Mac Pro 2023 = Another Mac Pro 2013 crisis and failure.

[theorist9]

One thing I briefly noticed but didn't spend any time looking into - when I was looking for Geekbench M2 Ultra numbers, I got the impression there were two clock rates of M2 Ultras in there.

Is it possible that the variant in the Mac Pro is clocked higher or something?

That's a subject of active speculation on another forum on which I post. Many there, including myself, were hoping Apple would take advantage of the high TDP's available in both the Studio and the MP to boost the clocks in their M2's (both CPU and GPU), to improve performance over the mobile devices.

But based on clocks shown on GB, it doesn't look like they did that for the Ultra Studio's CPU (they're all 3.7 GHz*, same as the M2 Max in the 16" MBP) (all other previously-released M2's are slower: 3.5 GHz; see screenshot). Nothing for the MP has been posted yet, so that remains to be seen. Certainly, significantly boosting just the MP would provide performance differentiation vs. the M2 Ultra Studio.

It's possible that the M2 can't be boosted (it may not be designed to handle the higher voltage). If so, we will need to wait for the M3.

Current M2 clocks, from Primate's website:

*Well, most are 3.7 GHz. There are some unusually low ones, but they can be ignored as bad data, because they give usually depressed SC scores (which we know should be ~2800, based on the M2 SC scores for previously-released devices; more specifically, based on the M1 scores, we'd expect the M2 Studio to be ~20-30 points higher than the M2 Max => ~2800).

 

[ADGrant]

And how do you feel about the plan to rewrite Windows Outlook in Edge WebView2? Probably the same person who came up with Windows 8 came with that one - I still don't understand who thought trashing a big bloated mess of a program with a broad ecosystem of third-party add-ins that businesses rely on every day was a good idea. Big legacy Windows Outlook is probably half the reason for choosing Office 365 over G Suite...

I think that is a terrible idea which will succeed about as well as Windows 8 did. For a lot of people though the killer Office 365 app is Excel, not Outlook.

 

[bobcomer]

Since none of us has a Mac Studio with an M2 Ultra, I don't see how we can know if there will be any thermal issues.

Just by the way the M1 Max has done in a studio and it's not a redesign. The sucker didn't get hot -- ever, no matter how I pushed it, and I push machines hard with my testing. The fan rarely even got above idle. Like I said before, it's way over engineered cooling-wise. And the copper heat sink just means it can dump heat faster.

I think we can safely assume that there will be no thermal issues with the Mac Pro because that cooling system was designed for a 14nm 28 core Intel CPU.

Yep, and I say the same about the studio ultra. I could possibly be wrong, of course, but you could be just as wrong since we haven't even seen an M processor in a Pro chassis and we don't know if they changed the cooling system in it either. My opinion is that neither will have a thermal issue.

 

[burgerrecords]

I wasn't really around at the time, but I don't think there was really a change from mainframe to x86 directly.

First you got minicomputers - DEC PDPs, VAXes, etc - and those took some of the mainframe's workloads away. Mainframes kept doing the big stuff.

Then RISC machines (e.g. SPARC) that continued to run *NIX ate the CISC minicomputers.

Then, a good chunk of the RISC guys running *NIX, plus the legacy of the minicomputer era's biggest player, DEC, made a very wrong bet on Itanium.

With most of the RISC guys on Itanium, Sun as the main remaining player, etc, x86, and more particularly, x64 (running either NT or Linux) starts seriously eating the RISC machines' lunch. And as that continues, the x64 machines get bigger, moving into bigger workloads that previously would have been reserved for the biggest boxes from Sun or whoever. (To put it another way - if you were starting eBay in the late 1990s and you needed a giant database server, you bought a Sun Enterprise 1000. Some years later, you might have bought a big x64 box, then a few years after that, you would have hosted your database in a public cloud.)

Meanwhile, the mainframes have kept mainframing in at least some of the core segments, e.g. banking, etc., and have been replaced directly in some other cases by virtualized/cloud x86 boxes.

But really, the world has probably been moving away from mainframes slowly starting at some point in the 1970s. And yet, it's worth noting, the IBM mainframes have outlasted all the minicomputer/RISC stuff. When's the last time DEC released a new VAX CPU? Or the last time Sun released a new SPARC chip? IBM released a new zArchitecture one a year or two ago...

I don't know why you keep saying "SoC" - the fact that the Apple silicon chips are SoCs is... not that significant. The i5-1038NG7 in the 2020 MacBook Pros is not that far from a SoC. It runs a different ISA, it has a different memory architecture, etc. And it has much bigger transistors. The AMD CPU in my Steam Deck is an x64 SoC, I believe - you could have put a SoC like that in a MacBook and it would be a SoC, yet it wouldn't be Apple Silicon or ARM or the underlying philosophy of Apple Silicon.

What's more important is that Apple Silicon indicates that smartphone technology is creeping its way up and replacing traditional PC technology. Just like, say, an x64 server running Solaris or OpenVMS is fundamentally just another x64 IBM PC compatible, an Apple Silicon Mac is fundamentally a smartphone with a big display and a physical keyboard running a Mac's GUI and adding some drivers for some types of peripherals not commonly used with smartphones. And the fact that such a thing... turns out to be much better at being a Mac and doing 98% of Mac workflows than a Mac built out of traditional PC technology says a lot. 15 years ago, a leading smartphone chip was a joke compared to Intel's industry-leading traditional PC technology; today, it's... the other way around.

Of course it wasn’t a single leap - that’s another knock on the analogy

I also agree x86 and soc + ram isn’t so substantially different.

i understand the direction things are headed (and it’s not going to end with Apple silicon as the only ARM used for personal computing of course) - but again the Apple fans were really hoping and expecting that macs up and down the line would be best at everything by the time the pro came out

 

[Gudi]

Like I said before, it's way over engineered cooling-wise. And the copper heat sink just means it can dump heat faster. […] My opinion is that neither will have a thermal issue.

Funny, how some people got so used to thermal issues that a computer without them seems to be "over-engineered".

 

[bobcomer]

Funny, how some people got so used to thermal issues that a computer without them seems to be "over-engineered".

I agree I'm used to it on laptops, but I don't have any desktops with thermal problems. The do speed up the fans when needed, but no throttling. The studio doesn't even need to speed up the fans -- that's why I said what I said.

 

[k27]

There’s no thermal limitations on Mac Studio (otherwise, please cite the source). In fact, cooling system is over-designed and Studios remain cool and quiet even under heavy load.

The cooling system is not over-desigened but apparently not undersized once in a while. Many Macs (especially under Johny Ives) have a too weak cooling system. It should be normal for a Mac not to throttle under load.

As far as I know, all MacBooks with Apple Silicon throttle when you put 100% load on CPU and GPU at the same time. This might be acceptable as a compromise for a laptop, but not for a desktop computer (Form follows function). I hope Apple doesn't come up with the idea (again) of degrading the cooling system to make a computer cheaper, smaller (or artificially slow it down).

 

[theorist9]

Workstations don't exist anymore. Threadripper hasn't had a new release in years and the TR 7000 series keeps slipping. Intel only put out Sapphire rapids because they are desperate to find any niche they beat AMD in, and they can beat 4 year old TR3000. Again, there are not enough people who want/need more than the MxUltra/x950X/xx900k but don't use server/datacenter hardware to make entering that market worthwhile. It sucks to be one of the people still stuck in the middle, who maybe could have used it. But your use case is extremely rare, and the cost of developing a new chip is extremely high. As a HEDT platform, the MacPro is DOA. But so is the whole HEDT market.

The MacPro is at the pointest end of Apple's most Niche product line. Stats I can find lump " All-in-one, blade, mobile, rack, and traditional desktop workstation" together with no further detail, so no doubt include a lot of consumer and server hardware as well, but they state that in 2022 only 7.7mm "Workstations" shipped globally. If even 25% of that was actual HEDT towers, that's under 2mm global units. Assuming ~8% Mac market share, that's ~150k potential MacPro customers. I feel this is a very generous estimate for total MacPro customers.

I can tell you're very passionate about your opinons, but remember these are your opinions, they're not facts. Here are mine:

1) You wrote "Workstations don't exist anymore". Yet HP, Dell, Lenovo, Boxx, Puget Systems, Digital Storm, and others all sell them. Small marketshare doesn't mean gone. You could give any one of these your credit card, and have a serious workstation at your door.

2) Yes, AMD hasn't released a workstation chip in years, but back in March the ASUS GM confirmed that AMD will be releasing a new generation of HEDT chips this year (https://wccftech.com/asus-rep-revea...er-7000-storm-peak-cpus-launching-in-2h-2023/), and these chips have been spotted in the wild (https://www.tomshardware.com/news/r...0-storm-peak-cpu-surfaces-with-64-zen-4-cores). Why would AMD do that (and why would ASUS's GM's care) if "workstations don't exist anymore"?

3) The MP doesn't have to be a large portion of Apple's sales for it to be justified. Yes, there are limits to the development dollars they can put into it. But a purely financial analysis misses the point. The 2019 MP was popular among high-end professional creatives, and keeping them within the Apple ecosystem, particularly given their influence, has marketing value to Apple. Plus it was a "halo" product, which also has marketing value. And if I were an Apple engineer, even if I didn't work on the MP myself, I expect I'd like the fact that Apple produced it. So it has internal morale value as well. The issue with the 2023 MP isn't that HEDT is dead; it's that Apple's execution was problematic. Apple knew this, which is why they glossed over it at WWDC (compare that with the lavish attention given to the announcement of the 2019 MP).

The value of the MP to Apple is similar to that of Formula I racing to Honda, or of the Bugatti division to the VW Group. Any modest losses from the MP could be considered part of Apple's marketing budget. Which is why its 2023 execution was disappointing.

Again, small doesn't mean dead. If it did, you might as well declare the exotic car market dead as well (it ain't). Incidentally, using your numbers (150k MP's/yr @$9k each), MP sales represent 0.3% of Apple's $400B annual revenue. That percentage is three times the ~0.1% that Bugatti sales represent out of the VW group's $250B annual revenue. And I'm sure Bugatti's R&D costs are substantial. Does that mean VW should dump Bugatti?

 

[k27]

As Siracusa said on ATP, they've essentially set up the Mac Pro to fail with this "update". In a few years the already shrinking pro customer base will shrink even further and Apple will then quietly kill it.

It's an absolute joke to call this product "Pro".

John Gruber:

""Pro", in Apple product marketing parlance, does mean professional sometimes (e.g. Mac Pro and iMac Pro). But sometimes it just means premium."

Going Pro

My prediction for new iPhone names next week: iPhone 11, 11 Pro, 11 Pro Max.

daringfireball.net

​

 

[Stevenyo]

I can tell you're very passionate about your opinons, but remember these are your opinions, they're not facts. Here are mine:

1) You wrote "Workstations don't exist anymore". Yet HP, Dell, Lenovo, Boxx, Puget Systems, Digital Storm, and others all sell them. Small marketshare doesn't mean gone. You could give any one of these your credit card, and have a serious workstation at your door.

2) Yes, AMD hasn't released a workstation chip in years, but back in March the ASUS GM confirmed that AMD will be releasing a new generation of HEDT chips this year (https://wccftech.com/asus-rep-revea...er-7000-storm-peak-cpus-launching-in-2h-2023/), and these chips have been spotted in the wild (https://www.tomshardware.com/news/r...0-storm-peak-cpu-surfaces-with-64-zen-4-cores). Why would AMD do that (and why would ASUS's GM's care) if "workstations don't exist anymore"?

3) The MP doesn't have to be a large portion of Apple's sales for it to be justified. Yes, there are limits to the development dollars they can put into it. But a purely financial analysis misses the point. The 2019 MP was popular among high-end professional creatives, and keeping them within the Apple ecosystem, particularly given their influence, has marketing value to Apple. Plus it was a "halo" product, which also has marketing value. And if I were an Apple engineer, even if I didn't work on the MP myself, I expect I'd like the fact that Apple produced it. So it has internal morale value as well. The issue with the 2023 MP isn't that HEDT is dead; it's that Apple's execution was problematic. Apple knew this, which is why they glossed over it at WWDC (compare that with the lavish attention given to the announcement of the 2019 MP).

The value of the MP to Apple is similar to that of Formula I racing to Honda, or of the Bugatti division to the VW Group. Any modest losses from the MP could be considered part of Apple's marketing budget. Which is why its 2023 execution was disappointing.

Again, small doesn't mean dead. If it did, you might as well declare the exotic car market dead as well (it ain't). Incidentally, using your numbers (150k MP's/yr @$9k each), MP sales represent 0.3% of Apple's $400B annual revenue. That percentage is three times the ~0.1% that Bugatti sales represent out of the VW group's $250B annual revenue. And I'm sure Bugatti's R&D costs are substantial. Does that mean VW should dump Bugatti?

I think it was clear “dead” and “doesn’t exist” refer to the market being too small to develop specific hardware for. the cost of desigNing, taping out and manufacturing a custom set of chips or SoCs for a couple hundred thousand units sold is just not possible give current RnD and manufacturing costs. Building a few bespoke Bugattis a year is nothing like designing and contracting to manufacture a bleeding edge sub 5nm cpu and gpu. Almost nothing in the world is as expensive or complex to make. A Bugatti can and is essentially built buy combining skilled craftsmen, a bit of engineering and a few custom parts, crankcase castings, forged internals, etc. With a budget in the 7 figure range you could start almost from scratch and get a totally custom supercar designed and built. To create all the parts needed to make a more trad tower MacPro with GPU and DIMM support possible would be in the 9 figure range, or about the totality of the revenue apple can expect from the product. While workstations aren’t truly dead, they are dying, even an amazing game changing workstation at an unbeatable price would not sell many units, the number of people who want a Mac tower at any price is vanishingly small compared even to MBP buyers, let alone ipad or iPhone buyers.

 

[iPadified]

3) The MP doesn't have to be a large portion of Apple's sales for it to be justified. Yes, there are limits to the development dollars they can put into it. But a purely financial analysis misses the point. The 2019 MP was popular among high-end professional creatives, and keeping them within the Apple ecosystem, particularly given their influence, has marketing value to Apple. Plus it was a "halo" product, which also has marketing value. And if I were an Apple engineer, even if I didn't work on the MP myself, I expect I'd like the fact that Apple produced it. So it has internal morale value as well. The issue with the 2023 MP isn't that HEDT is dead; it's that Apple's execution was problematic. Apple knew this, which is why they glossed over it at WWDC (compare that with the lavish attention given to the announcement of the 2019 MP).

The value of the MP to Apple is similar to that of Formula I racing to Honda, or of the Bugatti division to the VW Group. Any modest losses from the MP could be considered part of Apple's marketing budget. Which is why its 2023 execution was disappointing.

Mac Pro will likely not work as a halo product for iPhones, iPads, wearables and likely not Mac Book Air customers. Furthermore, the Mac Pro nerd is not the pinnacle of human existence in the Apple universe. The beautiful coffee shop people is and hence the Mac Pro will not work well as a Halo product.

The car analogy is problematic as consumer cars benefit of having high end cars as testbeds and development platform*. Apple uses the iPhone (consumer car) as development platform for new tech and then apply these, with modifications, to the high end Macs (Supercars).

*Interestingly, it seems that for electric cars, the rapid development is in the consumer space with supercars trailing.

 

[Gudi]

The studio doesn't even need to speed up the fans -- that's why I said what I said.

Which is exactly what you want from a computer in a recording studio.
If you ever hear the fans, it's not the $1,999+ Mac Studio.

 

[ADGrant]

Just by the way the M1 Max has done in a studio and it's not a redesign. The sucker didn't get hot -- ever, no matter how I pushed it, and I push machines hard with my testing. The fan rarely even got above idle. Like I said before, it's way over engineered cooling-wise. And the copper heat sink just means it can dump heat faster.

The M2 Ultra will run hotter than the M1 Ultra. Perhaps not enough to make much of a difference to the cooling system but we will find out soon.

 

[AlphaCentauri]

The M2 Ultra will run hotter than the M1 Ultra. Perhaps not enough to make much of a difference to the cooling system but we will find out soon.

And yet, instead of 25dB noise level at idle for Mac Studio M1 Ultra, Apple is now quoting 15dB for M2 Ultra. So they definitely thought it out and changed something in cooling system or changed fans profile.

 

[bobcomer]

Which is exactly what you want from a computer in a recording studio.
If you ever hear the fans, it's not the $1,999+ Mac Studio.

The only problem with that theory is the whine from some of the Mac Studios. And anyway, the fan is not silent at idle, it can be heard by a Mic if it's close, so I'm not sure I'd use one in a recording studio -- it would have to be outside anyway.

 

[bobcomer]

The M2 Ultra will run hotter than the M1 Ultra.

I haven't seen any proof of that.

Perhaps not enough to make much of a difference to the cooling system but we will find out soon.

Yes we will!

 

[bobcomer]

And yet, instead of 25dB noise level at idle for Mac Studio M1 Ultra, Apple is now quoting 15dB for M2 Ultra. So they definitely thought it out and changed something in cooling system or changed fans profile.

I suspect just the fan profile, as idle seemed too high to me. It'll be interesting to see how it does. Lowering the fan speed on an M1 Max to minimum 1100 RPM, definitely made it quieter.

 

[Gudi]

The only problem with that theory is the whine from some of the Mac Studios. And anyway, the fan is not silent at idle, it can be heard by a Mic if it's close, so I'm not sure I'd use one in a recording studio -- it would have to be outside anyway.

For years I lived with PCs which ran their fans at full speed right from the moment you pushed the power button. You had to buy a special noise-damped tower to make it somewhat bearable, but never silent. There can be no doubt that Apple always tried to make their Macs as quiet as possible and the Mac Studio is no exception. Yes, it is a little bit louder than other Macs, but it's also by far the fastest.

 

[bobcomer]

For years I lived with PCs which ran their fans at full speed right from the moment you pushed the power button.

I haven't seen a PC like that in a decade or two. Most have gone with bigger slower quieter fans.

My i9 desktop is quieter at idle than the Studio was. (Lenovo SFF) Of course it got a lot louder than the Studio when I push it hard! I never hear the fans at all on my Lenovo X13 with a Ryzen 7 processor.

 

[playtech1]

The only reason the 2023 Mac Pro exists is because it would be an embarrassment for Apple if it discontinued its top machine. That would have sent the message that Apple Silicon doesn't cut it at the very high-end.

Thing is, it's true: Apple Silicon was designed for mobile first, small desktop form factor second and high performance desktop a distant third. All very logical given the PC market, but it has left no additional performance on the table for a large desktop design like the Mac Pro, which can easily handle cooling massive amounts of watts.

 

[Gudi]

The only reason the 2023 Mac Pro exists is because it would be an embarrassment for Apple if it discontinued its top machine. That would have sent the message that Apple Silicon doesn't cut it at the very high-end.

M2 Ultra is literally twice as fast as the fastest Intel Xeon Mac Pro.

Thing is, it's true: Apple Silicon was designed for mobile first, small desktop form factor second and high performance desktop a distant third. All very logical given the PC market, but it has left no additional performance on the table for a large desktop design like the Mac Pro, which can easily handle cooling massive amounts of watts.

No, it was designed for higher performance and efficiency through tighter integration and lower modularity. The tower form factor is just no longer needed. Just like the microprocessor eliminated mainframes, the system on a chip eliminated PC boxes. Going bigger than a Mac Studio does not yield any performance gains. Which doesn't mean that the Studio is slow or slower than it could be.

 

[playtech1]

M2 Ultra is literally twice as fast as the fastest Intel Xeon Mac Pro.

Being faster than a 2019 Xeon isn't really much to write home about - I mean the original basic M1 had faster single core than the Mac Pro's Xeon W-3223 and similar multi-core. M2 Ultra seems about the same as a top-end consumer Intel CPU, which is fine but also not all that exciting.

Main issue is that the Intel CPU doesn't have the same RAM and GPU constraints as Mac Pro. At the workstation level there are also x64 CPUs that have way more cores than M2 Ultra.

No, it was designed for higher performance and efficiency through tighter integration and lower modularity. The tower form factor is just no longer needed. Just like the microprocessor eliminated mainframes, the system on a chip eliminated PC boxes. Going bigger than a Mac Studio does not yield any performance gains. Which doesn't mean that the Studio is slow or slower than it could be.

I disagree - IMHO Apple made trade-offs in its design to prefer cool and quiet to be the best laptop CPU, because that's where they make their money. They could have made different trade-offs that would have made it faster on the desktop but harder to power and cool and so worse in a laptop. A 5Ghz M2 Ultra would have amazing performance, but it just doesn't look like it's ever going to run that fast.

Don't get me wrong, I'm actually very happy with the trade-offs Apple made, but I don't think we should pretend they don't exist.

 

[thebart]

MP sales represent 0.3% of Apple's $400B annual revenue. That percentage is three times the ~0.1% that Bugatti sales represent out of the VW group's $250B annual revenue. And I'm sure Bugatti's R&D costs are substantial. Does that mean VW should dump Bugatti?

I didn't know VW owns Bugatti. In that sense, maybe they should sell it

But there's no mistaking the Mac pro being an Apple product. Honestly I don't know why pros even buy the MP, seeing how Apple has neglected it. And a few people who bought the 2019 MP must have felt pretty burned when the studio M1 did some tasks better at a fraction of the cost, noise, power consumption.

Why would someone sink money into a product Apple glossed over in two minutes? There are a couple of possibilities: a) this is a stop gap and they'll be a real ASi MP in a year or two, with at least an extreme chip, maybe expandable RAM, and either 3rd party GPU support or beefed yup GPU that is somewhat competitive in 3d rendering and ML. Or b) this is Apple giving the MP the kiss of death and they'll quietly drop it and when asked about it they'll say "well we listen to our users, who clearly prefer the studio."

Whichever way it goes, I'm sure the people who defend the 2023 MP will be back to vindicate Apple.

 

[theorist9]

I think it was clear “dead” and “doesn’t exist” refer to the market being too small to develop specific hardware for.

We can argue about the MP separately (and I did acknowledge that "there are limits to the development dollars they can put into it"). But it was also clear your statement was about the HEDT market as a whole, not merely Apple's HEDT market. And it was based on statements like "Threadripper hasn't had a new release in years and the TR 7000 series keeps slipping." But that discounts the fact that AMD has been working on TR 7000, and *is* releasing it (https://www.tomshardware.com/news/r...0-storm-peak-cpu-surfaces-with-64-zen-4-cores). Release dates often slip. That hardly means a market is "dead".

So again, how do you explain AMD's substantial investment in the TR 7000 if the market is "too small to develop specific hardware for"?

I'm reminded of Mark Twain's famous letter to the New York Journal in response to an inquiry about his possible death: "...I have even heard on good authority that I was dead....The report of my death was an exaggeration."

 

[VivienM]

There can be no doubt that Apple always tried to make their Macs as quiet as possible and the Mac Studio is no exception.

Always? I'd like to introduce you to my Power Mac G4 MDD, which is the loudest computer I've ever owned, Windows or Mac, in over 30 years. The second closest would probably have been a home-built Preshot Deleron around 2005-8 whose CPU fan would ramp up to 5000RPM if the room air was on the warm side, and I think the poor thermals in that machine probably have more to do with me having picked a less than optimal case, etc. - a professionally-engineered system would have run the same CPU just fine.

With the MDD, what happened is relatively clear - the G4's power consumption and heat output, especially if you had two of them, went up much more than a case design dating back to a single 400MHz G3 could handle, and the solution to just run the fans faster made for some serious noise. Something they corrected for with their next enclosure design.