Popular Mac Developer Slams Apple for 'Sad State of Macintosh Hardware'

[Bacillus]

...um maybe you need to rephrase your second to the last sentence or maybe use non-gender specific pronoun?

I assume his wife is called “wife” as nothing else would account for this referral (except stressing her female gender - which might be necessary in some cases...)

 

[Candykane75]

I really can't blame Apple. They've created such cachet around their brand and such a strong ecosystem that they're able to sell tons of whatever gimmicky mediocre mobile computing product they put on the market. What incentive could they possibly have to "do it right"? Whatever increase in sales they'd get from power users would be more than offset by the hit in brand associations and the losses from people who prioritize thin and shiny over powerful.

I don't have to like it, but as a corporate strategy, it absolutely makes sense.

well I can blame apple because is they would have innovated a little more these few years ... A Imac pro is nothing special they could have made it like a few years ago . they could have bin bothered to update the mac pro mac mini and macbook air , the are still offering them its a joke ...
if they had updated there computer line ups and innovated more ... they could have made a better surface studio i'm sure of that... now microsoft has the best all in one we can buy .. that's real innovation that how you lose costumers.
keyboard gate ...... an other thing.... non serviceable computers....
i think apple could have done a lot better but was doing all kinds of stupid stuff and neglected there core busnis
computers and phones. in sted of spending spend billions on nothing ... apple car apple tv apple who knows..

 

[chucker23n1]

well I can blame apple because is they would have innovated a little more these few years ... A Imac pro is nothing special they could have made it like a few years ago .

They could have, but Xeon-W CPUs are new, and Intel generally previously wasn't as interested in that niche of many-core desktop/workstation CPUs.

now microsoft has the best all in one we can buy

Surface Studio is interesting, but the iMac at less than a third the price tag ($1,099 vs. $3499) and goes all the way up to much beefier CPUs (18 cores vs. 4). It offers a much broader range. Surface Studio also lacks Thunderbolt, severely limiting its expansion options. (Its screen is also worse, it doesn't offer 10GigE, and so forth.) I'm honestly not sure why they introduced a Kaby Lake-based computer in October 2018?

It's an interesting AIO product, but given the massive lack of range in configuration options compared to the iMac, I really don't see how one can argue that it's invariably "the best".

And, alas, both iMac and Surface Studio have the questionable design decision of having all ports on the back.

 

[Count Blah]

I really can't blame Apple. They've created such cachet around their brand and such a strong ecosystem that they're able to sell tons of whatever gimmicky mediocre mobile computing product they put on the market. What incentive could they possibly have to "do it right"? Whatever increase in sales they'd get from power users would be more than offset by the hit in brand associations and the losses from people who prioritize thin and shiny over powerful.

I don't have to like it, but as a corporate strategy, it absolutely makes sense.

Except one totally depends on the other for content. And it’s sadly not the one Tim Cook thinks it is.

 

[Easttime]

Except one totally depends on the other for content.

There it is, summed up nicely in one succinct, beautifully parsimonious sentence. The great majority of content viewed on mobile devices probably is produced on desktops and docked laptops, which also produce the apps that the content is viewed on.

 

[flottenheimer]

...I also agree that the form factor of the iMacs lends to cooling issues which limit their trouble free operation to about 4-5 years. My old 2006 MBP still runs - obviously dated, but it runs and was a great hardware purchase 12 years ago. If Apple really put out a new cheese grater style Mac Pro, I'd gladly make it my next anchoring desktop.

Talking about durability. My 2006, super-compact 12-inch Apple PowerBook G4 (latest 1.5GHz revision) still works (not that I use it much). Battery hasn't popped. The drive still works. Never been repaired. Everything is fine.
Talk about high build quality for such a compact machine.

Looking forward to see Apple's next generation of desktops and laptops.

 

[-hh]

They could have, but Xeon-W CPUs are new, and Intel generally previously wasn't as interested in that niche of many-core desktop/workstation CPUs.

Not interested? Well ... I'm not quite so sure about that.

FYI, I'll freely admit that I've not bothered to keep up on all of the Intel CPU news, so please help me out here.

Let's start off with some real basics: just what is it that defines & differentiates a 'Xeon' versus another Intel CPU designs?

The reason I ask is that I recall reading somewhere that the support of ECC RAM was pretty much a "flip a bit" configuration change, which (if I'm recalling this correctly) means that ECC RAM support isn't really an architectural difference between the likes of an i7 vs a Xeon.

Similarly, locked or unlocked for overclocking ... I again don't see this as a meaningful differentiator for CPU architecture.

And so on ...

As I said, I don't know just how functionally different or similar the Xeon and i-Series CPU products are from Intel.

But I do know that it would be make sense from an Engineering design standpoint for them to maximize design structure modularity and interchangeability, but I don't know if this goes to its logical conclusion that both are made from the same mask, and it depends on the post-fab inspections to determine if a discrete part gets QA inspected into Bin "X" or Bin "i",

Because what would theoretically be "good for business" would be for Intel to reduce their proliferation of mask sets, and be able to make both CPUs from a single manufacturing line and mask. The CPUs would be differentiated by a combination of (a) post-fab inspections, (b) chip configuration options, and (c) what their PC customers want ... to determine if a particular CPU gets finished as an i7/i9/etc or as a Xeon. If Intel does have this degree of manufacturing flexibility, then the "niche" of Xeon-W's is an artificial narrowing.

 

[chucker23n1]

Not interested? Well ... I'm not quite so sure about that.

They seemed to react to competition from Ryzen.

FYI, I'll freely admit that I've not bothered to keep up on all of the Intel CPU news, so please help me out here.

Let's start off with some real basics: just what is it that defines & differentiates a 'Xeon' versus another Intel CPU designs?

My comment was in particular regarding not Xeon, but the new ~dozen-core (up to 18 cores) configurations in Skylake-X and Xeon-W. This is a very different configuration from what was available before.

The reason I ask is that I recall reading somewhere that the support of ECC RAM was pretty much a "flip a bit" configuration change, which (if I'm recalling this correctly) means that ECC RAM support isn't really an architectural difference between the likes of an i7 vs a Xeon.

That's basically true. There are lots of similar situations, such as i7s often "just" being i5s with HyperThreading enabled, or i5s being i3s with additional cores. And sometimes, due to scrap, cores get disabled and then binned as cheaper variants.

Core m3, i3, through i9, and Xeon (since the Core days) are all variants of the same microarchitectures with vastly different target audiences and price ranges. There's typically no need for the same power efficiency features of a 4.5 W CPU in a rack server with enough thermal room for a 120 W CPU. And there's no need for 18 cores in a 12-inch MacBook. ECC is theoretically useful for everyone, but particularly useful in the server area, because servers tend to have the highest uptimes, and higher uptime makes memory errors more likely to occur. (Plus, yes, sure, people are willing to spend more money for ECC, and Intel happily takes it.)

OTOH, Intel does have different concurrent microarchitectures — for example, their Intel Atom CPUs are a different branch of x86. (Core i* / current Xeon is derived from Pentium M, which in turn is a modern variant of Pentium III. Atom, OTOH, is derived from the even earlier original Pentium. Pentium 4 was basically a dead end.)

Confusingly, they currently use the Celeron and Pentium branding both for Core and for Atom products. Their marketing is horrible.

Similarly, locked or unlocked for overclocking ... I again don't see this as a meaningful differentiator for CPU architecture.

Yes. It isn't. It's a way to address different audiences. Most people don't need overclocking. Most of those who do care are OK with paying more.

As I said, I don't know just how functionally different or similar the Xeon and i-Series CPU products are from Intel.

Right. Think of it less as "(most)* Xeons are completely different CPUs" and more as: there's a broad range of CPUs in the same microarchitecture, currently Coffee Lake and siblings (Whiskey Lake, Amber Lake), and depending on your thermal and financial budget, you may want one more around $100 or $2,000, and one more around 4.5 W or 120 W. And if you need to go even below 4.5 W, there's a completely different range known as Atom, and an even smaller one known as Quark. If you need to go beyond Xeon, you can pick Itanium, but that's basically dead.

*) let's not even get into Xeon Phi, which basically isn't a CPU at all.

But I do know that it would be make sense from an Engineering design standpoint for them to maximize design structure modularity and interchangeability,

Sure, but as advisable as that sounds, that's hard enough to do in software, and harder yet in hardware.

but I don't know if this goes to its logical conclusion that both are made from the same mask, and it depends on the post-fab inspections to determine if a discrete part gets QA inspected into Bin "X" or Bin "i",

Because what would theoretically be "good for business" would be for Intel to reduce their proliferation of mask sets, and be able to make both CPUs from a single manufacturing line and mask. The CPUs would be differentiated by a combination of (a) post-fab inspections, (b) chip configuration options, and (c) what their PC customers want ... to determine if a particular CPU gets finished as an i7/i9/etc or as a Xeon. If Intel does have this degree of manufacturing flexibility, then the "niche" of Xeon-W's is an artificial narrowing.

You can remove cores and sell for cheaper, but you can't just keep adding cores inside the package and sell it at a higher price tag. It takes more engineering work than that.

 

[-hh]

They seemed to react to competition from Ryzen.

{big snip}
You can remove cores and sell for cheaper, but you can't just keep adding cores inside the package and sell it at a higher price tag. It takes more engineering work than that.

Thanks; you pretty much validated what I suspected to be the case, which is that there really isn't any huge Technology obstruction that prevents Apple from getting whatever permutation flavor of 'Xeon', so long as its specs are what I'll call an *interpolation* of existing designs rather than an extrapolation (extension): its more just a case of the lawyers sitting down to negotiate & work out the contracts for the custom design NRE, fab quantities, delivery schedule, unit prices, etc. Granted, Intel might not be interested in a limited run of only 50,000 custom CPUs for Apple, but turn up the offer's profit margin enough to sweeten the deal and they'll come around.

I'd have to go reference a wafer optimization project I worked on a couple of years ago .. but this also doesn't have to take all that long, even when including schedule to cut a new mask, etc. Assuming priority to get into stations, six months should be more than ample.

 

[chucker23n1]

Thanks; you pretty much validated what I suspected to be the case, which is that there really isn't any huge Technology obstruction that prevents Apple from getting whatever permutation flavor of 'Xeon',

Right, but the point is that before Xeon-W/Skylake-X, there wasn't anything like an 18-core desktop CPU from Intel. Not with Xeon, i7, i9 or anything branding.

You could've put a Broadwell Xeon E5 in there, but Intel didn't intend for that use case.

Only with Xeon-W did various vendors like Apple, Dell and HP decide to make "workstations" with that many cores.

its more just a case of the lawyers sitting down to negotiate & work out the contracts for the custom design NRE, fab quantities, delivery schedule, unit prices, etc. Granted, Intel might not be interested in a limited run of only 50,000 custom CPUs for Apple, but turn up the offer's profit margin enough to sweeten the deal and they'll come around.

Right, that's another point — Intel may not have been able to deliver the CPU at the desired volume. (Not that the iMac Pro is a particularly high-volume product.)

 

[LordVic]

Right, but the point is that before Xeon-W/Skylake-X, there wasn't anything like an 18-core desktop CPU from Intel. Not with Xeon, i7, i9 or anything branding.

You could've put a Broadwell Xeon E5 in there, but Intel didn't intend for that use case.

Only with Xeon-W did various vendors like Apple, Dell and HP decide to make "workstations" with that many cores.



Right, that's another point — Intel may not have been able to deliver the CPU at the desired volume. (Not that the iMac Pro is a particularly high-volume product.)

Not entirely true. As of Haswell at least, there were Xeon e5 and e7 based workstations available from some manufacturers. Sure, these chhips weren't "workstation" designed, but ultimately, as you two have covered, the Xeon architecture to the i* is fundamentally identical. And there were up to 18 core variants of these available.

I had purchased a few HP/Dell Workstations for developers during the time that featured these chips.
https://ark.intel.com/products/codename/42174/Haswell

 

[-hh]

its more just a case of the lawyers sitting down to negotiate & work out the contracts for the custom design NRE, fab quantities, delivery schedule, unit prices, etc. Granted, Intel might not be interested in a limited run of only 50,000 custom CPUs for Apple, but turn up the offer's profit margin enough to sweeten the deal and they'll come around.

Right, that's another point — Intel may not have been able to deliver the CPU at the desired volume. (Not that the iMac Pro is a particularly high-volume product.)

I think you might have misunderstood where I was going with that statement. My point was that if Apple was willing to pay enough, Intel could have fast-tracked both the work to design a derivative CPU ... as well as to have put it on a Fast Trick though their Foundry.

Personally, I've had wafer development contracts in a foundry setting (for process yield - - not in Silicone) and while you can eat up a lot of time mucking with deep process stuff, once that stage is done - - which it would be here for a derivative CPU design - - then its simply just getting priority on the fab floor to push it through.

Foundries like to talk about how long it takes to make anything, but most of that time is really spent waiting, not fundamental fabrication time. As such, with enough Management attention, you can get stuff go through much more quickly. For example, the foundry I worked with had a management policy for "Rocket Lots" and they were prioritized so as to be delivered in half the normal schedule ... and what we found working with them as they had a goof, there's more to be gained than just a Rocket: we saw what I'll call an "Oops, asking management for a 'Rocket' won't save us - we're in BIG trouble!", which shaved another 25% off of the already cut in half 'Rocket'. Putting this together, their WIP was in fact no better than roughly 1/3rd actual machine time and the other 2/3rds of the time was just spent waiting to get onto a machine.

TL;DR summary: its been 18 months since April 2017, so for quantities under 100K units delivered by 4Q2018, there really isn't any "CPU supply limitation" excuse. It meant that Apple wasn't willing to open their wallet to buy schedule.

 

[chucker23n1]

I think you might have misunderstood where I was going with that statement. My point was that if Apple was willing to pay enough, Intel could have fast-tracked both the work to design a derivative CPU ... as well as to have put it on a Fast Trick though their Foundry.

Oh sure. But it's hard to know what goes on behind the scenes. Was Apple only interested in making the iMac Pro because Intel told them they'd soon make the Xeon-W? Or did Intel plan the Xeon-W in part because Apple asked them to make that kind of CPU for them?

We don't know the lead times (some of the aspects of the iMac Pro suggest it was in development for years), and we don't know who talked to whom.

So it's quite possible that Xeon-W happened in part because Apple was willing to pay enough.

TL;DR summary: its been 18 months since April 2017, so for quantities under 100K units delivered by 4Q2018, there really isn't any "CPU supply limitation" excuse. It meant that Apple wasn't willing to open their wallet to buy schedule.

Sure. For the Mac Pro (not iMac), I don't think the CPU is the issue.

 

[Omega Mac]

I suggested maybe two years ago on the forum, that the Spaceship campus may been a project that was diverting a lot resources across Apple at the expense of day to day.

Now that's it complete I already feel there has been a uptick in output and focus. Another year should clarify. Simply put there are a ton of meetings an decisions spanning over years that no longer have to take place on the subject of a new HQ, surely that frees up everyone to get on with the important tasks.

Bumping my own post to blow my own Trumpet!

The Mac Mini Update and everything since finally validates my historical speculation. I really do think the Apple Campus distracted a lot of Apple Execs from keeping pace with serious day to day stuff and some slack and poor judgement was elevated to the top Much of the criticism of the past have been addressed somehow, maybe not to everyone liking and maybe not all but in many cases it has been addressed.

 

[Naimfan]

Looks like this thread can now be closed, as yesterdays announcements fixed all that was wrong with Mac hardware . . .

/s

 

[corebeliefs]

Looks like this thread can now be closed, as yesterdays announcements fixed all that was wrong with Mac hardware . . .

/s

Lol, right. I'm left wondering if I can buy a dual-core machine in the Air and still spend $1750+ with warranty and dongles. And the butterfly 3 keyboard ... what are anti-sales points called?