500W for ram / cpu is extreme maybe 2 cpu's also 10GB Ethernet ports is likely to be build in.
The top HP-Z has an 1450W PSU but that is likey to high end of an system for apple to go.
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
The top HP-Z has an 1450W PSU but that is likey to high end of an system for apple to go.
How do Dell, HP, and countless other system builders do it, then? The configuration you describe is entirely unremarkable in the world of workstation PCs, nearly all of which are plugged into a single 15A outlet.
Yeah whatever form factor Apple creates, they aren't going to create something that meets most or the top end of something like the Z8.
The Z6 tops out at 1000W and I think that's a far more likely power envelope. That would still let Apple produce a machine that can do dual CPUs and dual GPUs easily, so it's not like it's limited.
It's very limited if your system needs 6000 watts - using four high voltage (208V) 1500 watt power supplies.
Dual socket, dual GPU is a bit starved at 1400 watts input - and a 110V 15A circuit is capped at about 1240 watts. A dedicated 110V 20A circuit is capped at 1650 watts input.
And note that power supply specs are for the *output* from the PS - any inefficiencies mean a higher draw from the mains. (A 94% efficient 1500 watt PS will draw about 1600 watts from the mains.)
Last edited:
The W3175 is close to 300 watts on its own, without overclocking it and without the draw of anything else that's in the base (chipset, RAM, cooling). I don't think a 500 watt base is unrealistic prior to any GPU or storage draws (with that particular CPU - most other choices will have lower draws by 100 watts or more).
You can squeeze essentially any combination of 1 CPU and 1 GPU (certainly anything Apple would offer) inside 15A, even with a lot of RAM and storage plus some non-GPU PCIe cards.
Dual CPU OR dual GPU configurations will generally fit, with some extremely high-draw components getting tight (2 300+ watt GPUs, for example). It would be hard to come up with a configuration with only one dual component that wouldn't fit in the limitations of a 20A socket (and many higher-end workstations are 20A)
Configurations with two CPUs AND two GPUs will generally require 20A, with certain combinations having a very hard time fitting in 20A (two 300W CPUs and two 300W GPUs are really, really pushing it, and it would be very easy to go over with RAM, storage or PCIe).
Many configurations that use >2 GPUs (or CPUs, but I see very little chance that Apple supports quad-CPU configurations) will be over the limit.
The Z6 uses much lower power CPUs, if the W3175 rumor is right... They offer the Platinum 8180, which is a lower-clocked relative of the W3175 with a lower power draw (much more expensive because of its multiprocessing features) , but they don't support two of them! CPUs they support two of are limited to 150W apiece. They do something similar with GPUs - the real hogs are limited to one per system.
Many mining rigs that use multiple GPUs are connected to more than one 20A circuit, and some workstations presumably are as well.
You can squeeze essentially any combination of 1 CPU and 1 GPU (certainly anything Apple would offer) inside 15A, even with a lot of RAM and storage plus some non-GPU PCIe cards.
Dual CPU OR dual GPU configurations will generally fit, with some extremely high-draw components getting tight (2 300+ watt GPUs, for example). It would be hard to come up with a configuration with only one dual component that wouldn't fit in the limitations of a 20A socket (and many higher-end workstations are 20A)
Configurations with two CPUs AND two GPUs will generally require 20A, with certain combinations having a very hard time fitting in 20A (two 300W CPUs and two 300W GPUs are really, really pushing it, and it would be very easy to go over with RAM, storage or PCIe).
Many configurations that use >2 GPUs (or CPUs, but I see very little chance that Apple supports quad-CPU configurations) will be over the limit.
The Z6 uses much lower power CPUs, if the W3175 rumor is right... They offer the Platinum 8180, which is a lower-clocked relative of the W3175 with a lower power draw (much more expensive because of its multiprocessing features) , but they don't support two of them! CPUs they support two of are limited to 150W apiece. They do something similar with GPUs - the real hogs are limited to one per system.
Many mining rigs that use multiple GPUs are connected to more than one 20A circuit, and some workstations presumably are as well.
It's very limited if your system needs 6000 watts - using four high voltage (208V) 1500 watt power supplies.
Dual socket, dual GPU is a bit starved at 1400 watts input - and a 110V 15A circuit is capped at about 1240 watts. A dedicated 110V 20A circuit is capped at 1650 watts input.
And note that power supply specs are for the *output* from the PS - any inefficiencies mean a higher draw from the mains. (A 94% efficient 1500 watt PS will draw about 1600 watts from the mains.)
There is a 100% chance you will be disappointed in whatever Apple puts out, even if it were the second coming of the cheesegrater.
The Mac Pro always was a midrange tower. At best, they give a little more wiggle room to people who spec it higher, but the people looking for a $1400 entry-level workstation or a cluster machine with 48 cores are always going to be disappointed whatever comes.
Which is sad because the Z6 seems to cover most of that ground - going from $1400 i7/i9, all the way up to mega core count and cost Xeons.
Okay-not to be “that guy” but...
With today’s report that intel believes Apple will be transitioning to their own custom ARM, isn’t this a huge red flag for the Mac pro? I mean why put 3 years of R&D (even if it’s one guy in the basement) into a product that won’t exist longer than what it did in design. Wouldn’t it make more sense to go ahead and “preview” at WWDC with the developer announcement and say “oh, and we will be shipping our first ARM Mac this December” -hitting their self imposed 2019 product window. I mean at the most extreme I think this is the most likely scenario, unless they are involved in some sort of hybrid that can handle the transition. Hell why not have a machine that can switch between the two, pitch that to the developer audience. (I admit I think that is rather absurd but Apple 2019 who knows).
With today’s report that intel believes Apple will be transitioning to their own custom ARM, isn’t this a huge red flag for the Mac pro? I mean why put 3 years of R&D (even if it’s one guy in the basement) into a product that won’t exist longer than what it did in design. Wouldn’t it make more sense to go ahead and “preview” at WWDC with the developer announcement and say “oh, and we will be shipping our first ARM Mac this December” -hitting their self imposed 2019 product window. I mean at the most extreme I think this is the most likely scenario, unless they are involved in some sort of hybrid that can handle the transition. Hell why not have a machine that can switch between the two, pitch that to the developer audience. (I admit I think that is rather absurd but Apple 2019 who knows).
Mayve, with the stackable modular approach, there will be a transitory module with an ARM CPU that runs as a secondary unt to the main Brain module...? As Apple gets things going for HPC on ARM, pne eventually moves the Xeon Brain module to another task & the ARM Brain module takes over with the remaining modules (GPU / Storage / I/O / etc.)...?
and $3K-$5K+ workstation that is on the road to app store only + no windows boot and SLOW X86-64 vm's.
That Intel Xeon Brain module that just got replaced with the new HPC ARM Brain module, it can still boot to Windows...
I don't think Apple will be using ARM only in all their products.
Like ive said before :
ARM : MB Air,MB,iMac
Intel : MBP,iMacPro,Mac Pro
Like ive said before :
ARM : MB Air,MB,iMac
Intel : MBP,iMacPro,Mac Pro
The intel report didn’t seem to indicate that, and I would imagine they would have some insight on what their ‘buyers’ intentions are to a degree. If Apple was using ARM as a negotiating tactic and seemingly not getting anywhere I am sure they would have some signaling they they wouldn’t be a customer for much longer.
Additionally would Apple really want to support macOS on both architectures?
Last edited:
Apple supports two architectures right now and has been since the iPad .
With marzipan putting apps on different architectures is planned .
And keep in mind the majority of apple products are already ARM based.
I just can’t imagine having any new Mac being introduced this year only to have them unsupported after two or three years due to ARM only processors.
If apple were to continue support for intel based Macs past 2020 Apple would have to continue supporting intel.
Do you think that the delay (almost 2 years since last one?) of new iMacs is somehow related to the ARM cpu?
All depends on how long they're taking to switch over. It also makes a certain amount of sense to make the Mac Pro be the last Intel Mac just so the high-end people with legacy hardware needs have an option. If Apple is switching over, then I'm sure they've factored in ARM design decisions into the new Mac Pro design. It's fundamentally not that huge a difference in terms of physical attribute, especially since power efficiency and relatively cool operation of Apple's designed chips theoretically don't impact the Mac Pro.
The G5 to Intel resulted in dumping all those extra fans and shroud, but if you didn't know better there was little from the outside to distinguish a Mac Pro from a PowerMac.
So If this is true I wonder will it be on the new MP?
https://www.macrumors.com/2019/02/21/apple-custom-arm-based-chips-2020/
https://www.macrumors.com/2019/02/21/apple-custom-arm-based-chips-2020/
We don’t know that the three years have been spent designing the thing. Al we know for sure is they’ve invested three years in inviting their favourite bloggers to roundtables.
But your point kind of makes sense, if they’re going to make the switch to arm, that would explain the huge delay.
How crazy would be the notion of Apple coming up with their own GPUs based on clustered ARM chips?
SoftBank’s investment fund (owner of ARM Holdings) dumps entire $3.6 billion stake in Nvidia
https://www.cnbc.com/2019/02/06/softbank-vision-fund-sells-nvidia-stake.html
SoftBank’s investment fund (owner of ARM Holdings) dumps entire $3.6 billion stake in Nvidia
https://www.cnbc.com/2019/02/06/softbank-vision-fund-sells-nvidia-stake.html
Last edited:
While I fully believe Apple intends to transition macOS from x86 to ARM instruction sets, I expect the Mac Pro to be the last model transitioned over and only once core productivity apps have been fully integrated. So it will be x86 for years to come.
People don't install Boot Camp because they want to boot to Windows, they install Boot Camp because they want to run Windows applications. Windows 10 for ARM really only accomplishes the first one. That it "can still boot to Windows" is not really a feature at all.
Windows on ARM is a way for Microsoft to compete with the thin client/chromebook competition, not a path towards an Intel-less desktop workstation experience. It will get you a familiar Windows environment that can run a browser and a very minimal suite of applications.
Heck, until just three months ago it couldn't even run 64 bit ARM applications. Support for x86 apps (32 bit only) is nascent and comes with a performance penalty.
An ARM-based Mac Pro would be the effective end of Boot Camp. Linux dual boot would still be "ok," but likely compromised if you care about accelerated GPU.
Moreover, it's optimistic to imagine that an ARM platform version of VMware Fusion or Parallels would be coming any time soon (or at all).
macOS on ARM makes total sense to me for MacBooks and maybe even the MacBook Air. I think (hope) it's a non-starter for MacBook Pros and (i)Mac Pros though.
I could see the Mac Pro leading the charge, but it'd have to be an announcement like "hey look at this machine, it's X% faster than this Xeon and/or X% more power efficient. There's going to be new versions of Logic and Final Cut ready for this day-one when it ships, and here's (guy from creative application) who's already ported their software talking about how easy it is to do and the benefits they've seen to their workflows." (And some line about how legacy stuff won't be much slower.) In other words, they'd have to have done the software work themselves and have some sort of example for others, as well as the x months of lead time.
Does the workstation market really care about power efficiency in any machine that isn't running off a battery?