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.
Any chance for a smaller Mac Pro this year?
- Thread starter Radiating
- Start date
- Sort by reaction score
Some folks just want control not necessarily expandability. A box in which they can take some components out and put others in. How many or being able to put it all into one box isn't high on their value system.
But yes most of the modularity ideas are deeply flawed in where the "cuts" are made and insert bottlenecks or simply just increase overall system costs.
most of the calls for a 'smaller' box are just misdirection. it is really clamour for a less expensive box. By jettisoning parts of the box the implicit notion is that costs would go down.
If there is only one PCi-e slot there wouldn't be a high motivation for using an Xeon E5. A Xeon E3 processor could be used ( or mainstream desktop ). Given the lower thermal envelope the power and thermal management can be made smaller.
Could Apple do an Mac Pro "smaller" :
Xeon E3
16GB RAM
2 socket Thunderbolt
one PCI-e slot (with default card from bigger Mac Pro line-up)
dual 10GbE sockets.
no ODD
2.5" SSD options only for internal storage
and slide it into the $2099-2499 price zone? Yes, but they probably won't.
( It could be $1,899 or so without the PCI-e GPU card as BTO).
It would be capped at 4 cores (like the desktop/lower power options are these days). It would be "good enough" for a decent fraction of the xMac crowd that aren't primarily constrained on price. Apple can add value ( 10Gbe and SSDs ) to keep it out of the iMac's assigned price zone but the open question would be if there was enough of a groundswell available to carry both this and the single processor package Mac Pro. If "cheaper" ($800-1300) is all folks really care about then a smaller version is of dubious value to Apple.
What is needed is a "tinker tools" Thunderbolt box. A box that isn't too expensive that has a 5.25" bay , 1-2 3.25" bays , 1-2 2.5" bays , and perhaps one full size PCI-e card slot. This box would help the folks who want to open the lid and poke around with their trusty screwdriver. They can get their "control" needs satisfied even while using a mini or iMac.
Another response mentioned that a MBP would work while doing limited edits mobile. if folks wanted to drag along a larger external screen anyway a Mac Mini could play the role in the future too. Coupled to one of these "tinker" boxes folks and some TB aware driver support for graphics and it would have much of the modularity being clamored for. The only "problem" is that much of that already exists.
The modular arguments miss the point that the general trend for digital computers for last 60+ years has been smaller. Modularity has much more traction when the "computer" is large. Once smaller certain core functions CPU+GPU+Memory+OS/Apps storage all make much more sense to be part of the core device.
External 'working space', external files are fine as long as have high bandwidth, low latency connection.
Maybe smaller? But no chance for modular stackable. There just isn't any sort of connector that is able to handle that sort of bandwidth. And yes, I'm including Thunderbolt.
If Apple went the non-dual processor box route I'm not sure they'd stick with Xeon at all.
If Apple went the non-dual processor box route I'm not sure they'd stick with Xeon at all.
Agreed; even if TB would offer the bandwidth, it's still dead in the water right now .
Modular would really be sweet, though, but not in the way the usual cheap-Mac-tower-crowd wants it .
A top of the line base, with lots of high-speed slots, whatever tech those might be, and no low-cost version .
And then, affordable extension modules and cheap port extensions .
No additional fans, no power bricks clutter .
It's not going to happen in our lifetime; until then, the MacPro is a tad on the small side, it could use a couple more PCI slots and harddrive ports .
The real problem is that what Thunderbolt makes "modular" we already have. External drives, monitors, and RAID arrays. There isn't much of a reason Apple needs to do more. You can already buy a Mac Mini and plug drives and monitors into it.
If you want a desk full of external boxes the solution is already here. If you want to to be able to stack your external boxes, the solution is already here.
That would be dubious because that essentially caps the x86 core count at 4. The much smaller subset of core i7 desktop CPU package offerings that are derived from Xeon E5 designs aren't cheaper and really only differentiate themselves for the overclocker/tweaker crowd. I'd be shocked if Apple was going to switch gears and move in their direction. That seems highly unlikely. If Intel dropped the E5 1600 line up perhaps but as long as it exists and reasonably populated (i.e., not what Intel did with the Xeon 3600 series that was just one model for a substantive amount of time) Apple will likely continue with it.
Throw on top the fact that by Broadwell there aren't going to be any socket solutions for that desktop oriented space (about all BGA solutions). Sure, Apple is tracking to the solder more to the mainboard track in the other Macs, but they really don't buy much at all by tracking all of the Macs into that space for everything. They aren't going to be able to track the folks who do have applications that scale. Even if dump most of the "embarrassingly parallel" float computations off to GPGPU on a PCI-e card there are still going to be a small number of scalar critical sections for some workloads. On those, the larger thermal envelope will make a substantive difference.
The "soldered to the board" factor can just be the discrete GPU for the Mac Pro.
I'm not sure that Intel is going to grow the single socket targeted packages past very much past 6-8 cores. Seems more likely they will use additional transistors at some point to add an integrated GPU that can attack massively parallel float problems directly and with more efficient use of transistors. Right now HD4000-HD5000 OpenCL performance isn't much to write home about, but by Broadwell that should change. But yes if Intel keeps the whole E5 line up on a track where there are no iGPUs then eventually over time I can see Apple jumping off the track. It doesn't have to do with dual vs. single package set ups though.
Mmmm... If there is only one PCIe slot in the main box there would of course have to be a PCIe extension and another separate box for more cards to accommodate those users needing more slots. With thunderbolt already being a PCIe extension this wouldn't be very unexpected.
And for those thinking there is no connection fast enough PCIe extension boxes are already in existence and work just fine - TB or no... The PCIe specification v1.0 was designed with this in mind actually - and of course it's been updated with each revision as well.
Here's an article from 2007:http://www.rtcmagazine.com/articles/view/100910
Last edited:
They've just got to get a faster revision of Thunderbolt. The bottlenecks in video work are so frustrating.
Hehe ...
As for thunderbolt, daisy-chaining the few available devices to a single port or two, and paying a premium for mediocre products isn't exactly my idea of a proper modular system .
But what do I know ?
What I do know, any current Apple monitor won't make it into my workflow anytime soon, nor any of the available harddrive enclosures .
Of which there are 2, or 3 ? It could be a hundred, yet I still couldn't run a bunch of portable FW or USB harddrives off a decent hub - of which there are none .
With OWC, Sonnet and Magma TB->PCI-e cases you can use many existing enclosures from CalDigit, Sonnet, etc. using SAS cards and things of that nature. I have a Magma box loaded up and there are way more Thunderbolt choices than I thought. And it plays nice with my NEC monitor DP monitor.
Not perfect, but again, leaning on my MBP until we find out if another MP is going to happen or not.
Not perfect, but again, leaning on my MBP until we find out if another MP is going to happen or not.
Yeah, they probably wouldn't actually use thunderbolt... It's just that TB is already a PCIe extension technology which Apple is very involved with. So the know-how and familiarity is already in place if they decided to go that way.
I'm not saying they would, wouldn't, should or shouldn't as that would be speculation bordering stupidity. But they could if they wanted... And even offer more better faster than the current MacPros.. How would you like to see a breakout box with eight Tesla cards in it all operating at full speeds and yet not introducing any heat into the main system.
Totally doable.Yeah, the external GPU dream is a nice one.
I just am bummed the TB bandwidth limit lands right in the zone where RAIDs start becoming REALLY useful for heavy lifting.
Reminds me of how awesome I thought FW400 was when it took off in 1999. I bought a 4-port card for my PC for a mere $800, and then my AGP G4 came with one. Ugh...
I just am bummed the TB bandwidth limit lands right in the zone where RAIDs start becoming REALLY useful for heavy lifting.
Reminds me of how awesome I thought FW400 was when it took off in 1999. I bought a 4-port card for my PC for a mere $800, and then my AGP G4 came with one. Ugh...
There is either a Market for folks that need primarily just one slot (e.g., only a GPU PCI-e card that is later upgraded to get a performance boost in a couple of years) or there isn't.
Expansion boxes for PCI-e card slots aren't going to be cost effective way of approaching a market where the norm is 2-3 cards. That makes them uncompetitive. Especially in the context where there are a wide variety of options that do have 4 slots. 4 is a reasonable number to capture the vast majority of those that need more than one. The ones that are left over can go the more expensive option off of a x16 slot if necessary. That is far superior to a x4 v2.0 capped solution like Thunderbolt if the target is 6-8 slot usage of significant bandwidth.
So far TB is having trouble delivering on the cost effective solution. It has value where there are no options ( zero PCI-e slots like a Mini or MBP ), but are far more dubious once there is one.
The majority of PCI-e card topics on this forum revolve around GPU cards. That is highly indicative of where the singular card value market would be.
it is trivial to design a modified Mac Pro that is non-competitive and will only attract a narrow sub-niche market. That completely ignores the major problem the Mac pro faces of limited growth track record. A new product with no long term above average growth potential won't get off the ground.
It's an epic fail. The first version (never seen one though), was beautiful, made of unobtanium, and was priced accordingly. Only problem was that it got too hot with the separate closed boxes for the different components, and was never updated to include basics like USB3 ports. I bought one of the second version, and as has already been mentioned, it is a piece of expensive, but cheaply made plastic crap. It's still heavy, massive, and the plastic and redesign take much away from what was originally a stunning looking case.
The Thermaldesign really does show why it's worth paying the extra for what Apple produce. It sucks because you know you are paying dollars on the cent for their R+D, but then that's why they are pretty much the only tech company who don't have to worry about their bottom line at the moment.
I have a 1,1 Pro that I don't use any more, but you only have to look inside to see that it is a work of art - far more beautiful inside than out. Separate trays for the HDDs with screws already attached so you can mount the drive outside then just slide it in. PCIe slots that don't require any tools at all. RAM holders that slide out as well. Everything in its own thermal zone. Everything made out of aluminium so nothing flexes or feels cheap.
The Mac Pro that we have at the moment may well be the best designed computer in history. I have no real need for one, but if the 2013 version doesn't disappoint, I'll be getting one straight away.
David
I really do not think they NEED such a large box to keep dual processor support.
It was the case back in the G5 days, but those G5's has an enormous amountof heat to dissipate, put two of them in a case and you end up devoting more than half of the case to the cooling solution.
Today's Xeons are not the same lava spewing hair dryers that the G5s were.
Today's video cards and hard drives are not nearly as hot either.
I also think that while nerds like me have plenty of uses for every single connectivity option on a current Mac Pro, most folks never install any cards or use any of the drive bays.
Besides, there was NO REASON to make the iMac thinner, but they did anyway.
It was the case back in the G5 days, but those G5's has an enormous amountof heat to dissipate, put two of them in a case and you end up devoting more than half of the case to the cooling solution.
Today's Xeons are not the same lava spewing hair dryers that the G5s were.
Today's video cards and hard drives are not nearly as hot either.
I also think that while nerds like me have plenty of uses for every single connectivity option on a current Mac Pro, most folks never install any cards or use any of the drive bays.
Besides, there was NO REASON to make the iMac thinner, but they did anyway.
I have to question that, even though Apple might have the same opinion. All workstations are niche in terms of the whole computer market, but as a halo product it makes perfect sense for them, even if sales aren't skyrocketing. I can think of at least three reasons:
1. If someone has a MP for work, I bet it makes them far more likely to have an iPhone, iPad and MBP than almost anyone else.
2. Seeing a MP being used for cool stuff like editing the latest TV show on a 'behind the scenes' extra on a DVD must be worth a hundred sales of iMacs or MBPs.
3. The MP doesn't need a redesign. Granted, a refresh would be good, but it is already an excellent case. Dock Jony Ive a few days pay for having 'too short a haircut' or 'not spelling Johnny correctly', and you have enough money to adapt an intel reference mobo.
David
On a laptop, it would be great. But why on a desktop?
While this is true, if you pop open a Mac Pro case, you'll see that the leftover space has been given to other things.
Apple isn't wasting a lot of space in a Mac Pro case. It's hard to see how the case could be made smaller without sacrificing features. It's still smaller than comparable workstations from Dell or HP.
There are a few things that could be changed to reduce space, like 3.5" drives for 2.5", but to get dramatically smaller you have to start ditching features.
All speculative made-up tripe that doesn't add clarify or define anything - just totally speculative, made-up, and meaningless. Why?
Last edited:
Much smaller is unlikely, but some adjustments would be nice
I think we're unlikely to see a very much smaller Mac Pro, but I think it's not entirely unlikely that they'll remove the handles, so that it's low enough to be rack mounted sideways. It's currently just the handles that's keeping this from being possible.
Making the Mac Pro rackmountable would make sense given the discontinuation of the XServe.
It's also possible that with the likely removal of the optical drive, they could make it 3 rack units (133 mm) wide, instead of the current slightly over 4 units, although that would require mounting the hard drives rotated 90 degrees from what they are now (which is more like what most servers do). If they don't do that, I'd expect them to at least make it an even 4 units (177 mm instead of the current 206).
I think we're unlikely to see a very much smaller Mac Pro, but I think it's not entirely unlikely that they'll remove the handles, so that it's low enough to be rack mounted sideways. It's currently just the handles that's keeping this from being possible.
Making the Mac Pro rackmountable would make sense given the discontinuation of the XServe.
It's also possible that with the likely removal of the optical drive, they could make it 3 rack units (133 mm) wide, instead of the current slightly over 4 units, although that would require mounting the hard drives rotated 90 degrees from what they are now (which is more like what most servers do). If they don't do that, I'd expect them to at least make it an even 4 units (177 mm instead of the current 206).
Well, let's see, maybe because a nice GPU setup costs over $10K and it would be nice to move from computer to computer as needed, just like I do my RAIDs, capture cards and other I/O cards with my Magma TB enclosure.
if you've paid attention to the insides of the Mac pro since it has debuted, they have made very radical changes to the Mac Pro many times. It just seems like it has been the same because the external form factor has stayed pretty much the same. Even when they changed the power supply location and moved all the fans around the box stayed the same.
Comparing the size of the Mac Pro to workstations form HP and Dell isn't really fair. The Z800's and the like have space for a dozen hard drives and quite a few expansion cards. They also support many more legacy ports like serial and PS2 that Apple does not.
The main thing that makes Apples attractive as a workstation isnt it's hardware capability, HP and Dell blow it out of the water.
It's the workflow of the Mac OS which many creative professionals are used to.
You can question, but the questions would have to be grounded in some good rationale to have any impact.
Every sub-market of the whole computer market is smaller than the overall market. Workstation's percentage of the market isn't the issue. How much it grows is.
The Mac Pro way to expensive to be a practical halo product. A product you buy because bought another more highly valued product.
Depends upon what that person does for work and where they do most of there work. Folks billing others while moving around can probably do more billable activities with those 3 devices than a Mac Pro.
Even if stationary this is not a halo effect property.
Apple should build it because it is "cool". Not really. The Cube was a bust. The "20 anniversary Mac" was a bust. None of these "crotch grabbing" , isn't that way more cool than it is useful products really save Apple or any other company.
Products that folks lust after but can't buy do not produce much of a halo effect as much of the handwaving about that purports it does.
It does need a redesign because several aspects are based on dated technologies and thinking.
a. lacking 2.5" bays at this point is wrong. Back in the 2002-2005 era 2.5" disks were only "laptop" disks. Now they are not.
b. Similarly two 5.25" bays? Why? ( looking forward over next 5-10 years is that looking like a growing segment). In a poll of real Mac Pro usage what is the ratio of folks would hacked a 2.5" additional versus those who slid in a second 5.25" drive later on? Closely related how many folks completely removed the ODD drives to pack in more storage devices they actively use everyday?
c. Why are the handles gratuitously rack hostile. A large number of competing, similarly equipped HP, Dell, Lenovo, etc. workstations all fit in a standard rack in a horizontal position.
d. Friendly to MIMO wireless antenna designs? Not really.
e. Friendly to 250+W GPGPU cards? Not really. ( computation 'power' of the GPU is about as import to a workstation as the CPU packages ).
f. Thunderbolts pragmatic requirement for an embedded GPU will impact the internal layout. ( not technically necessary but if Apple is OCD about uniformly deploying TB then a case design issue for Mac Pro none-the-less. )
Completely redone to be totally foreign to the current design? No. But it is dated on functionality.
It is a secondary, marginally expansive role though. It would still primarily be a vertically oriented product in nominal usage. It would help to get higher utility by being more easily rackable, but it is a sell some more issue.
Making a vertical , heavier (with density relatively evenly distributed ) object thinner doesn't make alot of design sense. Given the height and weight it would be more prone to tip-over is shrink the width. Being rack mounted is not a primary target market. Most of them will stand up. The redesign shouldn't sacrifice utility as a vertically oriented product.
The major current problem right now is not being marginally over 4U 'wide'. It is being too tall ( wide in horizontal orientation) to fit in a standard rack. That is a much bigger issue than tap dancing between 3 and 4 U. Just as tall and exactly 4U wide doesn't help much. For the 1U XServe die-hards anything over 1-2U is still going to be a "grumble and moan" issue. The current 7U vertical orientation could save "2U" ( like dropping 5 -> 3 ) just be readjusting the container and the handles so that fit horizontally (e.g., increasing depth a bit to shrink height about 2" . )
The fans/cooling and associated noise constraints are just as high an design constraint than the current drive mounting. I don't think that will solve the problem.
Whether it is exactly 4U matter only if trying to pack a rack or half rack to the brim with the maximum amount of equipment. Most likely racking one or two ( in half rack and full rack respectively) Mac Pros will cover the vast majority of likely rack deployments. 1-2 Mac Pros is enough to run a small-medium business from a server standpoint. A
There isn't likely going to be large computational grids of Mac Pros that would require dense packing of a rack. If just want to crunch numbers in a rack in a noisy machine room there are much more cost effective ways of doing that than with a Mac Pro. Going "skinny" trying to crack that market isn't going to buy much of anything.
I think while all these are probably all good guesses as to changes on the new Mac Pro, none of them really produce the "mini" Mac Pro a lot of people are hoping for. It would get the Mac Pro closer to mid tower range, but the changes wouldn't be that dramatic.
I'm mixed on the handles. They're pretty handy (hah) if you've ever had the experience of lugging around a Dell workstation.
Here in Austin, the 2 Apple Stores are both awesomely inconvenient to get to. The handles are EXTREMELY handy and make a terrible experience just a little less terrible.