they made it fail with lackluster updates. Mp was very popular in the early days
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
they made it fail with lackluster updates. Mp was very popular in the early days
Of course there probably aren't any numbers to back this up but the theory is that although the Mac Pro was doing well early on it was not doing so well later as it became more expensive. And it was the expensiveness that drove down demand and sales leading Apple to start neglecting it. Vicious cycle. This is at least plausible.
My 2c..
Every pro desktop user not served by the iMac or Mac Mini needs the following in a machine:
CPU & RAM
Boot Drive
Graphics
Networking
Thunderbolt USB & Bluetooth
Marketing line: Every thing needed by every pro.
That's all that every user needs. Some need multiple drives, some need optical drives, some need multiple PCI cards, and Thunderbolt is capable for all those things. The only thing it can't do, due to a lack of PCI lanes, is externally hosted graphics cards.
It makes sense then that the next Pro Mac really only needs PCI slots for as many graphics cards as are expected to be used.
This style of machine would only need a single cable to connect it to a chassis that has drives and PCI slots - that could live under a desk or in a rack mount, and then when the Compute block needs upgrading, unplug 1 cable, plug the new machine in, and get back to work. It's a much more elegant solution than gutting your old Mac Pro to transfer all the cards etc across to the new one.
Would Apple make the expansion chassis? Doubtful, unless there was a way for it to lock to the compute block. I suspect it's similar to the xServe RAID situation.
What's in it for Apple? Much lower manufacturing, shipping & storage expenses. Effectively 100% coverage of the "I want a machine with pro performance without a built in monitor" crowd who might otherwise under-spend on a Mac Mini. It lets Apple concentrate their R&D opportunity cost on the cheapest solution with the widest applicability.
Every pro desktop user not served by the iMac or Mac Mini needs the following in a machine:
CPU & RAM
Boot Drive
Graphics
Networking
Thunderbolt USB & Bluetooth
Marketing line: Every thing needed by every pro.
That's all that every user needs. Some need multiple drives, some need optical drives, some need multiple PCI cards, and Thunderbolt is capable for all those things. The only thing it can't do, due to a lack of PCI lanes, is externally hosted graphics cards.
It makes sense then that the next Pro Mac really only needs PCI slots for as many graphics cards as are expected to be used.
This style of machine would only need a single cable to connect it to a chassis that has drives and PCI slots - that could live under a desk or in a rack mount, and then when the Compute block needs upgrading, unplug 1 cable, plug the new machine in, and get back to work. It's a much more elegant solution than gutting your old Mac Pro to transfer all the cards etc across to the new one.
Would Apple make the expansion chassis? Doubtful, unless there was a way for it to lock to the compute block. I suspect it's similar to the xServe RAID situation.
What's in it for Apple? Much lower manufacturing, shipping & storage expenses. Effectively 100% coverage of the "I want a machine with pro performance without a built in monitor" crowd who might otherwise under-spend on a Mac Mini. It lets Apple concentrate their R&D opportunity cost on the cheapest solution with the widest applicability.
Is it really that hard to make a computer where everything you need fits "in" the computer? It's now elegant quiet and contains everything..
I could see a switch to 2.5" drives, cutting an optical drive, maybe even an mSATA drive on the board for boot. But pieces you assemble, PCI box, Processor Box, Drive enclosure? What glaring problem does this solve? More to the point what is the MP's glaring design problem right now? What are you guys/gals fixing?
I could see a switch to 2.5" drives, cutting an optical drive, maybe even an mSATA drive on the board for boot. But pieces you assemble, PCI box, Processor Box, Drive enclosure? What glaring problem does this solve? More to the point what is the MP's glaring design problem right now? What are you guys/gals fixing?
I'm of the opinion that we still need something between the iMac and MacPro. Maybe the same expendability as the pro but with a common i serie CPU instead of an Ex Xeon. Call it a semi pro Mac.
I need multi-GPU for rendering and cuda based encoding but I don't have the extra money for those Xeon that cost 4x as much as an i7 3770 that would fit my need CPU wise.
I need multi-GPU for rendering and cuda based encoding but I don't have the extra money for those Xeon that cost 4x as much as an i7 3770 that would fit my need CPU wise.
For one, because expandable doesn't equal upgradable. There is nothing this rumors description that does excludes at least one of the GPUs being on a standard PCI-e card.
Expandable is increasing in number (e.g., going from 2 to 3 disks or 1 to two PCI-e cards inserted, etc. ). It is not a replacement metric.
The rumor outlines dual ( more than one ) GPUs. More than two in a current Mac Pro ? There is only two x16 slots. The others are x4.
Redrocket can be delivered via Thunderbolt.
http://www.sonnettech.com/news/pr2011/pr092211_redrocket.html
Likewise, RAID streams can be delivered over TB.
TB is same speed in both directions so if stream in on the rocket card and stream back out on the external RAID it isn't a problem.
It is quite likely it will have the bandwidth the current Mac Pros have, but it would just be a form factor change. There will be substantially bigger boxes you can buy where you can pack 'everything' into a single box, but that has always been the case.
Why a workstation would rationally drive a phone purchase I have no idea. The vast majority of iPhone (and iPod ) users don't own Macs.
They're fixing problems that exist on Apple's end.
Building a machine that could deal with everything any pro could possibly need internally drives up the price of the machine for options that most people don't use.
I'd bet 80% of Mac Pro users never install another drive, and that 95% never install drives into bays 3 and 4. I'd also bet 95% of users never install an additional PCI-E card.
From Apple's perspective, the Mac Pro is probably overly weighed down with expenses to try and target that 5%. It's just easier to let manufacturers make whatever Thunderbolt products are needed for those niche users instead of making everyone pay for things they won't use.
I've seen a lot of G5s (and more and more Mac Pros) that stay entirely stock from the day they ship to the day they hit the scrap heap. Aside from maybe some more RAM, not that many people are actually taking advantage of the internal expansion.
You know what I'd bet? That those numbers are not even close to reality. I find it very unlikely that someone spends $2500+ on a machine, who's main selling point over something $1500 is that you CAN add HDDs or PCI-E cards, but doesn't use those features.
I'd bet the exact opposite, and that you'd see the majority of Mac Pros out there have at least one user added part.
I've worked on a lot of Mac towers and most the ones that I've seen were stock. People buy them because they are fast or because they will last a long time. RAM upgrades aren't too uncommon, but I don't think I've ever seen a GPU upgrade except on my tower at home. Second hard drives are uncommon, but not rare, but third and fourth hard drives are extremely uncommon.
A lot of Mac Pros don't go to individual power users. They go to labs (schools and universities), they go to peoples desks at work, and they go to people's homes, where they sit unupgraded until it's time for a new machine.
Especially if you're a business that has network storage, you're not investing in more drive's for your employees Mac Pros. It's safer to have data in your local cloud with backups and faster access speeds than it is to have it sitting on your drive.
Having a local machine full of drives locally is probably going to be an antiquated way of doing things long term, especially with laptops becoming more popular.
User added or user replaced part ?
Gobs of folks swap out the 3 1GB DIMMs for 3 2 or 4 GB DIMMs. Likewise on the single storage drive. Likewise on dumping the Apple ODD for a newer Blu-Ray, faster ODD, or working ODD. Likewise dumping 3 year old GPU for new one. None of those is a device expansion.
I suspect 80 and 90 percent is high but it probably isn't the inverse either. Although, 2 drives minimal is also more likely a average minimal number. Again that doesn't necessarily mean the fixed number in this rumored updated doesn't match that number.
If Apple hasn't been completely incompetent and have done significant field configuration measurement they'll know better than any user what the median configuration numbers are.
SSDs nuke much of the "short stroking" that commonly inflated drive counts ( e.g., "short stroke" 3-4 HDD in RAID 0 to juice higher IOPs rates. ). You have to remove items like that to get an accurate projection for future needs. The solution to the low IOPs is to get rid of rotational drives not add more. The 'stopgap" solution was always suboptimal. It adds more of the problem to get around the problem. No way that is ever going to lead to a true optimal.
Very similar issue when GPU PCI/PCI-e cards could only drive 1-2 monitors. The card count is high, but really the suboptimal equipment limitations that is driving the higher card count.
Yep, I teach college motion graphics classes and there are labs full of stock MacPros. I've been to a couple other universities in town and same deal there. Mostly just RAM upgrades. All the drive bays and PCI slots are wasted there and they buy new ones every 3 years.
It's not ideal but I don't see how a 12/16/20 core XEON machine with SSD, 128GB of RAM and 2 Titan GPUs attached to a TB2.0 enclosure full of 10Gbe, RED, I/O cards and a RAID is so so incapable of professional work.
I don't like all the cables but I can certainly work with it, especially if it's fast.
It's not ideal but I don't see how a 12/16/20 core XEON machine with SSD, 128GB of RAM and 2 Titan GPUs attached to a TB2.0 enclosure full of 10Gbe, RED, I/O cards and a RAID is so so incapable of professional work.
I don't like all the cables but I can certainly work with it, especially if it's fast.
There is also the motivation that Thunderbolt solutions are useful across the whole mac line up. It can be attached to a MBA 11" or a Mac Pro. Thunderbolt to a large extent is just repacking what would have gone on a PCI-e x1 , x2 , or x4 card into an external box with the same set of interfaces balloon squeezed onto that other box (and probably an additional power supply).
Apple can move those niche users no to any machine in the Mac line up if that machine lines up with their workload.
When the niche user was coupled to a PCI-e card.... before lots of folks "had to" buy a Mac Pro to make it work. That leads to those users paying inflated Mac costs and very likely leads to them not really leveraging the expansion aspects all that much.... because they never needed them in the first place ( Mac Pro was wrong box in that respect and only "right box" since practically only PCI-e solution. ) [ ExpressCard next to Thunderbolt is a joke. That always was a cheesy kludge of USB and a single wimpy PCI link. ]
I suspect though the sampling is different if bias to just the top end single package model and the whole of the dual package end of the line up. Or perhaps more simply around $3,400+ versus < $3,400.
If the single and dual package groups are significantly different is usage patterns then it would make sense for Apple split that into two different boxes. I'm not sure the volumes and growth for each split group will work afterwards but status quo path isn't working either.
Last edited:
I think goMac is exactly right here - I wouldn't be surprised if most MacPros have never been expanded, except for RAM and maybe another video card. With respect to those that upgrade the heck out of these machines (I'm looking to be one of them as soon as we know what's happening next week), making a "slim" MacPro with some limited expansion maybe just the right redesign. As long as there's an elegant solution to expansion via modular, stackable, power-sharing units, along with a matching price-adjustment, I'd be a buyer. What I'd hate to see is a MacMini on steroids with expansion through a series of third-party units taking up desk and power-strip space with wire-spaghetti everywhere.
Rob
It doesn't have to be alot of cables. But I don't think TB 2.0 is going really solve the problem of condensing it all down into one enclosure. Thunderbolt 2.0 seems to be just reshuffling the deck chairs and not really a real throughput increase.
" ... It is achieved by combining the two previously independent 10Gbs channels into one 20Gbs bi-directional channel that supports data and/or display. ... "
http://blogs.intel.com/technology/2...ndwidth-enabling-4k-video-transfer-display-2/
If the TB controller still has a x4 PCI-e v2.0 connection to the personal computer device all they have done is reshuffle the transport bandwidth. The individual system throughput will be almost exactly the same.
The big change here is that "very big video + lightweight data" and " verg big data + mainstream video " users will get more bandwidth. However, mega big users of both are likely to bump into bandwidth cap limitations.
So this will work as long as that is "normal" capture 4K but just primarily to oversample and work with something smaller. Given the short term costs of 4K monitors that will probably work just fine for very large group of folks.
And should be easy to take 4K video off Thunderbolt network if really need the data bandwidth (i.e., just use a normal Display Port 1.2 link on a GPU PCI-e card edge. )
I was thinking about holding off on a new Macbook Pro this year because of the upcoming Thunderbolt upgrade, but that's not really that big a deal...
A Mac Pro in is current form does not preclude an xMac. Apple has shown no desire to create an xMac in the dozen years though
Depending on what you do, the very reason you work on them may be the reason they haven't been expanded.
It sounds like you're working on Mac Pros for people that don't really need Mac Pros. Again this is a guess, but I'd bet that's not very good pool of people to repressent the whole of Mac Pros. I've seen a lot in professional settings that have something like 64 GB of RAM, 3 or 4 HDDs, with RAID 1 or 0. GPU upgrades are less common, but its not exactly rare to see some sort of PCI-E card.
Currently I'm in academic research, and I can tell you most of the Mac Pros I've seen are pretty decked out by the user or university IT personal. The computers that sit in teaching labs or class rooms are iMacs or PCs, not Mac Pros. I've also had some experience in industry, and many Mac Pros I've come across are similarly upgraded. Management usually doesn't spring for the extra $1000+ for a Mac Pro, unless its really justified. And these are for corporations that may buy 100s to 1000s at a time.
Depends on the specifics of what's being done. Some types of work don't "share" well. Others do, but upper management is stupid. Some get off loaded to individual machines until a project is ready to "mint" in the central network.
True-ish. Many people now do all there work on a cluster (or in the cloud, if you like), and only need some sort of local storage solution for the raw data and the end product. So they can work on a laptop, if they want, and have a NAS + cluster access (thinking academia mostly here or smaller businesses). The weak link is the badwidth to move things around outside your local network and queue times once its there. That puts many people in a situation to at least get some work done locally on a workstation (or even mini-cluster). And I don't think this is a situation that is going to end any time soon.
But that's exactly my point. Apple wants to pair down the features to what everybody needs, and then whatever more niche uses people need is pushed off to Thunderbolt.
If you need more local storage, you get to decide that instead of Apple making an assumption at design time.
Where are you? Most departments I know haven't had that kind of money to throw around since at least 2007/08, or even the 90's. Replacing mostly idle $2500-$3000 machines every 3 years.... Hopefully this isn't a public institution.
----------
I'm not disagreeing with that point. "Most people" could probably be servered with just these kinds of solutions and it would save Apple a lot of money, while not sacrificing many sales.
I'm just saying I think the "80% of Mac Pros are stock(/unexpanded)" is bogus. At least in my experience.
It depends upon what it means for xMac. The "I don't like iMac so give me xMac" then no. Apple has little desire to just shuffle the deck chairs in the desktop market. Swap large blocks of iMac users for large blocks of xMac users. ( yeah there are folks who aren't in the intersection but the intersection is much larger than any one of those two subsets. )
I don't think it is $900-1,600 xMac as much as the 1,899-2,499 segment they have gradually given up as traded that range subset with the iMac. They did sell a "box with slots" into that range before during the last 12 years. The name "Power Mac" and "Mac Pro" was used in that space before. They may need a new name (or possibly old name + new suffix ) to cover just that range subset. if "x" stands for doesn't have a name yet, then it is a fit.
The open question how do they go back in and readjust allocating that range with the iMac folks. Given the whole range to the iMac was bad idea long term for both iMac and Mac Pro. (The Mac Pro has had and is having problems , but the iMac is only a couple of years away from running into the same buzzsaw of overall industry trends in that same price range. )
I still don't think so... Not many Mac Pros are going to specialty industries like yours.
I'd agree that maybe a lot of those people don't need Mac Pros. But if you want a machine with one or more external displays with a nice GPU, you don't have many choices from Apple these days.