MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)

[goMac]

Considering a 7nm intel anything is probably 2-3 decades away. I would certainly hope it would blow a current A12x out of the water.

However, there are 7 nm x86 CPUs available from other companies... Built in the exact same factories Apple is using, using the same techniques as A12x.

(Intel's upcoming 10 nm CPUs, due in about a year or two, are also supposed to be pretty close to 7 nm, due to Apple and Intel using different measuring techniques.)

 

[Kpjoslee]

A12x does a much better job of showing 7 nm vs 14 nm instead of ARM vs Intel.

I still think it will happen, but people are taking the wrong message from the A12x benchmarks. A 7 nm Intel i7 CPU would likely blow A12x out of the water.

I am fairly certain the next Mac Pro will not be ARM. Eventually the Mac Pro might be ARM, but it's going to be a while. I think we'll been on both ARM and x86 Macs at the same time for a while.

Yeah, that is also true, since Intel is currently stuck at 14nm. We will probably see the better picture when Intel 10nm shows up on holiday next year. But it showed Apple's variation of ARM architecture has lot higher performance ceiling than previously thought. It will be difficult to expect another large scale increase in IPC, but if they manage to scale up to higher power target and achieve decent increase in clock speed, we will definitely see the switch on Macbook line first.

Well, I wouldn't worry too much about it right now since it is 2 years away at earliest.

 

[goMac]

But it showed Apple's variation of ARM architecture has lot higher performance ceiling than previously thought.

Did it? It shows that a chip on a smaller process is faster. Don't get me wrong, I'm not trying to sell it short. But A12X definitely has not proved they have a workstation class chip. It's still pretty firmly in the realm of mobile performance, once you get past Intel being stuck on the same process.

 

[Slash-2CPU]

A12X has 10 Billion transistors and runs at a split 2.5GHz on 4 cores and 1.6GHz on 4 cores. 83mm^2 die size. 7nm nominal process.

18-core Xeon W 2191B in iMac Pro runs at 2.3-4.3GHz across all 18 cores. 484mm^2 die size. 14nm nominal process.

MUCH bigger piece of silicon.

Don't even get me started on how big a lie TSMC's 7nm process is and how it's closer to 11.5nm process. It's a half-node 7nm where a lot of the rest of the process is bigger than Intel's 14nm++. 14nm doesn't mean that the ENTIRE chip is wired with 14nm wire. Not even close. It's as much a marketing label as Intel's 9th gen chips. 9th gen? no. More like 8.005th gen

 

[Kpjoslee]

Did it? It shows that a chip on a smaller process is faster. Don't get me wrong, I'm not trying to sell it short. But A12X definitely has not proved they have a workstation class chip. It's still pretty firmly in the realm of mobile performance, once you get past Intel being stuck on the same process.

Per core performance. Chip on a smaller node will enable more performance on same die area but is not the only reason for increase in per core performance. Anandtech's test showed that A12, at least on SPEC2006 performance, is already on the level of modestly clocked Skylake.

https://www.anandtech.com/show/13392/the-iphone-xs-xs-max-review-unveiling-the-silicon-secrets/4

Of course that doesn't tell the whole story but at least it shows that the potential is there. If Apple's ARM architecture is yet to reach the IPC wall, then we could see it becoming really competitive against latest x86 architecture. We will see the first glimpse when first full-fledged photoshop launches on iOS next year, then we will be able to find out how it does in real world application.

 

[zephonic]

Did it? It shows that a chip on a smaller process is faster. Don't get me wrong, I'm not trying to sell it short. But A12X definitely has not proved they have a workstation class chip. It's still pretty firmly in the realm of mobile performance, once you get past Intel being stuck on the same process.

I guess what people find exciting is the fact that the A12X delivers that performance at 5W, and how that would scale up with proper cooling and no power restrictions. I think MacPro-type Xeons have a 130W TDP or so.

 

[koyoot]

I guess what people find exciting is the fact that the A12X delivers that performance at 5W, and how that would scale up with proper cooling and no power restrictions. I think MacPro-type Xeons have a 130W TDP or so.

I would like to ask: how ARM behaves when it has 100 amps delivered, instead of 5 amps? How it will behave with 150 amps?

It is different physical design, which also will affect architecture performance. Something powerful and efficient in low-power envelope, may not be equally amazing in high-power envelope.

 

[ManuelGomes]

I'll be ordering 2 iPad Pro later today. Guess I'll skip mMP with these monsters now
They're amazing piece of tech, but that bendgate issue again makes me wonder if I'll regret.
Since mMP should still be a year out, plenty of time to save money (keep saving that is...) for it.
I'm just afraid that, after all this time waiting, when it comes out it will be nothing to write home about.
Anyway, who waited all these years...
[doublepost=1543590129][/doublepost]Unrelated, but this being page 404 in this thread does it mean that it's not found?!
Sorry, couldn't resist it
back to the topic.

 

[filmak]

I'll be ordering 2 iPad Pro later today. Guess I'll skip mMP with these monsters now
They're amazing piece of tech, but that bendgate issue again makes me wonder if I'll regret.
Since mMP should still be a year out, plenty of time to save money (keep saving that is...) for it.
I'm just afraid that, after all this time waiting, when it comes out it will be nothing to write home about.
Anyway, who waited all these years...
[doublepost=1543590129][/doublepost]Unrelated, but this being page 404 in this thread does it mean that it's not found?!
Sorry, couldn't resist it
back to the topic.

Hello Manuel!
Ensure that you will purchase a nice case for your new iPad for better protection. Bendgate is present for many years as the devices ate becoming thinner and thinner.
Now, I hope that your savings will be enough for the future mMP, but Apple has its way with prices and I think that we will need something more than our savings once more.

[doublepost=1543590630][/doublepost]Have you bought the 11" or the 12.9" one?
I have bought the 11".

 

[Slash-2CPU]

I would like to ask: how ARM behaves when it has 100 amps delivered, instead of 5 amps? How it will behave with 150 amps?

It is different physical design, which also will affect architecture performance. Something powerful and efficient in low-power envelope, may not be equally amazing in high-power envelope.

Scaling doesn’t work that easily. That same logic says DJI builds really nice drones, so they should be able to build and design the next combat helicopter. Just build the drone bigger.

If it were that easy, there would be more than only Intel and AMD in the game right now. You’d have Apple, Qualcomm (Snapdragon), and Samsung(Exynos) all making 75-150W CPU’s for desktop or server use.

 

[goMac]

Of course that doesn't tell the whole story but at least it shows that the potential is there. If Apple's ARM architecture is yet to reach the IPC wall, then we could see it becoming really competitive against latest x86 architecture. We will see the first glimpse when first full-fledged photoshop launches on iOS next year, then we will be able to find out how it does in real world application.

So just to be clear, I'm not saying ARM is bad. There is potential for it to perform well.

What I am saying is I don't see a clear advantage over x86, especially x86 on the same process. And Apple doesn't have a workstation version at the ready. That's very different than the MacBook where they could probably drop in the A12X as it is today.

I'm not sure A12x proves much about the IPC wall. With the new process, they increased the frequency, cores, and cache. I'm sure there were some IPC gains, but it's hard to make a case that they're really coming up with something amazing for workstations.

 

[koyoot]

Scaling doesn’t work that easily. That same logic says DJI builds really nice drones, so they should be able to build and design the next combat helicopter. Just build the drone bigger.

If it were that easy, there would be more than only Intel and AMD in the game right now. You’d have Apple, Qualcomm (Snapdragon), and Samsung(Exynos) all making 75-150W CPU’s for desktop or server use.

This is the VERY reason why ARM is a joke on servers compared to x86.

 

[deconstruct60]

This is the VERY reason why ARM is a joke on servers compared to x86.

It is not a joke. It just isn't 'miraculously' better power utilization. Like the basic architecture implementation has some magic pixie dust in efficiency.

Cavium Thunder X2 180W > $1,000 price.
https://www.anandtech.com/show/12694/assessing-cavium-thunderx2-arm-server-reality/2

It isn't like getting high end compute server for 30W and $300 price tag.


For virtualized web servers and Hadoop nodes ARM solutions can be a player. Amazon is rolling out some AWS nodes with a lower end

"... Today, Hamilton said, "I’ve seen the potential for Arm-based server processors for more than a decade, but it takes time for all the right ingredients to come together." ...
... Semiconductor industry watcher David Schor shared SciMark and C-Raybenchmarks for the 16-core Graviton. In the SciMark testing, the AWS system-on-chip was twice as fast as a Raspberry Pi 3 Model B+ on Linux 4.14. ... "
https://www.theregister.co.uk/2018/11/27/amazon_aws_graviton_specs/

The solutions being applied to make them a better (or worse ) match. For a single user, workstation average workloads there isn't an obvious big win once get out of single core drag racing corner cases (with perhaps an overclock Apple A series) or uber extremely embarrassing parallel scatter-gather like Hadoop.

 

[Slash-2CPU]

ARM servers have a great niche for storage server nodes. That's why the Thunderx2 has 56 PCIe lanes that can be bifurcated down to x1. Hello, video streaming services. You could build a farm of these on 40GB Ethernet as data storage feeding a rack of Epyc or Xeon boxes.

Compared to other low per-core performance multi-core massively-parallel Xeon and Epyc it looks acceptable. It does however lack any higher-end SKU's that compete on the level of Epyc or monster Xeons like Broadwell-EX. Keep in mind also that Epyc was launched in early 2017 and Broadwell-EX is also up for a refresh soon. Comparing brand new hardware to 2-year-old stuff will always look good.

Oh, and no multi-socket capability either.

We're still several years away at a minimum from ARM being able to compete head-to-head with readily available hardware on single-thread or lightly-threaded typical end user scenarios against an i5-7400 or even an i7-7700K like you'll find in a top-level iMac

Not saying ARM-powered Macs will never happen, just very, very unlikely before 2025. Apple isn't THAT good at hiding secrets. They'd need a truly massive CPU development program. Someone would leak. Until i can buy one from Apple or amazon.com, it's all vaporware.

 

[AidenShaw]

This is the VERY reason why ARM is a joke on servers compared to x86.

And, for the record, x86 servers are dead. They're all x64 now.

 

[BlueTide]

A12x does a much better job of showing 7 nm vs 14 nm instead of ARM vs Intel.

I still think it will happen, but people are taking the wrong message from the A12x benchmarks. A 7 nm Intel i7 CPU would likely blow A12x out of the water.

I am fairly certain the next Mac Pro will not be ARM. Eventually the Mac Pro might be ARM, but it's going to be a while. I think we'll been on both ARM and x86 Macs at the same time for a while.

It requires careful management from Apple in any case. I hope they pull it off and I'd personally be very happy to have some action on the processor front - for academic interests, mostly:

• If you switch to ARM, and start with the low end, you just osbourned your high end. Who in their right mind would pay a lot for Mac Pro if they know it's replaced by ARM in few years. Unless, of course, you have acute business reason, but then... you should've switched to PC already.
• If you start with the high end, you need to somehow pull the major / relevant app along so that people can actually get stuff done with them.

x64 / x86 also gets a bit too much flack, IMHO. I'm not saying the ISA is great, but in practice it's been a relatively minor interpretation layer anyway to abstract how the processors work internally.

 

[mattspace]

• If you start with the high end, you need to somehow pull the major / relevant app along so that people can actually get stuff done with them.

maintaining the slimmest hope that "completely rethinking the mac pro" means a design that is the most slotty of slotboxes, and Apple makes a pci card similar to one that takes 4 m2 ssds, but instad of ssds, each is an ARM processing unit, something like an iPhone chipset, minus the parts that are needed to run a phone. That way you get a xeon workstation with minimum disruption, plus arm processing (if it's all its cracked up to be for pro tasks).

 

[goMac]

maintaining the slimmest hope that "completely rethinking the mac pro" means a design that is the most slotty of slotboxes, and Apple makes a pci card similar to one that takes 4 m2 ssds, but instad of ssds, each is an ARM processing unit, something like an iPhone chipset, minus the parts that are needed to run a phone. That way you get a xeon workstation with minimum disruption, plus arm processing (if it's all its cracked up to be for pro tasks).

But why?

There has been some movement around T.2 taking some load off the CPU and doing co-processing. But I'm not sure why you'd farm a bunch of work out to little ARM CPUs when you have a big honkin screaming fast Xeon CPU right there. Even if each little ARM board was as fast as a few of the Xeon cores, a single Xeon CPUs these days still has 20-30 cores alone.

Such a setup would make sense on a low end Intel box where you needed ARM CPUs to beef up a slow Intel CPU. But on a Mac Pro why would you want to bypass the speed of the Xeon for much slower and smaller ARM chips?

 

[mattspace]

But why?

I guess for the same reasons we farm stuff out to GPUs, or to allow Apple to spec a lower end Xeon, while still being able to keep the overall performance for tasks that are ARM-suitable, or as just one use case for a Flexible Mac Pro - ie a machine who's core philosophy is to do everything in a user-removable, standards-based units - ARM-blade monster as one option, multi-gpu rig etc as another.

 

[Biped]

Hmm Interesting.

MS just eclipsed Apple in market cap.

Satya Nadella is even more SJW friendly than the pasty privileged duo of Timmy and Jony. Excellent candidate to be the next cult level figure to sheepishly follow, Although Jobs will still be our first shepard, RIP Steve.

Welp gentlemen, gentlelaides, and gentleattackhelicopters, goodbye macpro, hello surfacepro ! An exciting new world of 'ecosystem' and 'services' catchphrases, overpriced and thermally throttled hardware, and software that listens on you except when you want it to ! Fresh but not unfamiliar.

 

[fuchsdh]

Scaling doesn’t work that easily. That same logic says DJI builds really nice drones, so they should be able to build and design the next combat helicopter. Just build the drone bigger.

If it were that easy, there would be more than only Intel and AMD in the game right now. You’d have Apple, Qualcomm (Snapdragon), and Samsung(Exynos) all making 75-150W CPU’s for desktop or server use.

The money isn't in making desktop CPUs when you're competing against two established figures.

And your drone comparison is simultaneously good and bad, because a small drone does not scale up to an attack helicopter—but larger drones are going to replace most manned vehicles at some point, especially in the military. So too does ARM and the like have the chance to totally usurp the old beast Intel.

 

[ibgb]

...

• If you switch to ARM, and start with the low end, you just osbourned your high end. Who in their right mind would pay a lot for Mac Pro if they know it's replaced by ARM in few years. Unless, of course, you have acute business reason, but then... you should've switched to PC already.
• 
  
• ....

I don't follow the logic.

The low end is already on ARM. You do know this? Why does that osborne anything? I think we all know that whatever they do with a Mac Pro will be replaced in a few years. That is what this whole thread is about: What will replace the current Mac Pro. ( Or maybe, WHEN will they replace the current Mac Pro.) If people need a Mac Pro, they will buy one and hope Apple replaces it with a better version in a few years, at least LESS THAN 5 years.

 

[vailr]

maintaining the slimmest hope that "completely rethinking the mac pro" means a design that is the most slotty of slotboxes, and Apple makes a pci card similar to one that takes 4 m2 ssds, but instad of ssds, each is an ARM processing unit, something like an iPhone chipset, minus the parts that are needed to run a phone. That way you get a xeon workstation with minimum disruption, plus arm processing (if it's all its cracked up to be for pro tasks).

What about the AMD "Rome" CPU approach?
Intel could fabricate a custom CPU having multiple ARM "chiplets" surrounding a central Xeon chip.

 

[koyoot]

What about the AMD "Rome" CPU approach?
Intel could fabricate a custom CPU having multiple ARM "chiplets" surrounding a central Xeon chip.

If anything Intel will do similar to chiplets, is by EMIB, and usage of GPUs or FPGAs.

 

[filmak]

What about the AMD "Rome" CPU approach?
Intel could fabricate a custom CPU having multiple ARM "chiplets" surrounding a central Xeon chip.

And the benefit for end users would be?