New Mac Pro
Hi guys,
here are some thoughts for discussion for the next Mac Pro:
- general concept: a radically new design that will redefine the Mac Pro
- no optical drive bays
- built-in flash memory for OS and Apps
- 3 additional 3.5" drive-bays (with hot-swap?)
- no slots, Thunderbolt will make up for this
- way smaller case
- lower price point
- release: October 2011 (maybe at a special Mac event?)
As with Final Cut Pro X, professionals will cry out foul!
Thoughts?
here are some thoughts for discussion for the next Mac Pro:
- general concept: a radically new design that will redefine the Mac Pro
- no optical drive bays
- built-in flash memory for OS and Apps
- 3 additional 3.5" drive-bays (with hot-swap?)
- no slots, Thunderbolt will make up for this
- way smaller case
- lower price point
- release: October 2011 (maybe at a special Mac event?)
As with Final Cut Pro X, professionals will cry out foul!
Thoughts?
Comments
Now, to business.
- general concept: a radically new design that will redefine the Mac Pro
I saw it then, but I was a fool. Keeping the case the same was the best idea. But that was before Apple abandoned the servers and most of their highest-end stuff. So who's to know this time around?
- no optical drive bays
OH PLEASE YES. But again, it's too early for that, I think. Even though anyone using a Mac Pro for high-end use will have towers of optical drive copiers that have robotic arms that can automatically burn thousands of discs without input from the user.
- built-in flash memory for OS and Apps
Way too soon for this, still. I think even I mentioned it back in late aught eight and early aught nine waiting for the Nehalem Mac Pro. I was even more early than you.
- 3 additional 3.5" drive-bays (with hot-swap?)
That'd be nice (and hot-swapping would be REQUIRED for the thing to be considered any sort of server replacement), but it's contingent on a new case design, so we'll see.
- no slots, Thunderbolt will make up for this
Oh, now you're just playing with us.
- way smaller case
I'd much rather have a LARGER case, you know? LARGER. Smaller just screams, "screw you, pros; you don't get to expand this thing at all."
And a larger case would allow for the full 12 RAM slots allowed in the chipset and create more room for the extra HDD bays.
- lower price point
I don't think this will ever happen again. The days of the $1,400 tower Mac are long, LONG gone.
- release: October 2012 (maybe at a special Mac event?)
If Apple kept the current models for the rest of this year and most of next, there wouldn't be anyone left to sell Mac Pros to. This is complete insanity.
Ivy Bridge server chips'll be out late Q1 2012, so that's likely when it's up for refresh.
OH PLEASE YES. But again, it's too early for that, I think.
They might leave it in but iDVD is not on the App Store, DVD Studio Pro is gone, the Minis have no optical, the mainstream Mac laptop has no optical. This might be the start of a trend.
It would immediately wipe $200 off the cost of the machine and shrink it down in height by about 1/5th. Less engineering, less wiring, less metal, less weight.
I'd much rather have a LARGER case, you know? LARGER. Smaller just screams, "screw you, pros; you don't get to expand this thing at all."
Smaller is also a way to change the perception that expansion is always a requirement for any work that uses the performance of a $2500+ machine. The quad-core Mini is already doing this:
Full Xeon chips, MXM graphics cards, 4 x Thunderbolt and a very small enclosure would be nice.
If Apple kept the current models for the rest of this year and most of next, there wouldn't be anyone left to sell Mac Pros to. This is complete insanity.
Ivy Bridge server chips'll be out late Q1 2012, so that's likely when it's up for refresh.
The roadmap says the Ivy Bridge EP chips will be out same time next year:
http://pc.watch.impress.co.jp/img/pc...tml/1.jpg.html
I reckon the Mac Pro will get a Sandy Bridge Xeon refresh in October.
Hi guys,
here are some thoughts for discussion for the next Mac Pro:
- general concept: a radically new design that will redefine the Mac Pro
I actually believe this is a requirement. The Pros form factor is about as handy as a Prime computer from the seventies. The case reflects thinking from decades ago.
- no optical drive bays
This is a tough one. I have no need for such a bay but I know many do. In the end I think a more radical modular approach is required.
- built-in flash memory for OS and Apps
Built in absolutely not. A SSD or a Blade SSD yes. For Apple I think SSDs are a huge win as the OS really benefits from the speed. Thus they are supporting SSDs everywhere.
Actually for the Mac Pro I'd like to see Apple go a step farther and define a standard for desktop SSDs that makes use of PCI-Express. Current SSDs are already bumping up against the limits SATA has.
- 3 additional 3.5" drive-bays (with hot-swap?)
Four would be better and they really should be laptop sized. It would be even nicer if the same frame could handle the PCI-Express SSDs alluded to above. This way one could choose between fast and bulk storage. I see us as still in a transitional period with magnetic not going away anytime soon.
- no slots, Thunderbolt will make up for this
BS!!!! It is getting to the point that everytime I see comments like this I want to grab a 2x4 and smack people about the head. You do realize that the lack of slots in a reasonably priced desktop takes Apple completely out of the running. It is a good and rational reason to select a PC solution.
Thunderbolt has problems that can not be solved anytime soon. For one it results in devices external to the chassis which is a bad thing. Second it is extremely slow compared to PCI-Express. Third it is expensive. Those are just three that come in right off the top of my mind.
Note this isn't being critical of TB but rather it is rejecting of the idea that it is a suitable replacement for internal slots.
- way smaller case
Yep! The current case is a nice bit if engineering but it has more volume than most Pro users need.
- lower price point
At least one model yes. I actually envision one model to serve general purpose users and one a blow out high performance machine.
- release: October 2012 (maybe at a special Mac event?)
Why so lOng? First they need to get TB out soon. Second by the end of the year or maybe early next year the chip market will be fleshed out with all sorts of chips for both the low end and high end.
As with Final Cut Pro X, professionals will cry out foul!
Those so called professionals can go pound salt. I never seen such a bunch of whining bastards in all my life.
Thoughts?
For the most part I like what you propose. The Pro is very much a dinosaur. I actuall flirt with designs in my mind and I sometimes think the best option would be a 2U half rack sized box. That is a lot smaller obviously but with modern tech it is very doable. By the way I wouldn't be striving for a lot of slots in this chassis but we do need to support at least two with one suitable for a GPU card. This would allow for the lowend guy to run off an integrated GPU and at the same time offer support for a supplemental GPU.
The neat thing here is that you could put two of these easily into a 2U rack space for those so inclined. However with a bit of attention to detail you could have a very nice looking stand alone chassis. The concept is easy to play with too, they could go to 3U and have a nicer layout inside. In the end they would have a machine usable in a wide array of applications.
They might leave it in but iDVD is not on the App Store, DVD Studio Pro is gone, the Minis have no optical, the mainstream Mac laptop has no optical. This might be the start of a trend.
This is all true but you know the so called Pros will make a big stink.
It would immediately wipe $200 off the cost of the machine and shrink it down in height by about 1/5th. Less engineering, less wiring, less metal, less weight.
I'm not sure how you get $200 it if a $20 drive but I agree with the rest. For the same reasons I'd like to see the 3.5" disk slots leave also.
Smaller is also a way to change the perception that expansion is always a requirement for any work that uses the performance of a $2500+ machine. The quad-core Mini is already doing this:
Smaller is good. However I think you need to reconsider the issue of expansion. Expansion is the whole point of a Pro machine other wise people would buy iMacs or Minis. Here I'm talking expansion in all the various ways be it RAM, PCI Express slots, "storage bays" or what have you. Without the ability to significantly out do the other machines with expansion there is little reason for the Mac Pro to exist. Simply putting a bunch of CPU power in a box does little good if it isn't supported properly.
Full Xeon chips, MXM graphics cards, 4 x Thunderbolt and a very small enclosure would be nice.
The MXM card can go as those have little draw when it comes to Pros. You did not mention RAM expansion which is a valued part of the current Pro.
The roadmap says the Ivy Bridge EP chips will be out same time next year:
Ivy Bridge holds a lot of promise for Laptops. I'm doing everything I can to resist buying before that hardware shows up. However I'm not so certain such chips will make sense in a Mac Pro. I'm actually hoping that AMD has really good luck with Bulldozer, especially bulldozer cores integrated into Fusion chips. Bulldozer is so different that Apple will likely have to do a bunch of testing but I see great potential for common workloads on Mac OS. Of course this means Apple would need TB hardware that would work with AMD systems but I'm under the impression that will not be a problem.
http://pc.watch.impress.co.jp/img/pc...tml/1.jpg.html
I reckon the Mac Pro will get a Sandy Bridge Xeon refresh in October.
Sounds about right but this could happen earlier, again mainly to support the TB initiative.
For the same reasons I'd like to see the 3.5" disk slots leave also.
How will you install a hard drive? How do you expect anyone to have any measurably sufficient amount of internal storage? 2.5" drives? Can't even get a terabyte that size without paying an arm and a leg, and that's as large as those get. We'll have 5TB 3.5" drives before long (and I realize that's the end of the road for them), so until you can buy the same amount of storage in SSD form without taking out a fifth mortgage, 3.5" drives are here to stay.
A
I don't think this will ever happen again. The days of the $1,400 tower Mac are long, LONG gone.
$1400 Mac Pro towers yes. $1400 iMac based tower sounds like just the computer I'm looking for.
How will you install a hard drive? How do you expect anyone to have any measurably sufficient amount of internal storage? 2.5" drives? Can't even get a terabyte that size without paying an arm and a leg, and that's as large as those get. We'll have 5TB 3.5" drives before long (and I realize that's the end of the road for them), so until you can buy the same amount of storage in SSD form without taking out a fifth mortgage, 3.5" drives are here to stay.
There are probably a number of reasons for that but one issue is power usage. Another is that more go into a box thus when combined in RAID form can offer awfully good performance. That is data center / server hardware.
I'd be the first to agree that this solution is suboptimal for bulk storage. However everybody seems to think the solution there is a TB connected RAID box. In the end part of the reason the Pro is so big is its internal drive bays. With 2.5" drives a bank of three takes up far less space and puts less of a demand upon the power supply. While not the best approach for bulk storage it can easily take care of the needs of many users.
The other way to look at this is that if the Mac Pro is to stay in production they need to move a lot more hardware. The way to do that is to lower the price to make it more attractive to a greater number of users. To do that they need to remove hardware targeted at Pros, especially if it can be moved to a TB connected box. So a Pro with three bays effectively takes care of the needs of a very large number of users while the Pros get to buy external RAID boxes.
the lack of slots in a reasonably priced desktop takes Apple completely out of the running. It is a good and rational reason to select a PC solution.
Out of the running for what? The mass exodus away from desktop to portable computing?
The market isn't there. 70% laptops, 30% desktops, < 5% towers (750k units per year).
Out of that 5% (if it is even that much), what percentage are installing PCI cards? People who make clusters e.g:
http://www.appleinsider.com/articles...teraflops.html
use infiniband cards:
http://en.wikipedia.org/wiki/InfiniBand
but stick an adaptor on a couple of Thunderbolt ports and bond the channels.
The internal storage expansion can be an issue with SSDs like the OCZ IBIS XL:
http://semiaccurate.com/2011/03/03/o...-off-4tb-ssds/
but who's really going to be using this (likely over $12,000 device) before Thunderbolt goes optical? Who really cares if SSDs with read/write speeds in excess of 700MB/s become mainstream and SATA is saturated? Isn't 700MB/s fast enough for the next few years?
One day not too far into the future, they will have 1TB mini SATA cards that go in the MBA and people will wonder why they have a machine the size of a fridge to put them in.
Thunderbolt results in devices external to the chassis which is a bad thing. Second it is extremely slow compared to PCI-Express. Third it is expensive.
It means you can buy a single peripheral and use it with a laptop, Mac Pro, iMac or Mini. It also means you don't have to design your product based on the PCI slot size - you can make them as small or as large as you want. You hit a much large volume of Mac users (20x more) so you can price products cheaper.
I'm not sure how you get $200 it if a $20 drive but I agree with the rest. For the same reasons I'd like to see the 3.5" disk slots leave also.
I noticed they only ship with one so It would be $100 - the drives are not $20. If you configure a second drive, they cost $100. Apple reduced the Mini by $100 by taking out the optical. The Mac Pro would also shrink in size so less metal used and less cabling.
Expansion is the whole point of a Pro machine other wise people would buy iMacs or Minis.
RAM, storage and high performance not PCI expansion. The PCI slots mainly are beneficial for GPUs but MXM cards take up much less space and high-end ones are more than half the performance of a high-end full-size GPU.
A 244W GTX 580 (fastest single-GPU desktop card) is only 75% faster than a 100W 6990M (fastest mobile GPU). Just like Intel is going ULV as standard, GPU manufacturers should go MXM as standard.
but stick an adaptor on a couple of Thunderbolt ports and bond the channels.
What does that mean?
Out of the running for what? The mass exodus away from desktop to portable computing?
the desktop is not dead and in industry you need hardware with slots. And no TB is not a solution.
The market isn't there. 70% laptops, 30% desktops, < 5% towers (750k units per year).
Those are Apples numbers but don't represent industry as a whole. In Apples case they don't have the hardware to market to actually generate solid sales. It is extremely difficult to sell All In Ones or slotless PCs to IT departments.
Out of that 5% (if it is even that much), what percentage are installing PCI cards? People who make clusters e.g:
In one installation I know of every PC has cards installed in it's PCI slots. This amounts to hundreds of machines at one installation. The problem is that Apple is totally excluded from such uses because of the lack of an economical computer with slots.
By the way this has nothing to do with clusters. There is a wide variety of uses for those slots. For example: networking cards, additional ports (USB these days), I/O cards, Vision system support cards and a host of other hardware. You never see Apple in these applications because the lack of slots in a reasonably sized and priced machine takes them out of the running.
http://www.appleinsider.com/articles...teraflops.html
use infiniband cards:
http://en.wikipedia.org/wiki/InfiniBand
but stick an adaptor on a couple of Thunderbolt ports and bond the channels.
Now you are way off in left field! You are trying to imply that the only use for those slots is to install infiniband cards yet I've never seen an infiniband card. Very poor argument if you ask me.
The internal storage expansion can be an issue with SSDs like the OCZ IBIS XL:
http://semiaccurate.com/2011/03/03/o...-off-4tb-ssds/
but who's really going to be using this (likely over $12,000 device) before Thunderbolt goes optical? Who really cares if SSDs with read/write speeds in excess of 700MB/s become mainstream and SATA is saturated? Isn't 700MB/s fast enough for the next few years?
They simple answer there is no! No 700MB/s will not be fast enough, we have already seen what a SSD can do for a laptop with a puny processor, for a desktop it could potentially do even more. Imagine a desktop consuming video data at 1GB/s.
The reality is this secondary storage has trailed in performance for more than a decade now. It has become the worst bottle neck on a PC, so getting performance back up there will lead to more innovation in software and open desktops up to uses that where previously impractible.
One day not too far into the future, they will have 1TB mini SATA cards that go in the MBA and people will wonder why they have a machine the size of a fridge to put them in.
This is one of the reasons I think now is a good time to transition to the 2.5" format. Yeah it is a problem for people that need lots of storage however various forms of solid state storage will displace magnetic storage in the near future.
The big problem right now is the lack of an industry standard for solid state storage printed circuit cards. Considering we are in a traditional time here it would be nice to have a high speed card standard that can use the same bays as the magnetic 2.5" drives.
It means you can buy a single peripheral and use it with a laptop, Mac Pro, iMac or Mini. It also means you don't have to design your product based on the PCI slot size - you can make them as small or as large as you want. You hit a much large volume of Mac users (20x more) so you can price products cheaper.
The problem is PCI cards are extremely cheap. TB solutions will never be comparably so.
I noticed they only ship with one so It would be $100 - the drives are not $20. If you configure a second drive, they cost $100. Apple reduced the Mini by $100 by taking out the optical. The Mac Pro would also shrink in size so less metal used and less cabling.
RAM, storage and high performance not PCI expansion. The PCI slots mainly are beneficial for GPUs but MXM cards take up much less space and high-end ones are more than half the performance of a high-end full-size GPU.
You keep going around and around with this stuff about PCI slots only being good for x,y & z yet that clearly isn't the case. PCI slots are useful to support any card you can get a driver for.
A 244W GTX 580 (fastest single-GPU desktop card) is only 75% faster than a 100W 6990M (fastest mobile GPU). Just like Intel is going ULV as standard, GPU manufacturers should go MXM as standard.
I'm not a GPU snob, nor do I have demanding OpenCL needs, however isn't the whole point of the Pro to support the needs of demanding users? If you tell such people that they will not be able to run the faster GPU hardware do you think they will stay around as customers?
I look at this way, by the middle of next year there will likely be processors out that suitably support my needs for a GPU right on the chip. I'm thinking Ivy Bridge in Intels case or AMD Fusion APUs with a new CPU core. That is me, it says nothing about the people that leverage GPU cards on a daily basis. Will those people be happy with a MXM card that might give them half the performance they where expecting. I doubt it honestly. The Pro simply isn't directed at the market that is happy with the current integrated GPU solutions nor will they be happy with future solutions.
What does that mean?
General this is done to go faster. Sometimes it is a good solution other times it isn't so good. Implementation details make all the difference.
What does that mean?
Channel bonding is using multiple data channels as one:
http://en.wikipedia.org/wiki/Channel_bonding
http://en.wikipedia.org/wiki/802.3ad
So even though a single Thunderbolt channel is 10Gbps, it's possible they can use multiple ports together for higher throughput.
There is no mass exodus away from desktop hardware in industry.
http://gigaom.com/apple/decline-of-the-desktop-mac/
Those are Apples numbers but don't represent industry as a whole.
The industry as a whole is also 70% laptop, 30% desktop. 350m units per year so 105m desktops of which 10 million AIOs + however many of those HTPC machines.
The PC industry still has the tower thing going but it's not a growth market by any stretch of the imagination.
There is a wide variety of uses for those slots. For example: networking cards, additional ports (USB these days), I/O cards, Vision system support cards and a host of other hardware. You never see Apple in these applications because the lack of slots in a reasonably sized and priced machine takes them out of the running.
I'm glad about that tbh. One look at this page only strengthens my opinion that this stuff needs to just go:
http://www.visionsystems.de/1_1_3_2_2.html
To me installing a specialised card for such low bandwidth ports is a silly thing to do. Driver support is a nightmare and you get stuck with horrible legacy hardware. The 'U' in USB is there for a reason - we should be moving closer and closer towards IO standardisation and minimalism and you don't get much more standard or minimal than a Thunderbolt port.
Imagine a desktop consuming video data at 1GB/s.
Ok but why is someone running 25x 1080p ProRes 4444 streams at the same time?
No 700MB/s will not be fast enough, we have already seen what a SSD can do for a laptop with a puny processor
We've seen what 150-200MB/s SSDs can do so we still have room over the next few years to actually saturate SATA in the mainstream before we need to transition to a faster connection.
This is one of the reasons I think now is a good time to transition to the 2.5" format.
Apple doesn't seem to be thinking this way yet - the iMac still uses the 3.5" drives - but I'm not a fan of 3.5" drives.
The problem is PCI cards are extremely cheap. TB solutions will never be comparably so.
That can only change with adoption. You will sell far more USB storage drives than PCI SSD drives and it's just due to accessibility. If people have the port and manufacturers support it, the price of the solutions will drop.
You keep going around and around with this stuff about PCI slots only being good for x,y & z yet that clearly isn't the case. PCI slots are useful to support any card you can get a driver for.
And limited to under 20% of the total computing market and shrinking.
isn't the whole point of the Pro to support the needs of demanding users? If you tell such people that they will not be able to run the faster GPU hardware do you think they will stay around as customers?
I don't think people would migrate away from a platform for the sake of dropping to 60% graphics performance. Nobody is buying a $2500+ machine to play video games and for other 3D tasks, the high-end mobile GPUs are perfectly capable. It's only really equivalent to being a year or two behind in performance, which Mac users have been used to for a long time and hasn't really affected anything.
The CPUs wouldn't be compromised, which is the main thing.
Probably the same case. It'll be interesting if they redesign it to be rackmounted per rumors, but it'll need a big case in any situation.
They should just forget about single socket machines. All machines should be 2 sockets, minimum, in a 8-core, 8-core, 12-core good, better, best lineup. A 16-core upgrade should be offered. Going 2-socket with 8 memory slots will necessitate a pretty large case, especially of they want it to be quite. A rack mount system should be a 3U thickness minimum.
They should keep the current 4 PCIe slot, 2 5.25" bay and 4 3.5" bay design.
It aint going to be cheap. A 2 socket 8 core system would probably be something like $2500.
catch and beat everything out on the market!
*Upgradable to a 1.8" SSD rather then a 2.5" drive for more battery room!*
^in a perfect world
remove the optical drive, use new battery technology (rollls ?), keep same body size, use extra space for battery room, insert imac processor and GPU. No mobile version, or low energy consumption, just full on power!
catch and beat everything out on the market!
*Upgradable to a 1.8" SSD rather then a 2.5" drive for more battery room!*
^in a perfect world
Mac. Pro.
Not your laptops, for heaven's sake.
DESKTOP USERS STILL EXIST.
Hopefully they'll go all out on the 2011 Mac Pro:
Yes! Apple really needs to overhaul the whole concept.
Probably the same case. It'll be interesting if they redesign it to be rackmounted per rumors, but it'll need a big case in any situation.
It doesn't "need" to be big. The current Pro is rather spacious. On top of that you can get a lot of performance out of lower power processors these days.
They should just forget about single socket machines.
This statement really bothers me. They need a single socket option to allow shipping the base Pro at a far lower price point.
Beyond the need for a lower price eight cores on a single die should be a reality in a few months. With the smaller processes the ability to get a lot of functionality out if a single core is significant. The whole point of a base machine would be to offer up a performance metric somewhere between a Mini and a Dual Socket Pro.
All machines should be 2 sockets, minimum, in a 8-core, 8-core, 12-core good, better, best lineup. A 16-core upgrade should be offered. Going 2-socket with 8 memory slots will necessitate a pretty large case, especially of they want it to be quite. A rack mount system should be a 3U thickness minimum.
That is all well and good for the high end of the market but it sucks for somebody just needing midstream performance and a good GPU.
They should keep the current 4 PCIe slot, 2 5.25" bay and 4 3.5" bay design.
Nope, not a chance. They would accomplish nothing by just throwing a new mother board in the box.
It aint going to be cheap. A 2 socket 8 core system would probably be something like $2500.
Which is no problem for people that need that sort of machine. However it is pretty obvious Mac Pro sales suck and are getting worst. They need to make a machine that is attractive to a wider audience.
I'd love to see better options for sound output since aftermarket (internal) sound cards wont work.
I'd love to see better options for sound output since aftermarket (internal) sound cards wont work.
Uh, what's wrong with the internal optical audio instead of trashy PCIe sound cards?
Uh, what's wrong with the internal optical audio instead of trashy PCIe sound cards?
Well, I have a real nice Sennheiser headset that wont work with optical and I use it for both my MAC and PC, I get much better sound on the PC because I don't have to use an adapter (external USB sound card).
Yeah, I could get a new headset but it's nice to have the option.
This statement really bothers me. They need a single socket option to allow shipping the base Pro at a far lower price point.
They do this currently with the 4-core 2.8 GHz Nehalem single socket Mac Pro. $2500. It's not a good deal, especially when for $3500, you can get 2 Westmere CPUs (8-cores total) and double the RAM.
If someone is buying it, their needs have to be pretty unique. It's an outrageous deal in the bad sense if you are buying it just for the expansion.
Beyond the need for a lower price eight cores on a single die should be a reality in a few months. With the smaller processes the ability to get a lot of functionality out if a single core is significant. The whole point of a base machine would be to offer up a performance metric somewhere between a Mini and a Dual Socket Pro.
How could this work when Apple doesn't sell a Mac Pro for less than $2500? And 2 low end 4-core Westmere CPUs will cheaper than an 8-core Sandy Bridge Xeon. The biggest question in my mind is that 8 cores is probably too much power for Apple to put in there low end Mac Pro.
Which is no problem for people that need that sort of machine. However it is pretty obvious Mac Pro sales suck and are getting worst. They need to make a machine that is attractive to a wider audience.
It appears Apple is happy with selling Mac Pros for $2500+ and offering consumers an iMac or a Mac mini.