I think it's more a question as to whether Apple's making enough money from their sale. Once could make the same arguments about the xserve, unless you could point to some sort of proof that Apple was losing money on them.
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.
MacPro-where next?
- Thread starter woodbine
- Start date
- Sort by reaction score
I think it's more a question as to whether Apple's making enough money from their sale. Once could make the same arguments about the xserve, unless you could point to some sort of proof that Apple was losing money on them.
I get your point, but there is something you're not considering.
You're looking solely at a comparison of PCIe lanes assignments (16, 8, and 4) for GPU usage. Not what happens when TB is added to the mix.
Granted, TB is bi-directional, but not each channel. So if the band is saturated on either the up or down channel, TB cannot dynamically re-assign the unused channel and use it as a uni-directional interface to increase the usable bandwidth while that request has priority.
So you've essentially half of the 4x lanes the TB chip connects to in either direction (2x lanes used for upstream, and the other 2x lanes for downstream), and this is before overhead and latency. Hence the 800MB/s per channel sustained throughput according to Intel.
At least with PCIe, you get all 4x lanes = more than double the bandwidth, as it's not burdened with the additional latency and overhead associated with TB (PCIe = bi-directional, but the physical links are uni-directional per transaction; handled by the transaction layer portion of the PCIe spec).
This isn't to say that TB is useless for attaching a monitor, as it will be fine for attaching a modest sized model from what I've seen of the tests done by Intel for things like web browsing or watching a movie. But I suspect if say an animator was trying to use it for viewing a multiple layer project in real time (output generates more than 800MB/s of data to the monitor), it would choke vs. a dedicated monitor interconnect.
TB adds it's own signal latency and overhead (i.e. protocol conversions) over and above the PCIe interconnect it's attached to.
So TB is far worse than PCIe. It definitely looks that way to me as well, and there's more to it IMO than just Apple's recent FCPX software (like you, I see it as professional software = fix the major bugs/missing features before you release it).
Specifically, however, I'm thinking in terms of economic viability. I'm expecting the additional complexity of workstation CPU's to increase Intel's pricing, and Apple's not known for low margins. So combining these two factors, the MSRP could reach the Ionosphere in short order in another couple of CPU generations. Once the MSRP is out of reach even for independent professional users, which I've the impression makes up the bulk of MP sales, it will reduce the MP's sales figures to the point it's not going to be a profitable market segment, and Apple will EOL the MP.
At some point, someone's budget limit is reached, and there's nothing else they can raid for the difference. The only alternative for Apple to this is either reduce their margins on the MP as we know it, change what the product is (likely cease to be a professional product), or replace it with something else (where an TB equipped iMac comes in).
Not to say such an iMac is ideal, but I think Apple will consider it a viable alternative given the alternative to retain the MP (won't be willing to reduce margins enough to keep the MP a viable product line). Particularly when considering some of their other ideas, such as product design = super compact (iMac clamshell), and "you'll get what we give you" attitude towards users.
Then there's the profits generated by the gadget side to consider, which they've been cleaning up with. Which doesn't bode well on how much attention a small professional user product like the MP receives (hint: noticed the increasing bug count/severity since ~2009?).
It's where the money is, so they'll naturally give this area more focus.
But I suspect it's taking their attention from less profitable markets, though still currently viable, to the point they won't remain viable for long (even if pricing isn't an issue, as I've noticed a notable increase in bugs in the last couple of years).
As you know, recent system updates have taken more than 365 days per.
So Jan 1, 2011 to Dec 31, 2012 year still counts as a year by your math?
If you count days, it comes to 730 (usually only 729 days, but 2012 is a leap year), which is 1 day short of 2 years.
I'm not indicating that a TB equipped iMac will be an ideal replacement for a MP, but taking the economics (scale of economy to be specific) into consideration, I'm not seeing a solid future for the MP by any means. Nor am I suggesting this is going to happen now. But not in the distant future either (TB may see a speed bump by then, making it even more attractive to Apple as a storage interconnect for professional users).
The entire workstation market is in the midst of some serious changes (increased socket count, TB, and Intel's quantity pricing), and it's going to be a "wait and see" scenario to see what shakes loose. It will all come down to the economics for niche segments, which the MP is.
I know we're of a different opine on this, but I've not seen the evidence to show the MP's sales figures are where they'd need to be in order for it to have a long term lifespan (need proof as to quarterly unit sales of the MP only to shift on this particular portion). But pricing on Sandy Bridge E parts will help too (indication as to what direction Intel will go with the more complex parts they've in the works), just in case Intel has a change of heart, and lowers their margins.
I've asked for verifiable proof before, but so far, nada. Apple and Intel don't seem too interested in reading MR and sending me an email or PM.
No.
Present MacPro's GPUs (HD5770 and HD5870) are faster (HD5770) and way faster (HD5870) than iMac's GPU (HD6970M). All the difference resides in that M, that means mobility GPU.
Ram size has little to nothing to do with GPU performance in today's games...
Future MacPro's GPU, probably a HD6970 or a HD7970 (the latter is going to come out by the end of the year), will increase furthermore performance gap....
Last edited:
Correct, probably because of the 128bit memory bus of the HD5770 that cripples it.
Last edited:
A few thoughts:
- I think Apple is in a really awkward place right now with the Pro market. A segment is calling for "bells and whistles" refreshes - TB, a new video card, etc. But if they do that, another segment is going to take the lack of a proper SandyBridge chip (versus the consumer-level chip they'd stick in its place) as a sign they're abandoning the market.
- Looking at how heavily they feature in retail stores is...flawed. That's like saying you don't think Volvo supports their trucking division because you don't see many at the dealership. Most Mac Pros, I'm nigh positive, are custom orders from the website.
- Since getting my Mac Pro, my productivity having a proper workstation has shot through the roof. I love it. If it leaves Apple's line, at least for work I'll switch to...another workstation. And my grant money, and funding for laptops, etc. will go with it.
- I think Apple is in a really awkward place right now with the Pro market. A segment is calling for "bells and whistles" refreshes - TB, a new video card, etc. But if they do that, another segment is going to take the lack of a proper SandyBridge chip (versus the consumer-level chip they'd stick in its place) as a sign they're abandoning the market.
- Looking at how heavily they feature in retail stores is...flawed. That's like saying you don't think Volvo supports their trucking division because you don't see many at the dealership. Most Mac Pros, I'm nigh positive, are custom orders from the website.
- Since getting my Mac Pro, my productivity having a proper workstation has shot through the roof. I love it. If it leaves Apple's line, at least for work I'll switch to...another workstation. And my grant money, and funding for laptops, etc. will go with it.
True, but the difference between volvo and apple is that volvo is continuing to support their trucking division and the contention here by a number of is that is not pay enough support on their MP line. While I agree with the other poster, in that apple is focusing on the sector that offers the most profits (consumers instead of pros) the fact remains that focusing more resources on other segments within apple leaves less resources for the MacPro
I don't think anyone would argue that the MP is a beast of a machine that is extremely fast. Its a great workstation.
As you know, recent system updates have taken more than 365 days per.
So Jan 1, 2011 to Dec 31, 2012 year still counts as a year by your math?
If you count days, it comes to 730 (usually only 729 days, but 2012 is a leap year), which is 1 day short of 2 years.
Yes, I mean once every calendar year.
Really? I think the Mac Pro will be the last Mac to go. The iPad is cannibalizing the low end machines, and will eventually work it's way up to the iMac, but the Mac Pro represents the "truck" sort of machine Steve Jobs has said is where the future of the Macintosh is going.
If anything, Steve's comments have implied that the high end Macs are going to be the last to go, not the first. Why would Apple keep the low end Macs that are going to be replaced by iDevices anyway?
I think (and this is more likely to be believed if the custom chip rumors are true) that Apple is going to look for ways to bring the Mac Pro price down. Basically bring the Mac Pro back towards what the G4 was, with like the G4, more expensive dual socket versions on the high end. The Mac Pro ships in single socket configurations now, but still with high end Xeon chips for no reason, which Apple can't be happy about.
Maybe some sort of i7/Xeon cross chip is in our future?
I know we're of a different opine on this, but I've not seen the evidence to show the MP's sales figures are where they'd need to be in order for it to have a long term lifespan (need proof as to quarterly unit sales of the MP only to shift on this particular portion). But pricing on Sandy Bridge E parts will help too (indication as to what direction Intel will go with the more complex parts they've in the works), just in case Intel has a change of heart, and lowers their margins.
I've asked for verifiable proof before, but so far, nada. Apple and Intel don't seem too interested in reading MR and sending me an email or PM.
Apple has already said they see the Mac's future as a high end Pro market (Steve Jobs on trucks vs. cars). I don't see how cutting the Mac Pro fits with Apple's vision at all.
Not to mention, I have heard pros on the Mac freak out about once a year about Apple not caring about them ever since OS X launched. Which is probably the reason I'm kind of grumpy about this entire argument. This is not new, and every time Apple has come through.
The FCPX complaining reminds me a LOT of the OS X complaining. It's almost verbatim. It's not that I don't understand, FCPX has issues just like OS X did also have serious issues. But Apple was actually listening (despite the naysayers) and they fixed all the big issues in OS X and then went in new amazing directions.
Everything will be fine, everyone just needs to calm down.
.
Everything will be fine, everyone just needs to calm down.
I'd like to believe you on this. I'd like to believe that the MacPro is just coming into its 'Golden Era' and not going into its 'Long Retreat'.
And since I get to decide on what I believe, I think I will choose to believe you. Thanks. I feel much calmer now.
Switching to i7s wouldn't help at all. SP Xeons cost as much as their non-Xeon counterparts. For instance W3530 is $294, just like i7-930. The only reason why SP Mac Pro costs what is costs is because of Apple's fat margins.
Now where did I put that facepalm pic....Yes, I mean once every calendar year.
I realize your logic, but it's not accurate (what you're calling a year can be well over a year as it can be up to 1 day short of 2 years
). Quite a difference in interpretation of a year without numbers as members will take it to mean 365 days. Reminds me of things like political logic, marketing jargon, modern accounting practices, ...
Twist, bend, manipulate information to fit whatever opine they desire rather than reality and offer it up as fact. Consider the XServe. They issued statements to calm potential buyers down just a few months IIRC before they announced it was EOL.
So I take such language with a grain of salt, as it's marketing (intended to get potential buyers to purchase = Apple ends up with less stock they need to unload). Their motives for this are based on financial needs/desires, not truth.
Their entire line of computers will change in the near future, including laptops (i.e. expect the MacBook to go, as the Air has caught up to it in terms of performance, MP <mainly due to cost reasons> and possibly the Mini <not sure if it's attracting the switchers as it once did> on the desktop side).
Models will only stick around if they remain profitable. Apple's reason for existing is to earn a profit, not lose money, just like any other corporation. I wouldn't suggest you forget fact this when dealing with any corporate entity. Notions such as loyalty mean nothing vs. the bottom line.
This sort of mentality is why things like outsourcing occurred in the first place (jobs to entire industries, particularly manufacturing).
When possible, Yes.
But their methods for cost reduction have via things like outsourcing (i.e. shifting to an ODM implementation for things like the MP).
For them to get a chip that's cheaper, they're going to have to take a step or two backwards. Even if it's custom, as performance costs money (complexity in the case of modern CPU's, as the days of just ramping the clocks are gone).
This is actually what we have now (since 2009).
Clock per clock (same socket), the difference between the i7 and Xeon variants is that the Xeon retains ECC functionality.And more importantly, the costs don't differ (i.e. same clock i7 or Xeon = same cost).
Now this doesn't mean they couldn't save a little bit of money, but that's due to using non-ECC memory, not the different CPU versions. But they could do this with the Xeon as well (Nehalem and Westmere Xeons can run non-ECC memory).
DP systems are stuck with Xeon due to the design requirements (i7 doesn't have a 2nd QPI, and will not work in DP systems due to the different chipset used; X58 = SP models, 5520 = DP models).
Statements designed for panic prevention I suspect.
If the financial side doesn't work, they'll EOL a product.
I saw the complaints as valid from an enterprise POV, as that many bugs, and the severity of some should never have been released to the public as a means of collecting more income for that quarter. The enterprise market is founded on stability as an absolute necessity, unlike the consumer market. This is short term thinking that's more in line with consumer products, and if they're not careful, will have negative long term consequences.
The one way this is a justifiable approach, is if they don't care and are in a "dash for the cash" mode (know that their professional products are near EOL; they may even have established dates
).The definitely treat the MP like a consumer machine, and it shows. But this type of behavior doesn't foster trust in this particular market segment (other vendors would be chastised for that sort of thing, and could actually go out of business if enough of their customers are pissed enough to abandon them). Now a pissed off user here or there doesn't mean much in the consumer side, as it's only a single machine or site license.
But in the enterprise market, those customers buy large, even massive quantities of systems and software licenses per year. So the impact of a single customer is significant to the bottom line.
I think this is the difference you're not quite grasping between these markets.
Now where did I put that facepalm pic....
I realize your logic, but it's not accurate (what you're calling a year can be well over a year as it can be up to 1 day short of 2 years
Reminds me of things like political logic, marketing jargon, modern accounting practices, ...
Yeah, but that kind of applies to all Apple machines. All Macs aren't necessarily updated exactly every 365 days, but they're updated once a year.
Again, if it's true of all Macs, it's not really "Mac Pro freak out" material.
Consider the XServe. They issued statements to calm potential buyers down just a few months IIRC before they announced it was EOL.
So I take such language with a grain of salt, as it's marketing (intended to get potential buyers to purchase = Apple ends up with less stock they need to unload). Their motives for this are based on financial needs/desires, not truth.
Calm buyers down? I'm not so sure about that. I worked in IT for a long time and I don't remember any of that. If anything, Apple's server solutions were so buggy that we all knew they either had to be fixed or they were going to be discontinued.
Say what you will about the Mac Pro, but bugginess isn't one of it's problems. It's a solid machine that someone is obviously paying attention to.
I don't think that's what Apple is going for. They could sacrifice the very very high end of the market that requires Xeons (which, let's face it, probably is not even very many people here at Macrumors), and instead use dual socket i7s. Instant price drop for very similar performance.
Now this doesn't mean they couldn't save a little bit of money, but that's due to using non-ECC memory, not the different CPU versions. But they could do this with the Xeon as well (Nehalem and Westmere Xeons can run non-ECC memory).
Sure, but what I think is more unnecessary is using Xeons in single socket machines. If they used i7s it would certainly help their release schedule too.
Unless, again, they had a custom chip...
Which, again, is baseless. Everything points to the Mac Pro being profitable. This just seems like more panic.
I don't know if this carries to the Mac Pro, though. The XServe family had extreme stability problems, which is no good in enterprise, and was probably why the XServe was cancelled. Is this applicable to the Mac Pro? Not really.
The one way this is a justifiable approach, is if they don't care and are in a "dash for the cash" mode (know that their professional products are near EOL; they may even have established dates
Dash for cash? Do you know how much money it took them to write FCPX? It was a massive project. It probably won't be profitable for years.
Your logic doesn't really fit. An extremely expensive ground up re-write to get them ready to dump the product? Huh?
Treat the Mac Pro like a consumer machine? How so?
I mean, there are faults, like the graphics card situation. But those faults are classic Apple, and have existed forever. Heck, remember everyone being upset the Power Mac G3 didn't ship with the Radeon?
Again, I hear these arguments about once a year.
So, to clarify, I worked in the video production software industry for a few years, hence my interest in this discussion...
Video customers can be huge, and certainly they could impact FCP X's bottom line. But, honestly? Looking at applying this back to the Mac Pro... A fair amount of Mac editors never used FCP to begin with. A lot of Avid, a lot of CS suite. Those customers still buy Mac Pros. FCP customers don't only buy Mac Pros either, a lot of iMacs, a lot of laptops....
I do think Apple probably sells a decent number of FCP/Mac Pro bundles, some to enterprise, but the Mac Pro market is much larger than video editors. You have image editors, audio editors, 3D modelers, scientists, developers... The Mac Pro could survive a total failure of FCPX, honestly (which I think is still unlikely.)
Having been around in Academia too, I think Apple sells a lot more Mac Pros than you're thinking. I worked for a university that put Mac Pros on every professor's desk. I think you're vastly underestimating the number of Mac Pros Apple sells.
And now that the XServe is dead, the Mac Pro is Apple's enterprise server as well. Are you suggesting Apple is going to fully abandon that market too? That move would make no sense, as having some sort of enterprise server is definitely tied to cooperate iMac/Macbook/iPad sales.
There is a lot of interconnects here that make this more complex than "Apple is killing FCP! Apple hates Pros! Apple wants to kill the Mac Pro!" I think people here are just viewing this through a FCPX lens instead of looking at the big picture.
If the Mac Pro gets a significant re-design, I'd say that's a good indication that if Apple is willing to make that kind of investment, the Mac Pro is not going anywhere for a long, long time.
Not entirely true. Mac Mini wasn't updated at all in 2008 but otherwise I think you are correct.
There are no dual-socket i7s, they all have either single DMI or QPI and you need two for DP system. Only Xeon 5000-series offers support for DP systems (as well as 7000-series and the new E7 series) and the second QPI link costs a hefty premium (W3680 is $999 while X5680 is $1663, both are 3.33GHz 6-cores).
Not necessarily. Again, the prices are the same when ignoring the minor extra from ECC RAM. Usually Xeon parts are released simultaneously with their iX counterparts (e.g. Gulftown and Westmere-EP were released on the same date). Nehalem was an exception and Xeons parts came months later but that was a huge change in terms of micro-architecture. Also, Apple would still need to wait for Xeons if they want to keep the DP Mac Pro alive.
But what would it help? You still need a second link for connection between the CPUs. Besides, like nano has already said several times, a custom chip would most likely be based on the designs of current chips (i.e. either Westmere-EP or Sandy Bridge), it's very unlikely that Intel would do a customized architecture just for Mac Pro (and if you think the price is already high, you could add a zero to the end because designing an architecture costs millions).
But there is the big if. I'm sure we all will agree that MP isn't going anywhere anytime soon if Apple shows some real love towards it.
This is where I go by the laws of mathematics (specifically that of the Gregorian Calendar which states: 1 year = 365 days). Within a year is less. Anything over (except for a leap year, which will raise it to 366 days), is greater than one year.
Simple.
I wouldn't make such a big deal if it were only over by a month or so, but when it hits the 450+ day mark, it's too far over to be stated as a year.
They needed to sell off the remaining stock. Had they not done this, it could have left them with more systems they would have had to get rid of at a discount.
As per why it went EOL, the bugginess didn't help. But the bigger issue was that it was a niche market that could not sustain itself.
It's actually a good example, as the lack of proper support, bug laden and ran hot (even for it's chassis), caused the sales volume to be eroded to the point where unit sales were too low to continue providing it (not enough profit, or worse, they let it get to the point they were losing money on each XServe sold).
- Problems with OS X (some major, such as the audio bug that caused it to run hot <and went on for ~ a year>, or how some updates broke RAID systems, which is a major issue in the enterprise market). Check.
- Problems with some firmware revisions not supporting PCIe storage controllers (would work with some revisions, end up broken under others). Check.
You'd demand someone's head if companies like Sun (aka Oracle), IBM, HP, or Dell pulled a stunt like that.
You mean they're not interested in high profit margins or making sure their product lines don't significantly interfere with/cannibalize one another?
No such animal exists.
DP and MP configurations are purely the realm of the Xeon line, and as an IT pro, I'm truly surprised that you don't already know this.
It won't make a difference on the SP systems in terms of performance or cost (ECC support in the Xeon versions is the only difference between i7 using the same socket).
Release schedules, maybe. It depends on whether or not they decide to delay the Xeon variant (depends on fab capacity and order requirements that need to be filled to meet supply contracts).
DP chips need to be able to connect to the chipset correctly, and this will mean it has the same issues as existing and upcoming parts (means of connecting to 2x CPU's).
I'm not saying that the MP isn't profitable right now (go back and read all the posts carefully, not skim).
What I am saying, is that the financial factors from Intel (increase in CPU complexity tends to translate into higher costs, and Intel wants what they think is a sufficient profit margin on top of this) and Apple (desire for high margins) indicates that the MSRP's are going to climb even further. Now if you think about this in context to Supply and Demand economics, the sales volume will be eroded away. Eventually to the point the MP is no longer financially viable (no longer able to provide an acceptable margin by Apple's standards). Once it reaches this point, they'll do the right thing by business logic, and EOL the MP (I just don't see Apple willing to take a loss on each MP sold).
You don't seem to be able to properly grasp the economic impact involved... The sales figures were too low to support the XServe, so they EOL'ed it (bugs played their part in reducing the sales volume per quarter to the level that dictated EOL).
Not as much as you might think...
They've dumped previous code. So Yes, technically this is more of a ground-up approach, but it's limited in what it does (missing features pros need). Then consider the fact that Apple laid off FCP developers while it was in development, and it indicates that it was half-baked from the start, and done on a minimal budget.
The went ahead and released it for one of two, possibly both reasons:
- Shifting to prosumer users (getting away from pros).
- Wanted an income off of it, despite the fact that it was Beta-ware for professionals.
Support (software <OS X and even their professional application suites> and hardware <see above>) is that of a consumer system, despite that it's actually an enterprise machine. I can't fathom how anyone could examine the existing support and consider it enterprise level.
No, not all graphics pros use FCP. But from what I'm seeing, the PC side gets more attention (software gets features before OS X, and more hardware options).
Avid and Adobe don't have a lot of love for Apple given how they've been treated. So if it's easier to add new features to the Windows version, it will likely get it first. Not to mention that Windows is a larger market segment = more potential for sales volume.
We don't know what the actual MP sales figures are, but the fact that it's not separated from all of the other computers, it's not a good sign IMO.
As per other markets besides creative professionals, I've not seen a lot of Macs of any kind used in the scientific field. PC versions are by far more common.
What departments?
All or just a few?
I ask, as I came up through engineering (and had access to other scientific Dept's., including medical as I was the tech that fixed their broken systems <mostly Compaq's>), and didn't see them used there. Granted, this has been awhile, but the software used is so specialized, that the Mac variants aren't many, and may not be without headaches (i.e. file formats won't load properly when used on another platform).
I've even seen this with something as simple as Office for Mac (opened a spreadsheet created under OS X, and it was a mess under a PC system - columns were wrong in particular).
Most Mac laptops I've seen in the corporate world are tied into a PC infrastructure.
So OS X Server may not be as important as you might think (niche of a niche in terms of user count).
I'm not sure about others, but I'm not relying on FCPX. Take a closer look at Intel's roadmaps, and you should begin to see that the workstation market is becoming more of a niche, which usually means higher pricing. Factor that in with Apple's margins and how high MSRP's will reduce the sales volume, the MP will become less and less profitable. To the point it's no longer a viable product.
Massive IF here, and not enough substance to support this ATM ("friend of a friend" sources are too unreliable to give such sources any merit).
How the heck do you think the MP is the future of the Mac, compared to laptops and the iMac? Not only does the MP constitute a faction of Macintosh sales...it's a an ever smaller percentage every year!
I think you've grossly misread or misinterpreted what Jobs said.
DaringFireball's John Gruber is the blogger closest to Jobs, and Gruber says: In short order there will only be two Macs: Air and iMac.
Had nothing to do with sales figures, it simply wasn't making any money. XServes, unlike the Mac Pro, were very expensive to design.
You're way off the mark here.
Being familiar with what is in FCPX, they started this work back in 10.4. Tons of OS APIs are in place just for FCPX. The app itself is very complex, very costly to write.
Minimal budget? No. No way in hell. Looking at FCPX it was a massive undertaking. Large portions of AVFoundation (which is not a small API) are targeted specifically at FCPX. I just don't see anything here which says minimal budget. We're looking at a ground up re-write of QuickTime for FCPX, as well as FCPX itself. Not cheap. Not cheap at all.
Even assuming it was somehow cheap, your idea basically breaks down to:
1. Redesign a product from scratch
2. Cut the product
3. ????
4. Profit!
Makes no sense, dude. I don't know why they cut engineers, but looking at FCPX it is not something they could have had a few interns do over their coffee breaks. They would have needed a large team over multiple departments.
It's an evolving app. Report that came out today basically said that all pro features will be restored except direct FCP project importing (XML importing will return.)
I think they wanted to show progress, and it didn't go like they had planned. Which fits with a long history of 1.0 Apple product releases that are really beta. This is pretty classic Apple.
Yes and no. Macs are popular because they're machines that have everything you need out of the box, and a fair number of video pros like Mac OS.
Ever used a Dell Dimension? They're not well designed machines.
Avid and Adobe don't have a lot of love for Apple given how they've been treated. So if it's easier to add new features to the Windows version, it will likely get it first. Not to mention that Windows is a larger market segment = more potential for sales volume.
Despite all the talk, Apple is still working with Adobe, and Adobe is still working with Apple. CS Suite is pretty neutral territory for them. I don't see Premiere or After Effects getting dropped.
Mini is not separated out either. I don't see it as a sign of anything, only that Apple lumps all their desktops together.
I don't know what to say. Macs are very common in science. It's the entire reason XGrid exists.
I said all professors. That means all departments.
I thought it was overkill, but whatever.
Coming up through engineering, I worked with a lot of Linux machines. But honestly? I saw a lot of Mac Pro cardboard boxes in the department too. And unless they just liked ordering the boxes by themselves...
I don't think people are using Mac Pros just for Office...
And Windows Server can't serve to iPads. Instant deal breaker for Apple. They have to do OS X services. And let's face it, Minis don't cut it for a lot of serving and Apple knows it. You can't avoid a powerful headless Mac.
(There are also a lot of OS X specific servers Windows Server can't do.)
Plus, do you really think Apple wants to give Microsoft that power over them? Even if they don't plan on strongly competing with Windows Server, they need that failsafe in place in case Microsoft screws them over.
Which may be why Apple wants a rumored custom chip. Stop supporting the ultra high end of the workstation market, bring it down to the large majority of pros. It would differentiate the Mac Pro as well.
Again, I think the general point here that you are missing is that if Apple is willing to sink money into redesigning something, it's probably not going anywhere.
Speaking from the academia side of things - I see a great many Macs in science and quantitative fields. A fairly heavy representation in places that benefit well from it being "UNIX but friendly". Statistics, bioinformatics, etc.
Less so engineering, and the fields that take engineers. I blame MATLAB
Based on the direction Apple is heading with the A4 and A5, and the advancements that are being made with these System-on-a-chip ARM processors, it is forseeable that Apple is heading to eventually desert intel and go with their own custom chips for the Mac computer, with OS X designed for the ARM/A7 or A8 chips by the year 2015 or 2016. After Ivy Bridge, I definitely see Apple heading in this direction. Mac OS X with evolve into something more like the iOS with time, and much more storage will be done in the Cloud. The only hiccup about this now is broadband speeds aren't quite where they could be...once we see 100MBit down/50MBit upload speeds offered to the consumers along with 4G LTE then Cloud computing will be very attractive. But I don't see Apple staying with intel and especially their Xeon offerings. Once 8-core and 12-core i7's come around we will see more of those targeted towards the upper-end. You already see alot of people on here ditching their 1st gen Mac Pros for the new 27" i7 iMac machines.
Apple has ditched XServe, Apple has discontinued matte display offerings (offering one glossy 27" option now), Apple upset their studio video editors with FCPX, they dropped Shake and Color, and they are no longer offering their software on disc (everything is going app-store/iOS/Internet/Cloud/iPhone/iPad, etc) That's the direction Apple is headed...total focus on the consumer and no attention to their serious professional, business, or enthusiasts that want the tower computing experience. It was something like 536 days between Mac Pro refreshes this last time, that should tell you something. The X5680 has been out awhile, the X5690 is out now along with the X3690, and yet Apple has not offered any of these chip options for their Mac Pros. If you want them, you have to add them yourself. A company like Apple should offer their customers these technologies when they are available, along with eSATA, Blu-Ray, SATA III, and other GPU options like the 6970HD which has been out for several months. Just more reasons why Apple DOES not really care much about about the Mac Pro and has not for awhile. These are not the days of the Power Macintosh anymore, this is a consumer, money-driven company now that is only concerned with how many iOS devices or iOS apps they can sell each day. It's all about numbers for Apple. It's just too bad.
Do you not understand how manufacturing economics works?
If they're not making money, they didn't meet their minimum target sales volume (units sold = break even point = no profit, but no loss either).
That minimum target is designed to cover the R&D, manufacturing, and all indirect costs associated with the number of units they initially order. If they don't sell that many machines, there's a loss. Anything over that, generates a profit, and where they want to be. Since they want high margins, they need to sell a fair number of systems above that.
Now I can understand some confusion with how the numbers are generated internally, as they're figuring their targets with an expected profit margin already built-in. But I'm trying to keep it as simple as possible to illustrate my point.
But since you're claiming they lost money, they didn't even meet the bare minimum target that keeps them from losing money (aka "In the Red", which is an old method of accounting; profits were written in black ink, losses in Red).
So how can you claim that Sales Volume (units sold) does not matter?
This is what I don't get from how you're explanation on the EOL causality of the XServe.
Dealing with inside information again?
No one else has access to it, so without a link to a source that supports it, it's against forum guidelines as Hellhammer recently mentioned elsewhere and can only be viewed as pure speculation (difference between "is" and "might" ways of presenting information; one comes off as fact without support, and the other as a conclusion that may or may not be correct in the end).
My point, is that even if they started from scratch (assumes it uses absolutely nothing already written in OS X in terms of API's), they didn't have to deal with spaghetti code, which tends to eat significant time during debugging phases. In my experience at least, starting from scratch allows for better expectations as to how much development time is spent on each section, and if done properly (not released as beta-ware), tends to have a better final result.
Then consider what they did release is beta-ware from what I'm seeing from various posts (mainly lack of features for the type of professional application it is, not things like loaded it and got a Kernel panic sorts of things).
So it appears there's a reduction of man-hours from what it should have been = not as expensive to do had they fully debugged it/added all the features professionals need at the initial release.
Now I'm not saying what they did do only cost Apple the equivalent of a couple of bags of peanuts, but it's not complete for professional software IMO. I'm actually more accustomed to such software retailing for much more than what they're charging for FCPX. Adobe and Avid both sell for at least $1k USD from what I've seen.
Now consider that electronic simulation software suites (by companies like National Instruments and Synopsys) go for $10k+ per isn't unheard of, and a single TCAD license (also from Synopsys) can actually cost as much as a house in some markets (stuff PCB and chip designers use respectively).
So thinking along these lines, FCPX @ $300USD is cheap, and and what seems to be an incomplete suite is the most reasonable explanation why (assumes Apple is at least trying to break even on it, rather than use it as a loss-leader to sell systems). If it is a loss leader, they could theoretically give it away if they didn't care how much they lost (not likely, given they're in business to make money
).Now considering this and other factors (Apple focusing on the consumer product lines, loss of the XServe, ... things that have already been discussed), it's no wonder graphics professionals see FCPX as an "old, dried out bone with no meat thrown to an old mangy dog" = Apple no longer cares about the professional market as they once did, if at all - sorts of sentiment.
Bad move for professional software IMO. If Synopsys or National Instruments blew it like this, I and others in the field would raise absolute hell. Of course as you now see how much this sort of software goes for, you can probably figure out why.
I guess it all comes down to what someone's definition of professional software is.
And that gets back to the point of Apple's definition vs. the rest of the world. What they call professional is pro-sumer quality, not professional by my and others' definitions in the enterprise industry.
Do you spend any real time outside of the Apple environment at all in the enterprise market?
If not, it could explain why the point I'm trying to make seems to be lost.
For me, no system has everything I need out of the box.
I always have to add additional hardware (think memory capacity, RAID systems, and bench instrument interfaces <GPIB adapter of some sort; USB to GPIB these days, so it's no longer an internal card>). And there's the software of course...
From what I'm seeing of true professional graphics professionals, they also have to add hardware and software as well, so I don't see the MP as "ready to go" out of the box for them either.
Now I realize what you're getting at (just add software), but as you know base systems from any vendor have bottlenecks that need to be addressed for professional users, namely increased memory and storage. Since these sorts of users add hardware to get it to a usable machine, I don't agree with your sentiment on this issue.
As per OS platform, that's dictated by the software they want to use. It just happens in the case of OS X, that means getting a MP if they need the slots (particularly for discrete graphics cards) and HDD bays for upgrades (based on hackintosh = bad idea for professionals earning a living IMO).
I've not said they've stopped working with one another, but there is strong evidence they don't like each other all that well. But they still do it as it's mutually beneficial (each company helps the other sell their products).
I don't get to university campuses much (last time was 1999), so most of my time with workstations is spent in engineering. The rest of my enterprise exposure doesn't use Macs either (mostly banking and telecommunications data centers for clients, such as Chase and AT&T).
What schools?
I'm curious to see what schools, and how many have shifted to Macs, as out in the wild, I don't see them too often. I do know of some used by medical professionals, but it's not many (local hospitals still use PC's for servers).
Friends/colleagues I keep in contact with in other areas of engineering don't either (mechanical and petroleum for example).
What kinds of engineering?
I've not seen them much in engineering (what the guys that did <technically CS, not engineers doing contract work for AT&T>, used Apple laptops and provided test data from cell towers in a .xls spreadsheet file). Just office staff otherwise in engineering offices.
To me, this is yet another clue that indicates Apple isn't really interested in the Server market though.
They could even choose to take OS X and license it out to other vendors (or at least one). Granted, this opens up other cans of worms (increased hardware support is a big one), but it's theoretically possible at some future date.
Specifically, I keep thinking of this possibility with their own Data Center they've created for iCloud. Without XServes, they'll need to run PC's from other vendors, as shelves full of MP's isn't really the way to go IMO (too much physical space required, even if they do come out with a convertible case <one that can be used either as a desktop or rackmount if you screw the ears to it>). Just not enough units per rack I suspect to keep it as cost effective as slimmer server rack-mountable cases would be able to offer. Floor space and electricity usage are major factors in Data Center operational costs, and has drawn a lot of attention lately.
Why do you think Intel is working so hard to keep power consumption down, and still increase the cost/performance ratio simultaneously = needs of large enterprise users?
If you consider if Apple really is getting away from Servers, it all makes sense. MS won't have power over what Apple does, as they won't be in that market.
The only area I can see Apple needing Server capabilities, is with the iCloud services, and you've already pointed out that it's not really reliant on what OS X Server even is. Which means Apple already has a solution to this problem.
Hardware is the only area that the Data Center will run into issues, and Apple could easily get that fixed by hacking their own OS + whatever they've created to handle iCloud servicing to run on whatever specific machines they'll be running.
They wouldn't need a custom chip for this though. Just select something that's already available, particularly a single processor based design.
The problem is, it's either going to have to be a Sandy Bridge E based part to keep the performance anywhere near what users already have with current SP MP's.
If they selected an LGA1155 for example, it would be a step backward vs. what users are accustomed to now in a base SP Quad. Yes, the SB's used in the current iMac are quick in terms of single threaded performance, but they fall flat on their face for I/O throughputs, and multi-threaded performance past what 4x cores can do. These versions of SB are meant primarily for mainstream computers and will be used as low-end Xeons (there is an LGA1155 due out).
So I can see where the idea of a custom chip comes in. The problem is, for a truly custom chip (totally different part from the ground up), it means a lot of R&D cost, and that's not likely divided by enough systems if it's only available in a MP.
Why do you think Intel is willing to sell their chips to multiple vendors?
It's not because they think it's a nice thing to do.
It's so they can recover their costs and still sell parts (keeps the R&D per unit low enough they're affordable to system vendors, and ultimately users once all the costs and margins are added). They have to use economy of scale to keep it affordable. I don't know about you, but I'm not willing to pay in the hundreds of thousands or more per CPU (all comes down to how many units will be manufactured). Now if they do a "custom" chip that's based on what they've already designed, such as the Sandy Bridge E, that is possible. But there's still cost issues to deal with, and to keep it low enough to be viable, there's either little changes (i.e. skip the IHS, which isn't changing the circuit topology), or cut something (i.e. eliminate the additional QPI's so it's an SP only design).
The reasoning behind this, is costs. They won't be reduced much, if at all (no IHS = naked = won't save much). Where the "if at all" comes in, is with cutting it down as a reduced part (reduce the transistor count as a means of getting more parts per wafer), as it could even increase a bit as there's still additional R&D involved that will only be divided by Apple's machines.
It all depends on how much smaller it gets as to whether or not Intel can get a significant increase in yield per wafer (more parts per wafer = less production cost per part, as the cost of shooting a wafer is ~ the same if it's 1 part or 1000 for the same process and layer count; small differences if there's less materials used, but the time and energy expenses are the same for the process). Now we're talking about a special run for Apple (they're quite willing to do this), but even if it works out (can get more parts per wafer), Apple still has to order enough of them to make it financially viable (how they'd be able to keep the cost per unit down, as there are minimum part counts to make it viable - why parts are made in lots).
Now as time goes on, a future ARM based design could get them most of the way (PA Semi is a licensee of ARM), and make modifications if the ARM design being considered is actually sufficient for workstation use. But this isn't possible right now.
Either way (now or future), it comes back to how many MP's models they can sell as to whether or not it's profitable.
In the case of the iPhone, it works out. I suspect it could for a future Mac consumer grade computer (laptop or even the iMac if they skip Intel or AMD). But they have the sales volume high enough it's feasible.
Unfortunately, the available information on the MP doesn't lend me to think this is the case.
I'm not sure they've sunk that much into FCPX, given what it is, and what they're selling it for (see above).
I don't see much of this now, but the school may matter as the 12 years that have passed since I was last even on a university campus.
Back then, they mainly used Linux on PC's from what I saw (didn't see any Macs in their offices, and those I was invited to, their homes either). They usually wrote the software themselves, as there wasn't anything available commercially (research afterall, so it's usually on something that's not been discovered already, save peer review).
Blame MATLAB!?!?! Did you fall off of your rocker and end up brain damaged?Less so engineering, and the fields that take engineers. I blame MATLAB
Seriously though, I suspect it's the most common professional software application across multiple engineering disciplines, and for good reason (for stuff a spreadsheet can't do).
Do you not understand how manufacturing economics works?
If they're not making money, they didn't meet their minimum target sales volume (units sold = break even point = no profit, but no loss either).
I understand that, but I think with the premium the Mac Pro sells for, Apple's target volume isn't nearly as high as you think it is.
XServe was an extremely buggy product. Unlike the Mac Pro, Apple was looking at having to fix a buggy product, in addition to continuing to produce it. I don't think it is analogous.
I've spent time working with Apple's server products. Most admins I know who have to do it full time don't like dealing with Apple servers. They are just plain unreliable as far as software goes.
Again, different factors than the Mac Pro.
I think this logic carries for long term strategy (ahem), but not short term. Even if you clean up spaghetti code a re-write is significantly more expensive than an upgrade, which is why so many companies keep upgrading. Re-writes are expensive.
I think you're brushing up against my point again. A re-write only makes sense in terms of a long term strategy, not a short term strategy.
I agree, it's beta ware. Doesn't change my point. Still expensive to write.
And, if they are planning on dumping it like you are saying, why are they wasting time re-adding the missing features?
I very strongly disagree. Please back up with proof. Go look at the APIs and come back with proof please.
At this point I'm pointing to the actual work that came out of FCPX, and you're going with a gut feeling. Until you start citing things, this discussion is going no where.
Ok, this is an entirely different point than "it was cheap to write". A lot of consumer software is expensive to write as well. I don't see the connection.
I'm not saying you shouldn't be mad about the feature set. But I am saying that the idea that this is a project about to be cut is pretty baseless.
Now consider that electronic simulation software suites (by companies like National Instruments and Synopsys) go for $10k+ per isn't unheard of, and a single TCAD license (also from Synopsys) can actually cost as much as a house in some markets (stuff PCB and chip designers use respectively).
So thinking along these lines, FCPX @ $300USD is cheap, and and what seems to be an incomplete suite is the most reasonable explanation why (assumes Apple is at least trying to break even on it, rather than use it as a loss-leader to sell systems). If it is a loss leader, they could theoretically give it away if they didn't care how much they lost (not likely, given they're in business to make money
Apple dropped the price of FCP very significantly when they first bought it out. Wasn't exactly a bad sign then.
Now considering this and other factors (Apple focusing on the consumer product lines, loss of the XServe, ... things that have already been discussed), it's no wonder graphics professionals see FCPX as an "old, dried out bone with no meat thrown to an old mangy dog" = Apple no longer cares about the professional market as they once did, if at all - sorts of sentiment.
Alright, let's back up here:
1) XServe is an entirely different market. Enterprise != professional creative user. I'm not sure why this keeps being brought into the discussion.
2) The Mac Pro user base is far wider than the FCPX user base.
There are a lot of broken connections here. Does cutting the XServe mean that Apple is about to cut the iMac? After all, the iMac was the premier desktop used by XServe clients.
See? Doesn't make sense. People are trying to read into things because they're worried about the Mac Pro.
The only way I'd relate the Mac Pro into this is that because the Mac Pro is now Apple's enterprise server, it makes it harder to cut. I think Apple knows they have to maintain at least one enterprise server line for their iOS strategy, at the very least. Windows Server doesn't have iOS sharing services, and I'm not betting on Microsoft adding those any time soon.
Again, Apple does this all the time. It's just FCP's turn.
Someone made this years ago:
http://www.misterbg.org/AppleProductCycle/
Also, here is another example for you.
XCode 4 was a very bad, from scratch re-write of an existing product. It cut many pro features, and had many developers up in arms. It was also very crashy.
Does this suggest Apple is about to cut iOS development, or get rid of XCode? Or, perhaps, is this just something Apple notoriously does with re-writes?
(In fact, Apple also promised fixes and features re-added with XCode 4 just like they are with FCPX, and they are actually starting to show up.)
No, it doesn't have anything to do at all with this. I don't care if people think FCPX is pro. I'm saying Apple is addressing people's concerns, and it's really kind of unrelated to this argument. You're trying to point to one pro product (that Apple is fixing) and trying to argue Apple is planning on cutting all Pro things. I'm pointing out why this is a bad argument. The only tangential thing to this is I believe Apple investing so much work in FCPX because they are keeping it around.
Sure, and I'm not sure why it's relevant. Windows Server is great and all and has solid hardware available but it still has no iOS sharing or deployment services.
If there is one thing we can agree on, it's that Apple is going to support iOS as much as they can. And you're telling me that they're going to drop the only enterprise hardware that can share to iOS?
Ok. Doesn't change that Mac Pro hardware is popular among pro and creative users.
Sure, and OS X is very popular among creative users and pros. So there is a large base of users there.
I think it's overblown. The Mac is still Adobe's premier platform, and they've put out a good amount of money porting over the remaining missing apps in their suite.
In addition, Apple is allowing Flash apps on the iPhone now. It's all really just down to the web plugin.
1999? Macs were very unpopular then, but a lot changed when Apple adopted UNIX.
University of Portland was the school which was putting Mac Pros on professors desks. I've seen a large number of Mac Pros at Portland State as well, especially in engineering. The high performance labs are still mostly Linux boxes, but I've seen Mac Pros there too.
My take is based on my education in computer science, in the engineering department.
I long ago gave up on them taking servers seriously, but again, this doesn't change that they have certain proprietary protocols that can only be served from a Mac Server.
No, I think they've totally abandon trying to tackle serving services of that scale. But honestly, the Mac Pro and the XServe were never hardware for those sorts of installations to begin with. I don't think Apple ever intended to go there. iTunes, for example, was for the most part not served on XServes.
Again, this all sounds well and good until you run into the "Apple has proprietary protocols and services for Mac and iOS" wall. Apple can't entirely leave the server market.
No, they don't... Read above. iCloud is a consumer replacement, but it doesn't solve what large organizations need for Mac and iOS.
How much Mac server work have you done? It sounds like you don't have much experience.
iCloud isn't even running OS X. Most of it is actually hosted off of Microsoft server. None of which is doing the custom proprietary stuff I mentioned earlier.
Unless they were planning on re-investing in the Mac Pro. Eventually the chip could also be brought down to the iMac.
I'm pretty sure they already said they'd be willing to do Apple specific chips if asked.
Please cite said information. Or stop saying that the Mac Pro is unprofitable.
Again, the API's built for Final Cut Pro X are publicly available. Please start citing. I'm looking at the APIs and seeing a significant amount of work, and I know the app coded against these APIs is not trivial.
You need to start backing this up with evidence.
If it's pre-OS X, it might as well have been an entire lifetime.
Last edited:
Hey nanofrog and goMac, post some photos of your keyboards, I'm sure that they're wearing out with this kind of threads, LOL.
P.D.: However, I'm reading all of them, seems quite interesting
P.D.: However, I'm reading all of them, seems quite interesting