindshare for the majority buying public though. People will want to make the same buying decisions that the pros did, even if their use cases are not the same.
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.
3nm Apple Silicon: What Is It and Why Does It Matter?
- Thread starter MacRumors
- Start date
- Sort by reaction score
I am starting to think how fast a M3 Ultra studio will be, given the M1 Ultra I have is an absolute joy to use.
The big word for my usage is 'Raytracing' - if implemented well it will be upgrade time.
The big word for my usage is 'Raytracing' - if implemented well it will be upgrade time.
When I worked in the semiconductor industry I learned that 4 nm = 1 beard second, the length a normal beard grows in one second. That’s pretty short distance.
(Math to back it up - 0.3 mm/day seems to be average growth of beard divided by 24*3600 approx 4E-9)
(Math to back it up - 0.3 mm/day seems to be average growth of beard divided by 24*3600 approx 4E-9)
"Must be Nvidia" ? What alternative universe are you living in? In this universe, Nvidia blew up their relationship with Apple. It isn't coming. There is no "must" that Apple needs to do here. It could be a 'nice to have' , but a hard requirement. No. There were no Nvidia MacOS support with the MP 2019 ( other than running off to Windows. And that is also out at the 'raw iron' boot level) .
Same issue with "must have" display GPU. Pretty high likelihood the baseline GPU will be the Apple iGPU. Whether there is other optional GPU ( e.g,. compute GPUs) may be up in the air , but are far as a hard requirment an apple iGPU is the modern Mac requirement at this point ( seemless native running of iOS software, assumptions on Unified Memory that Apple has heavily guided developers into adding to their applications fort he last 2.5 years , etc. )
It would probably be useful for the whole Mac line up if Apple upped their game on hypervisor and virtualization framework and allows a guest OS to take full responsibility for a 3rd party GPU ( e.g., Nvidia running solely on a Llnux OS guest through a direct ( or extremely thin) IOMMU static assignment. ). But that shouldn't be Mac Pro specific. Probably a more common use case than a MBA with a external TB PCI-e card enclosure, but not MP exclusive.
However, that is not native driver support inside of macOS proper.
Back in the era when Steve Jobs penned his "notes on Flash" there were a decent number of folks running around saying that the iPhone was going to flop if Apple didn't add flash. lots and lots of developers were using Flash and without it the iPhone would never get traction.
Somewhat similar boat here. The tasks doesn't strictly require it. It is completely possible to do AI/ML without using CUDA. There are some specific pieces of software eyeball deep in deeply proprietary code. But that software isn't really the 'task'.
In the Flash context Apple pushed open Web standards to fill the gap. It took years but HTML 5 / Javascript evolution , WebAssembly , WebGPU , etc there were not many whimpers when Flash finally died off. Apple is somewhat in a Metal versus CUDA dogfight here which really isn't as "open standards". Nvidia's 'exbrace, extend, extinguish" on OpenCL and the large moat to dig around the iPhone has put Apple on a more proprietary path this battle. Similarly Nvidia more than willing to put nudge Metal into loosing so that CUDA can win to dig a bigger moat around their hardware. Have two companies that don't strategically need each other each going in different directions.
Another post from some alternative universe. N3 grouping of fab processes will be followed by N2. This first N3 version, N3B is going to be design rule incompatible with the rest of the N3 family ( N3E .. etc) but the only "upside down" is that N3E eases off the aggressive shrink. (e.g., the SRAM/cache feature size will be the same as N5P. So that part of the chip won't get smaller at all. ) It isn't that the chips are getting much bigger, but they aren't getting a lot smaller either ( unless specifically leave certain features out). If the die size stays the same but the wafer costs are 20-40% more than costs get bigger.
Apple has already bloated the A14->A15 (A16 is still significantly above simple A-series normal using N4) and M1-> M2 , M1 Pro -> M2Pro , M1 Max -> M2 Max by adding more stuff to push performance to a relatively same N5 process.
N3 wafers cost more. So there is some pressure on Apple to use 'less wafer' with slightly smaller dies. There may be enough tension to add "more stuff" to keep the die sizes roughly the same . Since the caches are not shrinking much there isn't going to be gobs and gobs of 'free space' freed up by going to N3. Some ( and will likely get some more fixed functions), but folks trying to spin the "we are going to get the kitchen sink amount of new stuff" will likely be disappointed.
Why would you care? Unless it is for bragging rights, it doesn’t matter. If you need the higher end Macs then you are probably using them for tasks that do a lot of multi-core processing. In that scenario, more cores trumps faster cores.
Probably not. TSMC has openly stated that N3 has more tapeouts at this stage than N5 had. There are several companies working on N3 products. Most of them are targeting N3E which won't go into high volume production until 2H 2023 ( July-September ) timeframe. Other companies are getting samples back from the fab now. And will have access to modest size runs in 2H time frame.
Apple will probably beat them on volume. ( (in terms of hipster, trendy consumer product. ) But in terms of able to ship out N3 processors to targeted customers in late 2023, Apple isn't going to have an exclusive.
I suppose it will be even faster than the M2 Ultra that hasn't been announced yet. Still have to contend with this earlier news.
TSMC Struggling to Make Enough 3-Nanometer Chips for Apple
Apple supplier TSMC is "straining" to produce enough of its newest 3-nanometer chips to meet demand for Apple's upcoming devices,...
www.macrumors.com
i didn’t see that quote in the article you posted. While it is true that process naming does not have a direct and linear relation to actual feature sizes, it’s not true that it has “no relation”. Features and density of a 3nm chip will be smaller and more dense than a 5nm chip, not larger or randomly differently sized.
Looks like they are starting to catch up. Their first EUV-based CPU ("Meteor Lake") is reportedly in production for release later this year. It has a CPU-tile based on Intel 4, and GPU- and I/O-tiles based on TSMC 5nm. TSMC 3nm is still super expensive due to the low yield (but Apple may get a sweetheart deal to keep them happy). TSMC customers other than Apple (e.g. Qualcomm and AMD) are reportedly delaying 3nm adoption.Whatever the terms used it seems that TSMC is the only manufacturer that’s making advancements I reducing their node size, so much so that Intel is throwing in the towel and ordering 3nm from TSMC.
Intel seems to be struggling to go below 7nm (no matter what Intel names it).
Intel Will Rely on TSMC for its Rebound - EE Times
Intel will join Apple to order the world’s first 3-nm chips from TSMC as the Taiwan chip foundry ramps up its newest process this year.www.eetimes.com
Intel Reportedly Delays TSMC 3nm Orders for 15th Gen Arrow Lake CPUs
Intel is slightly adjusting its Arrow Lake rollout as it looks to cut costs and navigate weakened demand.www.tomshardware.com
Last edited:
Say what you will, the wheels would give it better airflow. It does need a nice polishing cloth and a sticker.
My first point, if you are doing scientific computing, you use CUDA. There is no alternative to this in the market. ATI/AMD is fantastic for desktop applications and playing games. that is not what a Mac Pro is for.
For the second part of my post about flipping the chip upside down and saying it was on a -3nm process, that was a joke. Can you say joke? Keep trying, you will get it one of these days.
It has no relation in the sense that there is no feature of the 3 nm chips that has a size of 3 nm, or anywhere close to that. For example, https://en.wikipedia.org/wiki/3_nm_process mentions 48 and 24 nm as actual feature sizes.
And the relative numbers are also misleading. If you look at the corresponding numbers of the 5 nm process, the 3 nm process should probably have been named 4 nm in relation at best.
In addition, the nm numbers of different manufacturers aren’t comparable, because they don’t share the same technological basis.
Yes, a smaller number means “better” in some capacity within the line of process nodes of a single manufacturer, but they may as well be calling them (n+1)th gen. The nm numbers really don’t give any more information.
Last edited:
Typo in the article:
"TSMC is expected to begin full commercial production of 3nm chips in the fourth quarter of 2022".
I guess, they meant 2023.
"TSMC is expected to begin full commercial production of 3nm chips in the fourth quarter of 2022".
I guess, they meant 2023.
But apple is a profit driven company (as all should be) and portables are 4/5 of the mac sold so portables are, sadly, highest priority for apple
Yeah it is "slow news Saturday" so a copy and paste article is quick to do. However, it is May 2023. "expected to begin full commercial production ... 4Q 2022 " is the wrong tense on the verb at this point. It has already happened. N3 is in commercial production.( TSMC reported that folks are paying for wafers or at least dies. It is generating revenue so 'commercial'. Just not the dominate revenue flow right now. ) The only quibbling to be done is about just how 'full' is "full production".
The lead production 'bake' time on these initial N3B dies is long. Reportedly around 4 months. So 'blank' wafers going in now (May ) are only going to pop out the other side packaged and ready to go in the Auigust - September timeframe.
TSMC years ago gave themselves until 2H 2022 to get started. So things are "going to plan", but pragmatically the plan has been shifted around 6 months. It appears a fair number of folks interpreted 2H 2022 to be August-October window for arrival. That didn't happen. It was December which is in 2H 2022 like Apple's 2013 December 20-ish Mac Pro arrival was in Fall 2013. Just a few more days and it would not have been in the 'plan'.
Because N3's bake time is substantively longer, it has taken longer to get to the point where the quality improvement data pipeline is fully flushed out.
Errr, reportedly for Gen 15 (Meteor Lake) nope. Laptops are coming first. And the Gen 14 is a "better than they though it would be" refresh of Gen 13. Laptop SoC are strategically more important to Intel right now than Core i9. AMD has 'slow paced' their work in Laptop SoC up until now. They were usually 'last' out the door on the early Ryzen generations. Partly because AMD was poor ( almost broke when Ryzen first started , but things have gotten gradually better). Intel has managed to fumble enough time and momentum that AMD has enough resources to be more aggressive in the mobile space.
Neigher end should to always promoted over the other. Part of this is timing as to when things are ready and how designs turn out. Apple's fixed in stone September evenut for the iPhone should lead to a time when the new M-series is ready to go but the iPhone SoC is not. Fab process updates don't come on an exactly 12 month schedule. In part, why iPhone A-series would have missed TSMC N3 even if it has been "on time" in early 2H 2022. Because 2H is too late for the A-series. Even more so when the bake time is about 4 months. ( i.e., would need to start in March to have something in June-July time frame. And Feburary to have something in June-July timefame. )
It is doubtful that the Ultra (and above) M-series will update on each iteration. The M2 has a 'hand me down' to an iPad. ( just like on the iphone side it is run down through the lower end iPads, AppleTV , etc. )
Where are 'hand me down' M2 Ultras going to go? It is too big for the chassis in the rest of the line up.
It is also 'too big' to have very low R&D costs , so if kill it too quickly the amortization costs over the even smaller number of packages made gets even higher.
Pretty good chance Apple is on an odd (or even ) path long term for the much larger and expensive packages. ( M1 -> M3 -> M5 or M2 -> M4 -> M6 ) Apple skipped on the Axx-X updates where the iPad Pro processor just took the process shrink iterations. ( ran 18-24 months zone ).
Yearly updates for the Mac Pro ... when has that happened in the last decade?
Yearly updates for the iMac Pro ... didn't happen either.
The Mini sat for over two years. The iMac 24" is STILL sitting with no update.
Apple has some yearly status quo on desktops really?????? not.
The more expenive the desktop system gets then the lower of volumes being bought. Lower volume product isn't likely going to be churned quickly by Apple. Can point at the Windows/PC general market all you want. That isn't what Apple is primarily selling into, nor are they doing anywhere near that kind of volume .
80% of users here won't notice a thing. seriously. why do so many people in these forums care so much? if your job requires it, won't the company pay for it? so many self employed here in macrumors
Is anyone besides Apple buying TSMC's first-gen N3, aka N3B?* Because if they aren't, it seems unlikely anyone but Apple will have TSMC 3 nm** products in 2023:
With N3B, there's expected to be a year delay between volume production and its appearance in products (2022 2H volume production => products expected 2H 2023). Even if we cut that in half for N3E, volume production in 2H 2023 implies we won't see products with N3E until 1H 2024.
*Anton Shilov of Anandtech wrote this back in Jan 2023; not sure if things have changed since then:
"Many analysts believe that the baseline N3 (also known as N3B) will be used by Apple either exclusively or almost exclusively"
[IIUC, the first variant was actually just plain N3, but it had problems and was cancelled, so the first production variant is N3B.]
TSMC's 3nm Journey: Slow Ramp, Huge Investments, Big Future
**I was explict in saying "TSMC 3 nm", because I don't know when the first products with Samsung 3 nm will be released.
Samsung's 3nm GAA node has yield rates of around 70 percent
Like Intel, Samsung has been struggling to find customers for its 3nm GAA process node. This can be partly explained by TSMC cutting prices for 3nm FinFET...
www.techspot.com
Last edited:
I'm stoked to get raytracing on AS SoC. I might ditch my gaming PC and just play whatever Mac games I can find. I should cut down on the gaming anyway. 
Just watch, Apple will release a M2 based Mac Pro and 2-3 months later M3 Mac Book Pros will come out that mop the floor of it.
I don't think anyone should buy the 1st Gen Apple Silicon Mac Pro unless it comes with M3 based hardware. And even then it still might not be as GPU powerful as the 2019 version fully decked with 2 W6800X Duos.
Apple should also release 7000 series AMD MPX cards, but they will not because then the 2019 Mac Pro will again be way faster GPU wise then whatever M variant they have coming.
Apple is damned if they do and damned if they don't. Literally the best thing they could do is release Apple Silicon Mac Pro and also updated Intel Mac Pro. Those who want to jump to Apple Silicon can do so and those who want to stay another few years on Intel + AMD can also. Win Win. (We all know this is a pipe dream though, because when has Apple ever done what was right for high end users?)
I am going to be a sad Mac user again if I have to build a hackintosh again and stick with Ventura until it's dead.
I don't think anyone should buy the 1st Gen Apple Silicon Mac Pro unless it comes with M3 based hardware. And even then it still might not be as GPU powerful as the 2019 version fully decked with 2 W6800X Duos.
Apple should also release 7000 series AMD MPX cards, but they will not because then the 2019 Mac Pro will again be way faster GPU wise then whatever M variant they have coming.
Apple is damned if they do and damned if they don't. Literally the best thing they could do is release Apple Silicon Mac Pro and also updated Intel Mac Pro. Those who want to jump to Apple Silicon can do so and those who want to stay another few years on Intel + AMD can also. Win Win. (We all know this is a pipe dream though, because when has Apple ever done what was right for high end users?)
I am going to be a sad Mac user again if I have to build a hackintosh again and stick with Ventura until it's dead.