Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
With two CPUs, although you get twice the aggregate memory bandwidth, you introduce additional memory latency. Also, on a single CPU the cores will share the top level cache, and thus with two CPUs maintaining coherency between those caches has an additional cost. Whether the dual socket system is faster or slower than an equivalent (i.e. same number of speed of cores) single socket system will depend on memory access patterns of the application.
In theory you are right.

In practice, nothing that you said matters.

You need to write an application that deliberately sabotages the cache hierarchy to see any NUMA effects. I've done it, and have a synthetic benchmark that shows a 20% hit for NUMA. Intel has done their homework on multi-socket cache and NUMA issues.

None of the application benchmarks that I've run see any significant difference.

Dual 6 core, 140W CPU with 3.6 GHz Base clock, 3.8 GHz would be faster than single 2.8 GHz 12 core CPU, with 140W TDP. Yes that is true. The problem is that it would consume twice the amount of power.
Yes, but unless you're trying to put dual sockets into a system that's severely constrained on both thermal and power issues - that can be a non-issue.

(Who sells systems that are severely constrained on thermal and power issues? ;) )

The other part of that is that idle power consumption is a small fraction of the TDP - so if all you're using are the extra PCIe lanes from the second socket the power difference could be minor. More likely that one more disk drive is using more power than the second socket.
 
  • Like
Reactions: tralfaz
Thank you for admitting as much. Case closed.
I have never claimed your post was invalid. Context of posts was invalid, and I was writing that ;)
Yes, but unless you're trying to put dual sockets into a system that's severely constrained on both thermal and power issues - that can be a non-issue.

(Who sells systems that are severely constrained on thermal and power issues? ;) )

The other part of that is that idle power consumption is a small fraction of the TDP - so if all you're using are the extra PCIe lanes from the second socket the power difference could be minor. More likely that one more disk drive is using more power than the second socket.
I am not questioning that. But ask yourself, in the context of Mac Pro: what Apple focused with Mac Pro 6.1? Highest possible performance at all costs, or highest possible performance in smallest possible footprint AND thermal envelope?
 
In theory you are right.

In practice, nothing that you said matters.

You need to write an application that deliberately sabotages the cache hierarchy to see any NUMA effects. I've done it, and have a synthetic benchmark that shows a 20% hit for NUMA. Intel has done their homework on multi-socket cache and NUMA issues.

None of the application benchmarks that I've run see any significant difference.

I agree that this is not a problem for most people, but I have most certainly hit NUMA issues many times for very complicated numerical codes I have written, and it is most certainly an issue that comes up in the numerical analysis community when the number of sockets is at least four.
 
I agree that this is not a problem for most people, but I have most certainly hit NUMA issues many times for very complicated numerical codes I have written, and it is most certainly an issue that comes up in the numerical analysis community when the number of sockets is at least four.
... I thought that we were discussing dual socket systems.
 
right now things point towards Apple waiting for Sierra and Kaby lake before any hardware updates arrive. after all this waiting, a few more months won't matter anymore. the ines who were about to jump ship have already done so.
 
I have been thinking about the reasons why Baffin is already fully supported by macOS Sierra beta builds, but there is no absolutely full support on Polaris architecture.

IMO it is simple. In September there will be iPhone/iOS event with release of final macOS Sierra. Possibly at the same event there will be premiere of new Macbook Pro, which should have the Baffin GPU inside it.

And then in October second event where they can introduce macOS Sierra 10.12.1, with completely new hardware.

If they will introduce Mac Pro 7.1 with dual Polaris 10 as the highest end offering I will be extremely disappointed.
 
TB, KBL-W will be here in 2018 at best, I believe, maybe even later with all the delays. SKL-W will come next year before that.
I'm not even sure Apple will wait for KBL for the rMBP, I hope they do but rumors of it being released some time soon makes me wonder if SKL will be in there instead. KBL would be awesome, with 200 series PCH with support for TB3, Optane and all the juicy stuff like HDR...

koyoot, either Apple already has KBL samples and has been working on the new rMBP with it (I hope so) or we're stuck with Skylake for some time. If it's SKL with 200 PCH, not too bad. But the iGPU lacks when it comes to the new features. And probably only the high end 15" will get Baffin.
I don't think they'll debut the rMBP in September though, they'll wait till the end of the year or so, hopefully.
the same for nMP, so that Vega is available for DX700.
[doublepost=1472133771][/doublepost]Yep, USB 3.1 Gen 2:
macos-sierra-usb-3-1-gen-2-text-string-mac-hardware2-635x510.png
 
It's grasping at straws... ;) kli7yyyyyyyyyyyyyy

(Last comment was from the coon that walked across the keyboard - maybe the Flying Spaghetti Monster sent a message through him.)

It's just as likely that the copy writer used parallel phrasing....

Oddly the Macbook didn't get the same phrasing though. What the heck.
 
Oddly the Macbook didn't get the same phrasing though. What the heck.
I thought that might be in part because the current MacBook is completely unrelated to earlier systems named MacBook.

Although that logic also applies to the current Mac Pro. :eek:

OMG - that mean that the MacBook is being discontinued.
 
I noticed the MacBook and this is the worst example, since there is already a MacBook 2016, which must also support Metal.
For all others it must be a matter of just in case a new one comes along.
Let's hope this is the case of the Mac Pro.
 
From Rumorland (believe it or not isn't matter to discuss, its rumor-based speculation):

Apple to introduce new Macbooks Pros along the new iPhone 7 same event, new desktops including new 21"4k TB3RD later in October/November including new Mac Pro, updated iMacs and All New form Factor Mac mini.
 
Last edited:
From Rumorland (believe it or not isn't matter to discuss, its rumor-based speculation):

Apple to introduce new Macbooks Pros along the new iPhone 7 same event, new desktops including new 21"4k TB3RD later in October/November including new Mac Pro, updated iMacs and All New form Factor Mac mini.

Does Apple really want the seemingly mediocre iPhone 7 to get buried and lost in a huge, juicy Macgasm?
On the other hand, maybe they do.
 
I wouldn't count on having rMBP being presented at the same event as iPhone 7.
But who knows?!
rMBP should be quite nice, if rumors are true regarding the OLED strip, embedded touch and all.
I was hoping for KBL to be there but if it comes in September chances are slim.
Also, upgraded resolution would be nice as well.
All TB3 and USB-C seems probable, nice.
 
Does Apple really want the seemingly mediocre iPhone 7 to get buried and lost in a huge, juicy Macgasm?
On the other hand, maybe they do.

I don't see why not, Apple isn't the Commonwealth where traditions are more important than the laws.

Apple needs to reinforce the APPLE Ecosystem as iPhone selling point, which better than to introduce all the typical iPhone-centric ecosystem (iPhone7,Watch 2, MacBook Pros), the New MacBooks dont need a multi-hour introduction 40 minutes is enough to introduce TB3(15), Touch ID(10) and the "FinderBar" (15).
Its the only way Apple has to push up the mediocre iPhone 7 is with an Rich ecosystem.

Further, The Macbook Pro Market urges for new a product much more than iPhone market.
 
From Rumorland (believe it or not isn't matter to discuss, its rumor-based speculation):

Apple to introduce new Macbooks Pros along the new iPhone 7 same event, new desktops including new 21"4k TB3RD later in October/November including new Mac Pro, updated iMacs and All New form Factor Mac mini.

from a rumour source I have seen, directly relating to this 4k 21" idea - there is 1 and only 1 display coming from Apple, it is 5K 27" eGPU TB3 based, with no ability to communicate over Displayport, and has been ready to go to manufacturing for months, but has been held up by the delays in Skylake-based Macbook Pros, which it's launching with. There is no 4K / 21" or Displayport-drivable at all display in the pipe.

It's going to be an interesting test of the credibility of various sources in the next couple of months.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.