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.
From Facebook's "We Want a New Mac Pro"
- Thread starter 5050
- Start date
- Sort by reaction score
Where is your proof and source for this? I highly doubt these skews have to do with Mac Pro or whatever it will be. Please back up your claim with evidence and sources.. otherwise, don't post it at all.
Expandability and upgradability are not necessarily the same thing. It is far more loss of the "big picture" to try to equate those two.
A Mac Pro to serve files less than a dozen clients over normal household network bandwidth is pretty much overkill. Maybe a Mac Pro serving up for a whole block or whole multi unit apartment building, but a household? Overkill for any new Mac Pro. In fact this facebook rumor description of a system is right up this application.
There is nothing about Thunderbolt that drives or demands that the drives be individually packaged. A TB device could have 8 , 12, 24 drives inside of one box on one TB connect for this type of file streaming of sequential files.
If want to purposely ineffectively leverage TB that is exactly what would do... buy a bunch of individually packaged drives.
There are no MAID (massive arrive of inactive disks ) oriented ( or JBOD for that matter ) TB enclosures yet, but doesn't mean their can't be if there is really demand for it.
But the whole "limited CPU portion that just needs boot drive + box to hold massive and incrementally expanding over time library" actually falls right into what is being described. That is actually not the higher end pro workstation market that needs high bandwidth and throughput.
A Mac Pro as a server just because you need the drive bays is a pretty awful idea. A Mac Mini + a big Thunderbolt multi drive enclosure (or, let's face it, just a NAS) is going to be cheaper and give you better performance.
A Mac Pro really only enters into play when you need to go faster than gigabit ethernet. At that point you start putting things on fiber channel. But in general, choosing a Mac Pro as a file server is pretty questionable.
If I was putting together a file server for an office, I'd want something with more than 3 free bays anyway for RAID 1+0.
If Apple simplifies the box so it is more like a headless iMac.
E5 1600 v2 vs E3 1200 v3 : cheaper. (****)
4 PCI-e slots vs 1-2 PCI-e slots : cheaper.
130W + 300W vs 80W + 150W : ( CPU + PCI-w TDP budget) cheaper (smaller power supply)
42 lbs vs. ~28 lbs : cheaper ( less high grade aluminum bought and and about 33% lower shipping costs. )
**** more so on high end of scale
E5 1660 v1 $ 1080
E3 1275 v2 $ 339
[granted a bit Apples vs. Oranges because 6 core versus 4 cores but even on low end.
E5 1620 v1 $294
E3 1225 v2 $209
trading off a bit of clock for newer micro-architecture 3.6 vs. 3.2
]
Even if Apple does move to E5 it is likely they'll be borrowing higher level of components from other parts of the Mac line. iMac GPUs for TB foundation. iMac disks. But yeah there would probably have to be a contribution of moving down from 40% mark-up on infrastructure to something lower to get some of the reduction. The entry level Mac Pros are juiced on price. They are align more so to gap the iMac prices and align with the rest of the Mac Pro line up than where the costs actually line up.
For the Mac Pro to survive it can't afford to juice prices like that. If demand drops low enough the product will get axed.
He mentioned 9to5mac. And as for the SKUs of shipping Mac Pros, well they're not a secret.Even that is lame because putting a 10GbE socket or two on a NAS box (or on a Mac Mini even ) also solves the problem. The newer Atom server line ups are going to come with 1GbE controllers built in. add 2-3 more 1GbE controller than can do link aggregation and a switch that can do simple load balancing to a single link aggregation and problem solved with anything fancy like fiber channel. All of that is basically off the shelf stuff (including switch).
10GbE isn't mainstream yet but at least the SMB switches are starting to show up now.
http://www.netgear.com/business/products/switches/prosafe-plus-switches/XS708E.aspx
The huge constraint in the "home network" system though is the client bandwidth. Given that is choked that leads to relatively low server overhead. As long as only have a small number of relatively slow clients.... there is nothing to tax a Mac Pro. (the current ones... let alone anything with 2013 parts ).
Hey look someone who cannot read the name of the site nor the preface of the breakdown.
"Please take it as mere speculation, looking for correspondences, reasonable conjectures, etc."
If this rumoured system is essentially a Mac Mini on steroids with dual CPU sockets and a single PCIe slot, then there's no reason the base E3 model can't start just above where the Mac Mini ends... Top of the line Mac Mini is $999... This Mac Pro could start at $1499 easily thanks to a much smaller chassis, smaller single logic board, smaller power supply, and integrated GPU. And if this is possible, then a dual processor variant could start at around $2499.
Take a look here: http://www.everymac.com/systems/apple/mac_pro/index-macpro.html
Listed under "Order No:" Many are BTO/CTO, but a few are MD7.
Also here's a couple of good reads regarding this rumor:
http://www.cpu-world.com/news_2013/2013043002_Intel_Ivy_Bridge-E_extreme_CPUs_to_launch_in_September.html
http://www.macworld.co.uk/mac/news/?newsid=3418637
But of course Adobe is refusing. My impression is that they've decided to increase profits by minimizing their code costs ... which means that OS-unique performance tweaks are a casualty.
(any one of several posts I could have cited).
I think that there's been a 'holy grail' of sorts that's been floating around for awhile in the form of some sort of "Modular Mac" where the customer would tailor how much power they need by how many Macs they buy and network together...FWIW, this was why I mentioned GCD - it looks like one of the likely technology enablers. Another necessary key enabler would be the inter-Mac interconnect ... this might be TB, or 10GbE.
In any case, if one were to get all of these hard Engineering nuts cracked, the vision would be that instead of having a half dozen model#s of Mac Pros with different CPUs (single + dual, plus different GHz), you would pick a single CPU that hits a sweet spot and build a boatload of desktop "MacModules" to sell.
Apple - manufacturing win ... down to a single SKU to build instead of several.
Apple - logistics win ... same single SKU to stock in inventory.
'High End' Consumer - "win" ... because they can buy as few/many Modules as their power requirements require, and they don't have to buy them all in a single lump. Not a pure win! because the cost of buying N units will probably be higher than the old "Best" Mac Pro SKU.
'Low End' Consumer - "win" ... because with the total volume increasing because the High End Consumer is now buying multiple Modules, the cost of a Module sould be less than the $2400 base of a Mac Pro, so a broading of the customer base can occur. Again, not a pure win! because one Module will probably have less content than the old base Mac Pro.
Apple's Mac Desktop Sales division - 'win' ... because their total sales volume will be greater than the sum of the old Mac Pro SKUs, if for no other reason than that they now get to report ~2 MacModule sales for every prior Dual-CPU Mac Pro they used to sell.
Mac OS X Software Development - probably a "push" ... they won't have four different Mac Pros anymore, but now they have to check out the MacModule in single, dual, triple, & larger clusters.
Finally...
Apple Cloud - - TBD (lose?) ... with the consumer retaining horsepower & storage locally, Apple won't be able to make money by renting storage (or bandwidth) to the consumer. Of course, the other implication of this is that for as long as Bandwidth remains slow & expensive in the USA Market, any dreams of Apple forcing a Cloud system down the throads of power users is misguided. But then again, that didn't stop Apple from stonewalling on Blu-Ray, either...
-hh
Yeah, all this is very true. Funny, now I'm struggling to come up with a reason the Mac Pro would be a good server. Even custom server software could still farm data out to a NAS or Thunderbolt box.
We'll see, but I don't think Apple are going to undercut their iMacs with headless cheaper desktop and I think that the $2,499 price will remain as it has for many years. Maybe it'll be better value this time or maybe it will be cheaper, but they know they can sell Macs at that price.
The Power Macs and iMacs were at almost the same prices for many years. Not to mention the Mac Mini has already undercut the iMac.
Of course once you add a monitor in, the Power Mac was not the same price.
Yeah I understand what people want and think the future is. It just baffles me why they think Apple will be the one to solve all the inherent parallelism issues that come with something like this. It would be a grand undertaking for a market that requires a lot of support and well, they don't seem to care about any market other than the consumer ones. It would be a massive shift in mentality. As soon as thunderbolt came out people started on about this, yet an interface is the least of the issues. Nice to dream perhaps, but I don't see modular computing solving the issues which the existing workstation market want from Apple.
It is a good server if have computational workload to put onto it. Streaming video to iPads is bubblegum work from a computational perspective. There is largely no computations there. Even ripping and de-DRM-ing DVD is pretty mainstream stuff.
That would in contrast to 100 active concurrent file serving users which add up to a substantial computational workload. Tons of smaller computations do add up to a big job. Even running ZEVO ( ZFS for OS X http://getgreenbytes.com/solutions/zevo/ ) doing full de-dupe and RAID-Z ... there would be a computational workload.
Okay, I'll be the nerd who checks all the numbers:
Past and current models, loosely marked by update year:
MAC PRO
------
MA356LL/A
----
MA970LL/A
----
MB871LL/A
MB535LL/A
----
MC250LL/A*
MC561LL/A
MC915LL/A
----
MD770LL/A
MD771LL/A
MD772LL/A
MAC MINI
------
MA205LL/A
MA206LL/A*
MA608LL/A
MB138LL/A
MB139LL/A
----
MB463LL/A*
----
MC238LL/A
MC239LL/A
MC408LL/A
----
MC270LL/A
MC438LL/A
----
MC815LL/A
MC816LL/A
MC936LL/A
MD387LL/A
MD388LL/A
MD389LL/A
IMAC
------
MB950LL/A*
MB952LL/A
MB953LL/A
----
MC508LL/A
MC509LL/A
MC510LL/A
MC511LL/A
----
MC309LL/A
MC812LL/A
MC813LL/A
MC814LL/A
----
MC978LL/A
----
MD093LL/A
MD094LL/A
----
MD095LL/A
MD096LL/A
----
ME699LL/A
AIR
------
MB003LL/A
MB543LL/A
MB940LL/A
----
MC233LL/A
MC234LL/A
---
MC505LL/A*
MC503LL/A*
----
MC968LL/A*
MC965LL/A*
----
MD223LL/A*
MD231LL/A*
RETINA MACBOOK
------
MC975LL/A
MC976LL/A
MD831LL/A
MD212LL/A*
----
ME662LL/A
ME664LL/A
ME665LL/A
MACBOOK PRO
------
MC374LL/A
MC375LL/A
MC371LL/A
MC372LL/A
MC373LL/A
MC024LL/A
----
MC700LL/A
MC724LL/A
MC721LL/A
MC723LL/A
MC725LL/A
----
MD313LL/A
MD314LL/A
MD318LL/A
MD322LL/A
MD311LL/A
----
MD101LL/A
MD102LL/A
MD103LL/A
MD104LL/A
LEAKED SKUs:
MD711LL/A — Better — USA
MD712LL/A — Best – USA
MD760LL/A — Better – USA
MD761LL/A — Best – USA
ME177LL/A — Better — USA
ME182LL/A — Best — USA
ME918LL/A — Good — USA
MAC PRO?
-every generation advances the SKU's letter next to the M (MA, MB, MC, MD, ME)
-the number next to that is seemingly random, but only increases from release to release within the same generation prefix (MA, MB, etc). it can't decrease unless you start a new generation.
-new mac pro is not consistent with first set of leaked numbers, because that would require the numerical suffix increment downwards, which it never does.
-new mac pro is very consistent with the second set of leaked numbers. ME is correct for the next generation letter name. and if you look at the group of SKUs for the 2010 models it has almost the exact same pattern:
MC250LL/A*
MC561LL/A
MC915LL/A
With the first being a single CPU, the second being a dual CPU, and the third being a server. The only thing is that "Good" is not a name you would give to a server.
WHAT ELSE COULD IT BE?
-definitely not an iMac for either set of leaked SKUs
-a new mac mini? Mac mini is due for an ME model, and it has the same pattern of 2 closely spaced model numbers plus a higher numbered server model in the same generational release.
-a new macbook air? the first set is consistent with a specs bump for the current model of macbook air. not a new generation though. the second set is not the pattern you usually see for notebook releases, although it could be.
-a new macbook pro? the first set is consistent with a specs bump for the current model of macbookpro but not a new generation.
-retina macbook pro? nothing about any of the numbers is consistent with a release for the retina macbook pro.
SUMMARY:
-the first set of SKUs most likely refers to non-generation advancing specs bumps for the macbook air and pro lines.
-the second set of SKUs most likely refers to a 3-point mac pro generational update (single double server) but it COULD refer to a 3-point mac mini generational update.
-the "Good Better Best" titles are somewhat confounding, as you'd never call a server model "good." It could refer to a low-end Mac Pro or Mac mini as opposed to a server model. Or the "Good Better Best" could be made up information. In fact, all these SKUs could be a sham just to give us something to talk about. Well played, shamsters.
Last edited:
No internal expandabity ??? WHAT ?!?!?
No internal expandabity means my next machine will be a Hackint0sh, without question.
The whole point of the Mac Pro IS internal expandabity. If Apple has the audacity to actually release a machine like that, they can kiss my $$$$ goodbye.
Ill build the machine that I want. To hell with them.
No internal expandabity means my next machine will be a Hackint0sh, without question.
The whole point of the Mac Pro IS internal expandabity. If Apple has the audacity to actually release a machine like that, they can kiss my $$$$ goodbye.
Ill build the machine that I want. To hell with them.
True, but would 100 active concurrent users be more than a quad core i7 could handle? 10 years ago 100 concurrent users would have been a feat (I'm thinking all the Netboot server demos that used to happen with 100 clients), but these days I'm not so sure.
Thanks for the very informative post. I brought up the "leaked" SKUs from 9to5mac.com hoping we could make some sense of them. I was surprised that no rumor site had investigated them yet. I really hope that the SKUs you pointed out belong to Mac Pros, but I also know the Fall is the safer bet with the arrival of the new xeons.
Chuckle, Apple's XServe transition guide had 100 users on a Mac Mini with a 2.66 GHz Core 2 Duo.
www.apple.com/xserve/pdf/L422277A_Xserve_Guide.pdf
Can argue AFP filesharing workload isn't media stream workload, but probably right kind of order of magnitude differentials for what a Mac Pro server would be allocated for versus something with more mainstream Core i (or Xeon E3) solution.
There is a reason why OS X Server deployments exploded deployments once Apple made it affordable on the Mini. OS X Server deployments went up after the XServe got canceled; not down. It actually works for a ton of folks. There really wasn't a strong motivating factor as to why folks needed to spend $3K to serve up files and some centralized serving for a group of 4-10 folks.
That is what I was going to say.
Given that however, the problem is once you're in the high computational workload category, OS X is a really puzzling choice of OS, and you can get a *better* server elsewhere.