There are faster CPUs available.
There is faster memory available.
There are faster GPUs available.
There is faster storage available.
How far out of date should it get to warrant an update?
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.
nMP 2015: where is it?
- Thread starter ManuelGomes
- Start date
- Sort by reaction score
There are faster CPUs available.
There is faster memory available.
There are faster GPUs available.
There is faster storage available.
How far out of date should it get to warrant an update?
Which AMD gpus can fit in 125W thermal envelope AND be faster than Tahiti GPU, that are in Mac Pros?
Apple prefers to update whole computer, not just parts of it.
If they're intent on sticking with AMD they could easily downclock Hawaii hard enough to fit into the power restrictions.
Downclocking to 400-500 MHz? It would make it less powerful than the Tahiti chip.
2816 cores*2/clock rate = 2816 Gflops. Tahiti is ~3500 Gflops.
That's best case scenario but there are plenty of examples of spec bumps.
Hell, the new SSD in the air is way faster than the one in the nMP, would it kill them to start using those?
Another option would be to use a design that didn't demand such crippling tradeoffs but i suppose that ship has sailed.
I already put one in my nMP. Instant 40% speed boost.
If they're not all getting them in 2 months it will be rather sad.
About 3% faster on average (slower in some benchmarks)
About 1% faster in real-world
About 20% faster but running hotter (= down clock)
About 20%-40% faster (depending on which current nMP SSD you compare to)
While it's still possible Apple will update the nMP with this stuff, the real-world performance improvements will be negligible for most people.
Given the emphasis on GPUs in this Mac, they may have shifted away from refreshing based on Intel's roadmap to refreshing based on AMD's roadmap and thus are waiting for AMD's next-gen GPU architecture... which has been delayed due to a shortage of 20nm silicon production. (Ironically, it's likely due to Apple having bought up all available supply for the A8).
when there's more debt in the world than money, of course we're going to see it in action.. greece argentina are localized bursts of the bubble.
all the countries are in debt.. most, by far, people are in debt.. what's so special about 'national' debt? just another version of debt that everyone is mucking through.. it's a stupid system.
(or, a genius system if you designed/control it and are into that sort of thing)
----------
maybe maybe_not.. but it appears that's how they're going to do it even if you don't think it's best for the company.
how many mac pros have been sold?
(just out of curiosity.. has that stat been released anywhere?)
i'm pretty sure this forum is a horrible sampling of the typical mac pro user/buyer.. most people don't think or care about this stuff or certainly not to the degree the forum hypes it up to..
most people are more like "ok.. i want a mac.. and i want a badass computer.. i'll buy the mac pro" ..and they buy it and use it and it's great for them.. they spent a lot of money but the computer is actually badass and will still be in 5 years..
all the people that hyperfocus on computers dayinanddayout and complain about them etc are the people that tend to congregate in places like this.
point being- if you talk about apple customers, you gotta realize youre not the typical buyer or user.
I'm not sure your assumptions are generally true among Mac customers. Of the Mac customers I know (at least those that might by anything more than a Macbook Air), while I'm certainly more tech-inclined than the majority of them, most of them could what intel generation of CPU they are on. I know some who will just go buy that bad ass Mac Pro even if rumors are its getting updated tomorrow or if it hasn't been updated in 2 years. But I rarely run into those people. At least not for Mac Pro, maybe for a MacBook Air. But usually people that need a Mac Pro (even a MacBook Pro), tend look up a thing or two about what's going on with those computers before purchasing one. Even my own parents, who aren't tech savy at all, knew to ask me before they got a Air (and I saved them from buying only about 1-2 weeks before a refresh a couple summers ago).
That's been my experience anyway.
The idle rich buy things for reasons like "I want a bad ass one".
Professionals spending $3-10K of their own money on a tool to make money tend to do some research first. It is an investment, and just like investing in a stock, you wouldn't want to invest in something that isn't going to make you money.
Professionals spending $3-10K of their own money on a tool to make money tend to do some research first. It is an investment, and just like investing in a stock, you wouldn't want to invest in something that isn't going to make you money.
What is your point exactly here guys? I know plenty of leading London based graphics teams with the new nMP - They're happy with their choice. It was their hardware of choice and I don't know of one that regrets it.
Stereotyping is right down there with sarcasm - I'm not rich and I bought one, what does that say?
It's easy to hate when you don't even own a nMP.
Stereotyping is right down there with sarcasm - I'm not rich and I bought one, what does that say?
It's easy to hate when you don't even own a nMP.
That probably depends on which guy youre talking to here.
Im certainly not bashing anyone for choosing something that fits them. Certainly it fits some people well and good for them. And while I wouldnt say I hate this machine, I will say it makes exactly zero sense for me. Owning one first isnt going to change that. What I do gets by just fine on any random $50 graphics card because all I need to do is drive a simple display (4K/5K video means nothing to me). So why would I need 2 graphics cards that probably add $1000 to the price tag?
And weve been through CPU gains that Apple is passing up. They could probably be putting 2x 10-core Haswell chips in the mid level Mac Pro. Instead, because of the 2x GPUs and tiny form factor they are stuck with 1 CPU socket regardless, and because of those GPUs they are still waiting to do an update while Intel has moved on. Meaning Apples stuck with 1x 8 core in the mid level Mac Pro.
What I do is CPU limited and to a lesser extent IO/storage space limited. This Mac Pro (not to mention the long delay between real updates from 2010-2014) has left me behind and owning one first wouldnt change that.
I disagree. Pros make do with what they have or what is available. Just because you make money with your computer, doesn't necessarily mean you can justify the cost. Want 10% increase with your old equipment? Use your old machine 10% more. Take 10% less breaks.
Did well this year and going to have to pay Uncle Sam 10k anyways...might was well buy a nMP.
I've never seen what type of CPU a person has impact their bottom line. If that was the case, every tech house in the land would constantly be begging for constant updates.
This can't be emphasized too much.
The question shouldn't be "how much GPU X has to be downclocked to fit the thermal envelope of the trash can enclosure?", it should be "how on earth did Apple execs allow an 'innovate, my ass' enclosure that is in essence non-upgradeable?".
Wherever you stand on your opinion of the MP6,1 tradeoffs - most should agree that if those tradeoffs make it infeasible to upgrade, then mistakes were made.
I'd love an upgradable nMP as much as anyone, but Apple has made all their recent computers less and less upgradable. And I'm pretty confident it was a conscious decision on their part, not a mistake. Apples philosophy about computers as appliances has a long history, it just took a long time to bleed over to the Mac Pro. They are clearly willing to sacrifice sales to folks who insist on upgradeability as a requirement. It seems if you want Apple hardware, unfortunately this is the compromise you need to be willing to make.
Barefeats found same numbers same so I can get them out there now.
As more proof that the machine wasn't designed with the future in mind, consider these numbers.
SM951 in cMP in a PCIE 4 lane adapter 1,500 read/write speeds
SM951 in nMP 1,300-1,400 speeds
SM951 in TB2 enclosure to nMP 1200-1300 speeds
If Apple had run PCIE 3.0 to that little slot it would be hitting 2,100. Instead the slot seems to not even have full time PCIE 2.0 x4.
But if you want the best speeds with a 2015 Apple SSD, put it in a 2009 Mac Pro. Pretty sad.
As more proof that the machine wasn't designed with the future in mind, consider these numbers.
SM951 in cMP in a PCIE 4 lane adapter 1,500 read/write speeds
SM951 in nMP 1,300-1,400 speeds
SM951 in TB2 enclosure to nMP 1200-1300 speeds
If Apple had run PCIE 3.0 to that little slot it would be hitting 2,100. Instead the slot seems to not even have full time PCIE 2.0 x4.
But if you want the best speeds with a 2015 Apple SSD, put it in a 2009 Mac Pro. Pretty sad.
A "conscious decision" can still be a mistake.
It was a "conscious decision" for Coca Cola to have introduced "New Coke".
ps: in this context I don't mean "upgradeable" to mean "user upgradeable" - the issue is that it seems that the MP6,1 is so crippled on power and thermals that Apple is unable to easily make an updated system - even though newer CPUs and GPUs are here.
Last edited:
really? you don't think you guys have emphasized it enough by now? it's been what, two years?
"Milky Way 2" - the World's Fastest Supercomputer
Built for the National Supercomputing Center in Guangzhou China, the "Milky Way 2" system is powered by 32,000 of the upcoming 12-core Intel Xeon processors E5-2600 v2 based on Ivy Bridge architecture, and 48,000 Intel Xeon Phi coprocessors, with a total system power of 17.8 MW. Not only is it the fastest, but also one of the most power-efficient systems on the Top500 list. The system uses "neo-heterogeneous architecture," whereby the hardware architecture has multiple classes of compute capabilities that are accessed by a common programming model, streamlining development and optimization processes an advantage not possible when using a combination of CPUs and GPU accelerators.
Built for the National Supercomputing Center in Guangzhou China, the "Milky Way 2" system is powered by 32,000 of the upcoming 12-core Intel Xeon processors E5-2600 v2 based on Ivy Bridge architecture, and 48,000 Intel Xeon Phi coprocessors, with a total system power of 17.8 MW. Not only is it the fastest, but also one of the most power-efficient systems on the Top500 list. The system uses "neo-heterogeneous architecture," whereby the hardware architecture has multiple classes of compute capabilities that are accessed by a common programming model, streamlining development and optimization processes an advantage not possible when using a combination of CPUs and GPU accelerators.
A "conscious decision" can still be a mistake.
It was a "conscious decision" for Coca Cola to have introduced "New Coke".
ps: in this context I don't mean "upgradeable" to mean "user upgradeable" - the issue is that it seems that the MP6,1 is so crippled on power and thermals that Apple is unable to easily make an updated system - even though newer CPUs and GPUs are here.
I see, yeah that makes sense. Maybe one day Apple will go back to the "classic" formula, but given the Mac Pro is millions of parsecs from the center of the Apple product universe, I doubt it.
In terms of GPUs... All the recent GPUs (Tonga) keep trying to extract more performance from the same 28nm process they've been on forever now and are running hotter than the Tahiti GPUs that the nMP thermals were designed for. Everyone expected next-gen GPUs at 20nm would be shipping months ago, but TSMC says 20nm yeilds of large die chips aren't good and rumors are that GPUs will skip 20nm for 16nm which means even more delays. Although I wonder... with Apple selling a plane load of iPhones daily, perhaps the A8 has consumed all available 20nm FAB production... And might for several months yet. I haven't looked into it, but this generation of GPUs (28nm) has got to be one of the longest in history.
The more concerning thing is... If GPU yields at 20nm are no good, what makes anyone think they will be much better at 16nm? Perhaps the days of monsterous GPU cores are gone and more modular designs combining several smaller GPU dies is the way things will need to go in the future?
Last edited: