The problem with this statement is it doesn't define what situation applies to best. For example it's not best at running software written for macOS.
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.
MP All Models Is Apple ****ing themselves over the ThreadRipper 3 yet?
- Thread starter macjunkie2013
- Start date
- Sort by reaction score
https://www.anandtech.com/show/1512...phire-rapids-cpus-with-six-ponte-vecchio-gpus
Well, Aurora Supercomputer, which is going to be finished in 2021, will have Sapphire Rapids Xeons, so I am sure Sapphire Rapids Xeons will be available in 2021. Since you can't win $500m contract on vaporwares.
Not "Finished." "Delivered." Huge difference. Delivered means working proof of concept. That may be 10-15% of the CPU's and a lot of empty racks.
You can win a $500m contract on vaporware. Intel just did.
Vaporware:
- software or hardware that has been advertised but is not yet available to buy, either because it is only a concept or because it is still being written or designed.
I will just respond in this way.
Show me a SINGLE roadmap where Van Gogh is touted by AMD. A single one. We have seen Renoir, Dali, Matisse, Vermeer. Van Gogh does not exist on a single AMD, official rodmap. Why?
Because it is semi Custom product. And Semi-Custom products can only appear in products ordered by specific clients. Which we have historically seen, and you are trying to diminish that. It even does not come in FP5 socket, like other BGA AMD products come.
I will put the stake higher. If you will find a Single product roadmap where Van Gogh exist, I will publicly admit the fact it means nothing that we have AMD APUs in Mac os Catalina Beta. Find a single one official product roadmap. Then we can talk about it being something more than just semi-custom project.
Secondly. Too small volume? That APU will land not only in MacBook Pro's, but can also land in Mac Mini's and 4K iMac's. That is not a small volume.
Thirdly. Semi-Custom AMD's business works in a simple way. Its clients who are ordering specific products, and paying for their development. It is AMD clients who are ordering TSMC's wafers for those products. AMD gets only small percentage from licence's. Never before in their business we have seen a situation, where, they would, as you have put it: "AMD easily put those folks, who demanded bigger GPU in their APUs into bigger group". Why? Go back to previous two points. Van Gogh is not a official roadmap product.
If it is in MacOS Kexts, it could've been ordered ONLY by Apple. There is no hand waving here. There is your lack of understanding of the situation.
im glad the Mac mini pro worked for you and many others. But many left the Mac because it didn’t work for them. Apple realized this too late and then wen on an apology tour that their design failed. Yea, that is a marketing and sales disaster.
I never got the impression that the 2013 was a sales disaster. They had a lot of very large orders for them, up until the end.
I don't think it hit their targets, but it seemed like they sold more than enough of them to make a good profit.
The design was a disaster in that they couldn't continue the series. But that's different than a sales disaster.
Yea, the traschcan had no where near the volume of the previous models. To be fair, it never got any updates and the previous cheese grater did, but nonetheless, sales wise, it was a relative flop. Furthermore, it's still a design disaster. A marketing disaster. And an operations disaster. Not shipping a replacement for over 6 years because you designed yourself into a "thermal corner" or more colloquially you **** the bed', that is a disaster.
Last edited by a moderator:
Can you link the apple financial report you are referencing to make your statement about sales? I need to see. The units made, units sold, units expected to sale, and other ROI numbers to decide if the Tｒａｓｈｃａｎ was successful or not as an investment for apple. I have read comments made by apple executives that they admit boxing themselves into a specific ｄｅｓｉｇｎ．
2. That is specifically announced when unready, in order to dissuade customers from committing to competitor’s solutions. E.G. 2017 (announced) Mac Pro, Adobe Photoshop “full / real Photoshop” for iOS.
I will never understand folks calling this an “apology tour”.
Well, I guess Apple did go out of their way to book visits to meet those five journalists... wait. No they didn’t, the journalists traveled to Apple‘s Campus.Not much of a tour! LOL
What will drive any Apple decision to move to AMD is MOBILE (notebooks - they're totally happy with ARM on iOS, as they should be).
Apple's sales breakdown is 80% notebook, 20% desktop. I'm almost sure that each major notebook model (16" Pro, 13"Pro, Air) independently outsells all desktop Macs combined.
AMD has nothing that would fit in the Retina MacBook Air, much less a MacBook refresh.
AMD has an inferior mobile APU (with markedly superior GPU) that would work for the 13" Pro - it would be a CPU performance downgrade, but the GPU upgrade is possibly worth it. However, note that the Surface Laptop (the only premium Ryzen notebook) isn't getting great reviews - people are going to Microsoft's Business store to get Ice Lake instead. A Zen 2 version of the low-power AMD APUs is due next year, and that should be a better choice.
AMD has literally nothing that would work for the 16" MBP. How would pros feel if the fire-breathing 8 core notebook workstation with 64 GB of RAM and an 8 TB drive that has left them feeling heard was snatched away and replaced by a low-power quad core with most of its power in an integrated GPU?
Yes, it's theoretically possible to run one of the Ryzen APUs with a discrete GPU, but why would you want to? Their strongest feature is the integrated GPU (perfect for the 13" MBP, not for the 16")
Even while the 15" MBP was drawing complaints for butterfly keyboards and old Polaris GPUs, CPU performance marched on. The 2016 was disappointing - just like a 2015, 2017 got a surprising performance boost, 2018 was significantly faster (extra cores) and could take more RAM (about time), but had initial cooling problems, 2019 was very fast. Now that there's a new design with excellent initial reviews, a 50% performance hit isn't what Apple needs.
On the desktop side, new Ryzens are mostly a very good fit, assuming the iMac Pro can cool a Threadripper. If not, there's not enough CPU differentiation to maintain the iMac Pro as a separate product (it becomes a high-end CTO of the 27" iMac with a Ryzen 9 3950X, which might be OK) - only the Mac Pro gets Threadripper in that case.
The other issue on the desktop side is the supply problems with the 3900X and the delayed launch of the 3950X. If you assume that 3900X and/or 3950X are maybe available in iMac Pro quantities, but not in mainstream 27" iMac quantities, the top chip on the 27" iMac becomes a Ryzen 7 3700X, which is about equivalent to a an i9-9900K. No big loss, no big gain. There are some gains lower in the line, but the big gains are to two exotic systems that make up a small fraction of Apple's sales.
The iMac Pro gets much better price/performance (and probably becomes a high-end CTO iMac) It gets a 12-core Ryzen 9 3900X as its base CPU, with a $400 upgrade to a 3950X. It only gains a little speed compared to similar core count Xeon-W configurations, but the 12 core comes in at the price of the 8 core base model, and the 16 core costs like the existing 10 core, but performs like the 18 core.
The Mac Pro gets much better chips. Instead of being 8 to 28 cores, it starts at its current base price with a 24 core Threadripper, has a $1000 upgrade to 32 cores, a $2000 upgrade to 48 cores and a $3000 upgrade to 64 cores...
The problem is that moving to AMD would eliminate 5 million or more MacBook Airs per year, cripple 5 million or so 16" MacBook Pros and really benefit 100,000(?) Mac Pros and a million or less high-end iMacs/iMac Pros.
If Apple were to move the MacBook Air and maybe some Minis to ARM, everything that can to AMD, and they were willing to accept two architectures (ARM and AMD), the only remaining problem would be the 16" MacBook Pro. Do they keep one machine on Intel? Can they convince AMD to make some special chips?
Apple's sales breakdown is 80% notebook, 20% desktop. I'm almost sure that each major notebook model (16" Pro, 13"Pro, Air) independently outsells all desktop Macs combined.
AMD has nothing that would fit in the Retina MacBook Air, much less a MacBook refresh.
AMD has an inferior mobile APU (with markedly superior GPU) that would work for the 13" Pro - it would be a CPU performance downgrade, but the GPU upgrade is possibly worth it. However, note that the Surface Laptop (the only premium Ryzen notebook) isn't getting great reviews - people are going to Microsoft's Business store to get Ice Lake instead. A Zen 2 version of the low-power AMD APUs is due next year, and that should be a better choice.
AMD has literally nothing that would work for the 16" MBP. How would pros feel if the fire-breathing 8 core notebook workstation with 64 GB of RAM and an 8 TB drive that has left them feeling heard was snatched away and replaced by a low-power quad core with most of its power in an integrated GPU?
Yes, it's theoretically possible to run one of the Ryzen APUs with a discrete GPU, but why would you want to? Their strongest feature is the integrated GPU (perfect for the 13" MBP, not for the 16")
Even while the 15" MBP was drawing complaints for butterfly keyboards and old Polaris GPUs, CPU performance marched on. The 2016 was disappointing - just like a 2015, 2017 got a surprising performance boost, 2018 was significantly faster (extra cores) and could take more RAM (about time), but had initial cooling problems, 2019 was very fast. Now that there's a new design with excellent initial reviews, a 50% performance hit isn't what Apple needs.
On the desktop side, new Ryzens are mostly a very good fit, assuming the iMac Pro can cool a Threadripper. If not, there's not enough CPU differentiation to maintain the iMac Pro as a separate product (it becomes a high-end CTO of the 27" iMac with a Ryzen 9 3950X, which might be OK) - only the Mac Pro gets Threadripper in that case.
The other issue on the desktop side is the supply problems with the 3900X and the delayed launch of the 3950X. If you assume that 3900X and/or 3950X are maybe available in iMac Pro quantities, but not in mainstream 27" iMac quantities, the top chip on the 27" iMac becomes a Ryzen 7 3700X, which is about equivalent to a an i9-9900K. No big loss, no big gain. There are some gains lower in the line, but the big gains are to two exotic systems that make up a small fraction of Apple's sales.
The iMac Pro gets much better price/performance (and probably becomes a high-end CTO iMac) It gets a 12-core Ryzen 9 3900X as its base CPU, with a $400 upgrade to a 3950X. It only gains a little speed compared to similar core count Xeon-W configurations, but the 12 core comes in at the price of the 8 core base model, and the 16 core costs like the existing 10 core, but performs like the 18 core.
The Mac Pro gets much better chips. Instead of being 8 to 28 cores, it starts at its current base price with a 24 core Threadripper, has a $1000 upgrade to 32 cores, a $2000 upgrade to 48 cores and a $3000 upgrade to 64 cores...
The problem is that moving to AMD would eliminate 5 million or more MacBook Airs per year, cripple 5 million or so 16" MacBook Pros and really benefit 100,000(?) Mac Pros and a million or less high-end iMacs/iMac Pros.
If Apple were to move the MacBook Air and maybe some Minis to ARM, everything that can to AMD, and they were willing to accept two architectures (ARM and AMD), the only remaining problem would be the 16" MacBook Pro. Do they keep one machine on Intel? Can they convince AMD to make some special chips?
No need to think they can't use multiple suppliers as they do with pretty much everything else.
It's the high-end that is suffering from Intel, mobile is fine as long as AMD have nothing noteworthy.
E.G. Xeons that are 2-3 generations ahead and 2 process nodes beyond what Intel can currently mass produce.
I don't know what Intel's contract with Apple looks like, but I suspect Apple gets much better pricing by agreeing to be all Intel. They also haven't updated any desktop since Zen 2 came out.
The MacBook Pro has (recently) been the exception to Apple being slow to release updates. The 2018, 2018 Vega, 2019 15" and 2019 16" all came out soon after the CPUs or GPUs were ready (this was after a period of neglected updates). The longest gap in that period was the 2018 CPU refresh, which was early July when the CPUs were late April. There have been two CPU refreshes and two GPU refreshes (counting the November 2019 redesign as a GPU refresh, which it was, among other things) since July 2018.
The 2019 iMac also uses current CPUs at the top end and current upgrade GPUs - the stock GPU is ancient, and some of the lower end CPU choices are a generation old. The iMac with the 19-9900K came out about 5-6 months after its CPU was ready (although the previous CPU generation got skipped).
The MacBook Pro has (recently) been the exception to Apple being slow to release updates. The 2018, 2018 Vega, 2019 15" and 2019 16" all came out soon after the CPUs or GPUs were ready (this was after a period of neglected updates). The longest gap in that period was the 2018 CPU refresh, which was early July when the CPUs were late April. There have been two CPU refreshes and two GPU refreshes (counting the November 2019 redesign as a GPU refresh, which it was, among other things) since July 2018.
The 2019 iMac also uses current CPUs at the top end and current upgrade GPUs - the stock GPU is ancient, and some of the lower end CPU choices are a generation old. The iMac with the 19-9900K came out about 5-6 months after its CPU was ready (although the previous CPU generation got skipped).
Where do the rumored Apple ARM Macs fit into this discussion?
https://www.macrumors.com/guide/arm-macs/
https://www.macrumors.com/guide/arm-macs/
Everybody assumed that when Bloomberg posted an article about Apple ditching Intel and going with their own designed Chips it has to be ARM based.
But building Semi-Custom x86 products with AMD's newest technology also constitutes for the idea of "Apploe designed" chips. They are ordered by Apple, design is paid by Apple, and it will be Apple who will order the wafers on TSMC's processes, which will mean that Apple will have capacity required to feed the volume.
Switching from x86 to ARM is much more expensive than just switching x86 vendors. Not only for Apple, but also - everybody else. Switch from x86 to ARM will effectively break the software compatibility.
The latest initiative to run iPad/iPhone Apps on Macs is emulating ARM Apps on x86. If Apple would switch from x86 to ARM, they would the other way around. Emulate x86 software on ARM. Why are they not doing this?
But building Semi-Custom x86 products with AMD's newest technology also constitutes for the idea of "Apploe designed" chips. They are ordered by Apple, design is paid by Apple, and it will be Apple who will order the wafers on TSMC's processes, which will mean that Apple will have capacity required to feed the volume.
Switching from x86 to ARM is much more expensive than just switching x86 vendors. Not only for Apple, but also - everybody else. Switch from x86 to ARM will effectively break the software compatibility.
The latest initiative to run iPad/iPhone Apps on Macs is emulating ARM Apps on x86. If Apple would switch from x86 to ARM, they would the other way around. Emulate x86 software on ARM. Why are they not doing this?
AMD's newest technology still doesn't perform as well per watt as Apple's ARM's solutions, though. I think the biggest "hint" of the future of macOS is the transition to 64-bit only code. Both Intel's and AMD's solutions support 32-bit and lower. If they were continuing with either, they could have kept 32-bit in Catalina. They did the same with iOS, first required 64-bit code on a chip that supported 64 and 32, and then released the 64-bit only version of the systems (iPhones and iPads). I guess Semi-Custom x86 COULD mean AMD that's 64 only, though...
Are they emulating ARM on x86 OR recompiling the code for Intel? I was under the impression that it was the latter. The Xcode simulator is performing emulation, but when you build an app for Intel, it's not building ARM code then running it in emulation, it building Intel executable code.
Intel already expecting Apple to start ditching Intel CPU from 2020. I think Macrumors have this article.
The project call Catalyst from Apple enables to use of iOS app on macOS. We've been using a lot of iOS apps since Yosemite.
[automerge]1575230627[/automerge]
That's because APU has an actual graphics card separately.
No amount of pie in the sky thinking, wish fulfillment or secret codenames in macOS betas are going to trump the fact that Apple wants to control its own destiny, which means moving to ARM/Arm/Axx CPUs, not AMD and their product roadmap. They are already hostage to their GPU roadmap...
AMD, as far as they have come, does not have the CPU portfolio or the volume manufacturing Apple would want or need. AMD mobile is nowhere as far along as Intel. Apple is not going to sell some things on Intel, some on AMD, it’s confusing to the end user and means that models end up competing against themselves to an extent. Explaining the nuances of a 6-core Ryzen being faster than an 8-core Core i9 mean absolutely zero to 90% of Apple’s potential Mac purchasers because trying to explain it to someone who is simply looking for their next MacBook Air just does not care enough.
Sorry, but there is no reality in which Apple switches to all AMD - the portfolio isn’t there and the volume manufacturing isn’t there.
No.
Catalyst does not emulate ARM. It does not run existing iOS apps.
Apps have to be recompiled. They're still running in a giant compatibility environment, but they're not being emulated, and it's still not the same environment as iOS. The Catalyst environment is basically half Mac, half iOS. But it's not at all the same platform or code target as your iPhone or iPad.
Not all iOS apps will work in Catalyst either. Some require a moderate amount of porting.
It's only been part of the OS since Mojave. Yosemite was way before this.
SwiftUI also seems like the actual goal for Apple, not Catalyst. Catalyst is a quick stop gap hack to hold things together until developers can start writing multi platform apps in SwiftUI. SwiftUI is also compiled for the native CPU.
Last edited: