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 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
Yes, please tell us who your customers are that are pirating MacOS. I want to know who has zero respect for the intellectual property of others so I can be sure none of my intellectual property ever flows through their businesses.
He won’t answer because of the privacy of those large companies running hackintosh, I bet big Hollywood studios are among them.
Things we never lack off:
ARM-mac dreamers
8K- Thunderbolt Display dreamers
Modular Lego-like dreamers...
Last breakthough dreamers...
..No chance keep dreaming
All-AMD hopefuls...
nVidia comeback hopefuls..
xMac hopefuls...
OpenCL, HIP, CUDA, GPU-Tensorflow etc hopefuls...
... Little hope at least you have right to dream, sadly we have to wake up without these.
ARM-mac dreamers
8K- Thunderbolt Display dreamers
Modular Lego-like dreamers...
Last breakthough dreamers...
..No chance keep dreaming
All-AMD hopefuls...
nVidia comeback hopefuls..
xMac hopefuls...
OpenCL, HIP, CUDA, GPU-Tensorflow etc hopefuls...
... Little hope at least you have right to dream, sadly we have to wake up without these.
I'd agree with that. I know of some very large organisations that are running hooky software and also a few that try to distance themselves by employing sub contractors that they know are using hooky software instead.
Given the fact that since the trashcan Apple has screwed over those who wanted a decent, not thermally constrained and upgradable desktop, I don't blame them.
I was clearly sarcastic, in my experience ther's no way large companies choose hackintosh or assembled PC
I assume all of that was directed at me. I can tell you right now that more than one executive at Apple has publicly stated that they are desktop class. But my point was that you cannot blame your lack of performance on someone else like that.
Enough,
ARM-mac dreamers, Read: there is no ARM-powered Mac in Apple's roadmap, the reason is ARM' scalar performance still halfway to x86-64 scalar performance, thus means a typical single threaded GUI routine non viable for smp (multi-threaded) re-write will run half slow as on a true desktop x86 CPU, too many low-level GUI related macos routines are by definition single threaded, it means can't or are hardly to be rewrite for smp, so despite some ARM' CPU s claiming desktop performance, this is related to smp performance not to scalar performance, I'd like someone here sometime tried macOS in 5k with 24 cores iMac pro, and sameday macOS on 5K i9 iMac (8core, faster scalar), while the scalar performance among a 24core Xeon and a 8 core i9 isn't as dramatic, its easy to realize what would be an much slower ARM' Mac experience.
An ARM-mac will not deliver the user experience Apple offers even with i5 MBA, slower desktop/laptop is something windows or Linux can afford as their userbase is wider, and where the definition of "usable pc" largely varies.
Expect more and even more auxiliary ARM' cores for low power sub-system tasks (as background downloads), even for very optimized applications as video transcoding, but not for kernel neither GUI or mainstream processing.
I mean apple toyed with arm Macs as a just-in-case readiness excersise or even as a feasibility research for macOS servers re-launch, or considering ultra-low cost educational markets, but with AMD-Intel performance race reloaded, when some ARM cpu reaches some Intel/AMD performance , those beaten Intel CPU should have been replaced with much faster ones long ago.
Few facts, when arm promised years ago to reach Intel single core performance they targeted 8-16 core Xeon-D Intel's efficiency champ, but even today no arm cpu has faster scalar performance, Intel only beaten at smp performance (at the exoevse of twice or more cores/Chip), since then AMD and Intel (2 years) increased ipc scalar performance at about 21 an 39%, so nowadays arm dream to catch x86 st/ipc/scalar performance is even farthest, and maybe will never happen, arm seems aware on that and now is focusing it development on ML related optimizations and their own tpus.
ARM-mac dreamers, Read: there is no ARM-powered Mac in Apple's roadmap, the reason is ARM' scalar performance still halfway to x86-64 scalar performance, thus means a typical single threaded GUI routine non viable for smp (multi-threaded) re-write will run half slow as on a true desktop x86 CPU, too many low-level GUI related macos routines are by definition single threaded, it means can't or are hardly to be rewrite for smp, so despite some ARM' CPU s claiming desktop performance, this is related to smp performance not to scalar performance, I'd like someone here sometime tried macOS in 5k with 24 cores iMac pro, and sameday macOS on 5K i9 iMac (8core, faster scalar), while the scalar performance among a 24core Xeon and a 8 core i9 isn't as dramatic, its easy to realize what would be an much slower ARM' Mac experience.
An ARM-mac will not deliver the user experience Apple offers even with i5 MBA, slower desktop/laptop is something windows or Linux can afford as their userbase is wider, and where the definition of "usable pc" largely varies.
Expect more and even more auxiliary ARM' cores for low power sub-system tasks (as background downloads), even for very optimized applications as video transcoding, but not for kernel neither GUI or mainstream processing.
I mean apple toyed with arm Macs as a just-in-case readiness excersise or even as a feasibility research for macOS servers re-launch, or considering ultra-low cost educational markets, but with AMD-Intel performance race reloaded, when some ARM cpu reaches some Intel/AMD performance , those beaten Intel CPU should have been replaced with much faster ones long ago.
Few facts, when arm promised years ago to reach Intel single core performance they targeted 8-16 core Xeon-D Intel's efficiency champ, but even today no arm cpu has faster scalar performance, Intel only beaten at smp performance (at the exoevse of twice or more cores/Chip), since then AMD and Intel (2 years) increased ipc scalar performance at about 21 an 39%, so nowadays arm dream to catch x86 st/ipc/scalar performance is even farthest, and maybe will never happen, arm seems aware on that and now is focusing it development on ML related optimizations and their own tpus.
Last edited:
If only Apple could license the ARM architecture and build their own design from scratch that had great single thread performance.
(They did, and they did. A13 has extremely good single thread performance. With all the fancy sorts of branch prediction that desktop chips had.)
That said, I'm still very skeptical about it coming to something like the Mac Pro anytime soon. There still are certain characteristics of Intel's architecture that make it ideal for pro work.
But all the stuff you're talking about means exactly jack to someone who wants to do email on their MacBook Air. SMP performance blah blah blah all they know is that their ARM MBA has twice the battery life and to them feels fast browsing YouTube. That's where we'll see ARM come into play.
The iPad has already proven Apple could do a well performing ARM device on the low end that is competitive with Intel. I would even bet the iPad Pro will be the basis of the ARM MBA platform.
...for about 8 seconds. Until we see an A(x) chip with a proper cooler attached, demonstrate that it's capable of loading to 100% for minutes, hours etc, I'm not buying the "desktop class" hype. That's too much like Apple's usual PR move of trying to redefining words to suit their preferred meaning.
It’s trivial to remove the heat generated by a few watts of power consumption, even with an improper cooler
A little more difficult when you get to a few hundred watts, but still a long-solved problem with well known solutions.Not sure what you refer to wrt Apple’s PR “redefining words to suit their preferred meaning” but that’s another subject altogether.
Last edited:
That’s not necessarily true.It’s trivial to remove the heat generated by a few watts of power consumption, even with an improper cooler
Not sure what you refer to wrt Apple’s PR “redefining words to suit their preferred meaning” but that’s another subject altogether.
if you have a few cubic inches for a heat sink then yes, but if you only have cubic centimetres for the same heat dissipation then, no.
The OP is correct. Until we see the results then the Apple spin means little.
Why wouldn’t a desktop class Ax have a desktop class cooling solution? Why would it be limited to a few cubic centimeters?
Both your and OP’s concern regarding the difficulty of actively cooling an Ax chip is unwarranted.
Hilarious that you’d say that knowing the Apple track record with cooling problems.
Never the less my statement is true. Just because an Ax is good at what it’s doing doesn’t mean it’ll scale well.
iMac Pro cools just fine, and even the 80+ Watts of power dissipated by the 8-core I9-9980HK and GPU in the 15” MBP is effectively handled by a laptop cooling solution.
You can laugh all you want, but there’s absolutely no basis for thinking that the first 8 seconds of performance can’t be maintained for minutes, hours or days with active cooling—or a passive cooling solution of adequate design/thermal mass, for that matter.
I know it may be upsetting to see the incredible single-core and multi-core benchmarks for Ax chips, but pretending Apple’s silicon isn’t highly performant, and couldn’t be easily improved with the “proper cooling” contemplated by OP, doesn’t change reality. ¯\_(ツ)_/¯
Correct. It may be upsetting. To me, it is not. But I’m stunned that you can’t see that on paper does not equal in reality.
You’re stunned to think a part that operates at 3-5 Watts could be continuously cooled? OK, maybe you are?
On paper or in practice, this is not difficult. Physics/thermodynamics is a thing. It’s not really up for debate, and it’s certainly nothing to be “stunned” by, in my opinion, but yours obviously differs. Some opinions are right, some are wrong ¯\_(ツ)_/¯
I still think this will play out as some form of coprocessor - almost (but obviously much more complex) than the hi/low core combinations on iPads, where some cores are focused on crazy energy efficiency some are focused on performance, and the system adjusts based on needs.
You're right that nothing Intel/x86 is required to browse the web or read email, but IMO the chances of Apple having a split product line with some on x86 and some on ARM, without a 'transitioning everything' plan, is effectively 0.
We differ, that’s fine. But what’s not up for debate are the following two statements;
Practice and theory are often different, (as I alluded to earlier).
Companies make imperfect designs all the time.
They are not opinions.
It’s trivial to remove the heat generated by a few watts of power consumption, even with an improper cooler
On iOS, if you spike the processor for more than a few seconds, your app gets killed by the system - that's why I don't buy a lot of the "desktop-class" claims about A(x) chips - noone has demonstrated an A(x) chip running outside of the very contrived constraints of iOS, which does not facilitate "desktop computing" by design. Can the A(x) actually handle being at 100% for a minute, an hour, 3 days etc.
Before talking about how inevitable A(x) desktops are, it would be useful to have actual evidence of them being able to survive and perform under the sort of duty cycle actual desktop chips experience.
I agree with both of those statements, and neither of them are necessarily relevant to my contention that there’s no reason to think Apple’s Ax can’t be adequately cooled with the same active (and passive) cooling techniques that are used to cool every other chip.
With the expertise and demonstrated capabilities of Srouji’s silicon development group, I can’t imagine they wouldn’t be unable to design and package a CPU/GPU that would run at sufficiently high sustained clock rates at moderate power levels, say 50-100 Watts.
I don’t understand what you think would prevent Apple’s chips from being adequately cooled by active cooling. My 1st gen iPad Pro can get quite warm when playing GPU-intensive games but there is certainly a decent level of sustained performance even with passive cooling.
Maybe @cmaier will pop in if he has the time (and inclination) to comment on the prospects of Apple scaling up their silicon to, say, 50-150W of power dissipation. It would be very informative I’m sure to hear from someone with actual CPU design experience.
SNAPPY!
Un-obtainium OPTane
"This is a huge deal. Adding Intel Optane DCPMM to workstations will allow for higher memory capacities and faster storage. With AMD pushing higher core counts and features like PCIe Gen4 in 2019, Intel is engaging in asymmetric warfare by bringing in Optane DCPMMs which are an Intel-only technology. While AMD may win the core count wars, Intel will have something different to offer for memory and storage that AMD cannot match."
Coming to a Macpro next gen in 2026? Will Apple refresh designs faster than every 7 years?
Un-obtainium OPTane
"This is a huge deal. Adding Intel Optane DCPMM to workstations will allow for higher memory capacities and faster storage. With AMD pushing higher core counts and features like PCIe Gen4 in 2019, Intel is engaging in asymmetric warfare by bringing in Optane DCPMMs which are an Intel-only technology. While AMD may win the core count wars, Intel will have something different to offer for memory and storage that AMD cannot match."
Coming to a Macpro next gen in 2026? Will Apple refresh designs faster than every 7 years?