Did Apple rush an 'inferior' SOC to market ?

[ProgRocker]

New to Apple and trying to understand what I'm getting for a large sum of cash. I'm hearing that Apple went with N3B instead of waiting for the 'better' N3E chips. What are your thoughts on this ? Is it better to wait and see if they move over to N3E later in '24 ? Or will they continue on with N3B until the M4's are out ?


Apple M3 Deep Dive

 

[elvisimprsntr]

Probably. Rumors are N3 yields are not as high as previous lithography, so it’s likely Apple and TSMC had to lower specs and bin the chips to improve wafer yields. I’m good with my ‘20 M1

 

[Kazgarth]

Do you think Apple suddenly decided to use N3B and produce M3?

They have contracts several years in advance. Priority on new node.

Which has it's ups and downs, cutting edge performance, but with terrible yield and cost.

 

[MBAir2010]

RUSH!

dont know if  rushed on the Market and Socks,
but Geddy Lee will have a 4 part TV show airing next month.
we miss he dry wry sense of humor!

BTW: the video is better and funnier with the volume off!

 

[Apple_Robert]

I would say stop watching YouTube clickbait drivel. There is no need to wait on a purchase.

 

[bcortens]

I would say stop watching YouTube clickbait drivel. There is no need to wait on a purchase.

I would say if you have M2 you should wait for reviews for your application needs before buying. If you're coming from M1 however the M3 looks like a fantastic upgrade.

 

[altaic]

You’ve been a forum member for five+ years. You’re not new to Apple.

 

[APCX]

These takes are getting desperate.

 

[bcortens]

Having watched that video I think he's being a bit disengeous about some of the reasoning for things.

I have some doubts about the 150 GB/s bus on the Pro being done because the N3B process yielded poorly. It seems like it would take too long from design lock-in to tape=out for that to have been changed based on yields. I think Apple made the decision several years ago to move the Pro down in the performance hierarchy and put more distance between the Max and the Pro and less distance between the regular M3 and and the M3 Pro. The only way the 150GB/s bus can be a yield based decision is if the Pro actually has a 200 GB/s bus but they just disabled part of it.

I dislike this move and wish Apple had not cut the memory bus on any of their SoC generations as I think it is a bit hostile because they should be continuing to raise the floor in memory performance to continually increase the capability of the base models.

At 5:15ish he complains of the configurator stealthily upgrading memory capacity as you toggle between SoC choices but this was also true when switching between the Pro and Max SoCs for the M1. Sure it is new that it is also the case within an SoC but it isn't as though it is that new or stealthy or unwarranted. This point of his seems like complaining for the sake of complaining.

He repeatedly asserts that N3B is yielding super poorly and costing more than N3E will, I mean yeah N3E will cost less but it also not likely available at Apple volumes till next fall (given the way N3B rolled out). The only article I could find in my quick searching suggests that cost per transistor is down by 15% moving to N3B (Transistor cost)

Given how close N3B and N3E are in terms of density, while N3E is slightly less dense it is not by much, they will likely build the A18 and M4 on N3E and that will just be another year on year improvement.

Edit: In conclusion, no Apple did not rush an inferior SoC to market.

 

[JustAnExpat]

RUSH!

dont know if  rushed on the Market and Socks,
but Geddy Lee will have a 4 part TV show airing next month.
we miss he dry wry sense of humor!

BTW: the video is better and funnier with the volume off!

Really?? Source on that? I'm interested.

 

[GMShadow]

You’ve been a forum member for five+ years. You’re not new to Apple.

They also asked about this in July...perhaps doing 'research' for a video.

 

[MBAir2010]

Really?? Source on that? I'm interested.

progrock.....Geddy Lee TV

can't waits!

 

[neuropsychguy]

Future tech will almost certainly be better than current tech. Using current tech is using inferior tech. Therefore, companies should only use future tech and not current tech.

 

[antiprotest]

More than ever, I do believe that Apple cuts corners in design, materials, quality assurance (both hardware and software), but you are still likely to get a satisfactory product.

As their product margins increase though, customers are getting less value for the money paid.

Around my city, I see $7-12 pastries. They are worth $1.50-$4 imo, but people still buy them, because either they don't have real choice, or the products are still better than the alternatives, especially when other variables are considered (such as location, convenience, etc.).

Businesses know this and exploit the customers. Of course, they can blame it on inflation, on climate change and environmental considerations, claim superior quality, etc., but they charge that much mostly because they can.

Apple does the same thing. You will have to decide for yourself.

 

[jz0309]

New to Apple and trying to understand what I'm getting for a large sum of cash. I'm hearing that Apple went with N3B instead of waiting for the 'better' N3E chips. What are your thoughts on this ? Is it better to wait and see if they move over to N3E later in '24 ? Or will they continue on with N3B until the M4's are out ?


Apple M3 Deep Dive

What difference does it make for YOUR USE case?????

 

[LambdaTheImpossible]

Ignoring the video entirely, yields are always lower than ideal when you reach first production. You hope to be able to make gains over time as production issues are ironed out. But you have to start paying back all that money at some point so you need to bring the product to market at a price point where you can actually do that. And that means not throwing 80% of your stock in the trash.

What the SKU breakdown of the M3 stinks of is that the yields were pretty poor and they had to segment the market enough to use up the crap parts. It wouldn't surprise me if the M3 is actually binned M3 Pro parts at this point. The fact it'll be in the same form factor and package in the MBP14 (for production reasons) makes me suspicious.

 

[Pet3rK]

Ignoring the video entirely, yields are always lower than ideal when you reach first production. You hope to be able to make gains over time as production issues are ironed out. But you have to start paying back all that money at some point so you need to bring the product to market at a price point where you can actually do that. And that means not throwing 80% of your stock in the trash.

What the SKU breakdown of the M3 stinks of is that the yields were pretty poor and they had to segment the market enough to use up the crap parts. It wouldn't surprise me if the M3 is actually binned M3 Pro parts at this point. The fact it'll be in the same form factor and package in the MBP14 (for production reasons) makes me suspicious.

The base chip on MBP14 is the replacement for tbMBP. No one should be surprised that they used the MBP14 chassis.

 

[Confused-User]

I can feel my IQ dropping reading this nonsense. This has been discussed at great length in other threads here. There is zero evidence that there are any major problems with N3B. Posters in this thread supporting that view simply have no idea what they're talking about and are too lazy to research.

N3B yields are clearly good enough for Apple to be rolling out M3s. It would be a very easy call to not do that if there weren't enough wafers. Any suggestion that binning means a bad process is just completely ignorant.

The M3 series is mostly where we thought it was going to be after the A17 shipped. Clocks are a little lower than they could be in the laptops, but it will be unsurprising if they're higher in the desktops. They are demonstrating surprisingly high multi-thread scores on the Pro and Max, possibly even shockingly high - we need more data but it's looking extremely good. Single-thread scores are also good.

If you have patience, read this thread, where this is all being covered in exhaustive detail.

 

[LambdaTheImpossible]

The base chip on MBP14 is the replacement for tbMBP. No one should be surprised that they used the MBP14 chassis.

Of course. But the point is they stick an M3 chip in there. I suspect it'll have the same die and carrier as the M3 Pro. Just with more broken bits on it.

Time will tell on that of course.

 

[bcortens]

Of course. But the point is they stick an M3 chip in there. I suspect it'll have the same die and carrier as the M3 Pro. Just with more broken bits on it.

Time will tell on that of course.

I consider this very unlikely. The Pro SoC is quoted at 12 Billion more transistors than the M3, they are not eating an extra 33% per chip cost overhead just to avoid making a dedicated M3 SoC...

Edit: Apple also showed a different logic board layout for the M3 vs the M3 Pro 14" chassis. The M3 only has one fan and a completely different internal layout, again, more evidence that it is not a cut down Pro.

 

[Aries79]

I'm betting that the N3E will be used for the M4. The next big step forward will be for the 2nm M5.

 

[gpat]

They just wanted to push out one more iteration.
The alternative would've been to just skip this process altogether and wait one more year for M3 N3E.

 

[NT1440]

New to Apple and trying to understand what I'm getting for a large sum of cash. I'm hearing that Apple went with N3B instead of waiting for the 'better' N3E chips. What are your thoughts on this ? Is it better to wait and see if they move over to N3E later in '24 ? Or will they continue on with N3B until the M4's are out ?


Apple M3 Deep Dive

Apple didn’t “rush” anything. Their capital expense, being the only major OEM to not pull out of 3nm production, *is the only reason that 3nm production became economically viable in the first place*.

 

[leman]

It wouldn't surprise me if the M3 is actually binned M3 Pro parts at this point. The fact it'll be in the same form factor and package in the MBP14 (for production reasons) makes me suspicious.

You think Apple would be lying to the customers like that? That’s a very odd thing to assume.

 

[leman]

I'm hearing that Apple went with N3B instead of waiting for the 'better' N3E chips.

N3B is the node developed for Apple and in cooperation with Apple. Without Apple there would be no N3B. I don’t think it makes much sense to claim that Apple rushed something by adopting the technology they have commissioned in the first place.