I see the Mac Pro having the same case or similar enough case (glossy black on the case somewhere?
). Other than that, I see no reason for change of the case.
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.
The 2009 (Gainestown) Mac Pro: Everything We Know
- Thread starter Tallest Skil
- Start date
- Sort by reaction score
- Status
I see the Mac Pro having the same case or similar enough case (glossy black on the case somewhere?
). Other than that, I see no reason for change of the case.
I see the Mac Pro having the same case or similar enough case (glossy black on the case somewhere?
Nehalem is triple channel instead of dual channel and uses standard DDR3 DIMMs instead of FB-DIMMS so the risers are out of the question. The current case won't support 12 regular DDR DIMMs without getting rid of half of the drive bays and a couple of the PCI-E slots. Pros wouldn't like that very much. Its going to need a different and probably bigger case for the new platform.
Well, I guess that is fine. As long as things improve, the appearance should not matter that much.
Although, they could make the current Mac Pro with desktop CPUs and sell it for a cheaper price (when the new Mac Pros come out). This would be the same situation as the current white MacBook vs Aluminium MacBook.
I don't think the case would have to be larger... the FB-DIMMs take up so much room, and you could fit a lot of DDR3 DIMMs in place of 8 FB-DIMMs. If anything, I would say they could theoretically make the case smaller. not saying they will.
You could actually still use riser cards. No problems other than the engineering that has to go into it routing the wires.
Well, I guess that is fine. As long as things improve, the appearance should not matter that much.
Although, they could make the current Mac Pro with desktop CPUs and sell it for a cheaper price (when the new Mac Pros come out). This would be the same situation as the current white MacBook vs Aluminium MacBook.
I don't see selling the old Mac Pro with desktop CPUs, but I could definitely see a split coming in the Mac Pro ranks with both a single CPU bloomsfield machine in a case that is smaller than the current Mac Pro and a dual CPU gainestown machine that is larger than the current Mac Pro. Like the Macbook Pros and the iMacs, the twin Mac Pros would share a common styling. They could also differentiate the lines by using Quadro/Fire cards exclusively in the dual socket version and GeForce/Radeon in the single socket version.
In the xServe that may be true. However, the FB-DIMMs in the Mac Pro are mounted on a riser card to save space. Fitting six regular DIMMs in the space of four FB-DIMMs may be possible. That being said, there's a pretty good chance that standard DIMMs cannot be mounted on riser cards like the FB-DIMMs and may have to be mounted on the motherboard. If they're mounted on the motherboard, they'll take up a significantly higher amount of space.
I haven't seen it done with regular DIMMs though. All the articles I can find hint that the use of riser cards on the Mac Pro is due to the use of a serial connection in FB-DIMMs.
Bingo. I've been thinking about this possibility for some time, and although I don't see it coming, I think it would be quite good.
The smaller Mac Pro could also have less HD bays than the larger one. Also given the wide variation in Gainestowns, the smaller Mac Pro could even use 2 cheap Gainestowns (although that may not be as appealing).
There might be some obscure Professor in a Uni lab working on it as we speak.But that's string theory, and I don't remember Apple having a Plancktech engineering team... although... those guys from PA Semi might... no, no; that's not right... forget it.
It depends.
The current model uses a slightly modified (locations) SSI CEB board (10.5" x 12"). Risers were needed to fit the memory on that small an area.
Apple could chose to either keep risers with DDR3, or go with a more traditional layout. Risers are technically possible, though it would require a custom connector. If they decide to place the DIMM's directly on the board, they'll have to go larger (SSI EEB/E-ATX; 12" X 13"), forcing the need for a larger case. Given Ive's designs lately (thin, sleek, small...), I don't think this will be the way Apple chooses to go.
Though this might make a lot of people happy, as they'd finally get their dreamed of "eMac", I doubt it.
It would almost certainly cut into the iMac sales, and I can't think of a reason Apple would want that, given the investment they've already spent on developing that market.
It wouldn't be a problem in the Xserve. MP already covered above.
FB-DIMMs are serial, and only use 69 pins of the 240 pin PCB.
So Yes, FB-DIMM's are easier, but DDR3 isn't impossible. It would require a custom connector (quite large @720 pins). Say 360 pins per side, 1.0 mm pad & spacing, double sided. Add in notches, so ~4.5 cm wide realistically. This seems technically possible to me. At 2x of these per machine, even as custom parts, it could prove financially viable when compared to the cost of switching to larger boards. (Additional $$ for the board & case).
There might be some obscure Professor in a Uni lab working on it as we speak.
It depends.
The current model uses a slightly modified (locations) SSI CEB board (10.5" x 12"). Risers were needed to fit the memory on that small an area.
Apple could chose to either keep risers with DDR3, or go with a more traditional layout. Risers are technically possible, though it would require a custom connector. If they decide to place the DIMM's directly on the board, they'll have to go larger (SSI EEB/E-ATX; 12" X 13"), forcing the need for a larger case. Given Ive's designs lately (thin, sleek, small...), I don't think this will be the way Apple chooses to go.
No offense to Ive, but I hope they keep his input here to styling only. He and this market and not even close to on the same page.
Though this might make a lot of people happy, as they'd finally get their dreamed of "eMac", I doubt it.
It would almost certainly cut into the iMac sales, and I can't think of a reason Apple would want that, given the investment they've already spent on developing that market.
I can think of one: user retention. Having owned one, there is simply no way I'm going to make that mistake again and I'm not alone. Some users may successfully transition to the iMac, but for others its a total mismatch.
FB-DIMMs are serial, and only use 69 pins of the 240 pin PCB.
At 2x of these per machine, even as custom parts, it could prove financially viable when compared to the cost of switching to larger boards. (Additional $$ for the board & case).
At 720 pins, 2 cards would only be good for a single socket machine. For a dual socket machine, you'd need either 4 720 pin 3-DIMM cards or two 1440-pin 6-DIMM cards.
The likelyhood of risers for DDR3 is slim to nil. FBDIMM's made it easy for a number of reasons that are not shared with non-buffered RAM. The engineering complexities far outweigh the alternative of a case redesign.
It may not be necessary to make the case larger, a different layout of the system internals may be all that's necessary. Having said that, a larger case wouldn't be a bad thing, if it makes good use of the added space.
It may not be necessary to make the case larger, a different layout of the system internals may be all that's necessary. Having said that, a larger case wouldn't be a bad thing, if it makes good use of the added space.
None taken.
Leave styling to Ive, and the design work to the engineers. I'd love to see an eMac. I'm just thinking that Apple wouldn't do it, as they'd expect it would damage sales of the iMac. (Reducing profits, and even recovering R&D, part supplier contracts, tooling, etc. spent to develop it).
I understand.
From posts I've noticed, some, if not many, would like the ability to easily upgrade components in a desktop class machine.
I would as well, particularly if I could get away with using one. I like saving $$$ when I can. From what I gathered looking at the data sheets, No.
In this case, the DDR3 is attached directly to the CPU's memory controller. No more going through the chipset. Yay! LGA 1366 = 1366 pins, and not all are even used (remaining are reserved for future functions).
So how would you connect 1440 pins to a component that doesn't have that many?
Simple. It doesn't require 1440.
It only needs 240 pins per channel, so 3 channels = 720. The memory controller switches between the DIMM's. Since each CPU has it's own memory attached, two banks will be seen. The real question is, which physical methodology will we see?
Risers (2 most likely), or go with a larger board to fit them directly to it. Given Apple seems to be on a "small size" trend, I'm still thinking the risers are a strong possibility.
If I had it my way, it would just be a larger board. Makes the argument for a larger case viable, and then extend it to more drive locations while we're at it.
(So I'm greedy).
None taken.
I wish it worked like that/
From posts I've noticed, some, if not many, would like the ability to easily upgrade components in a desktop class machine.
Honestly, Apple got less money out of me from the iMac than they would have with a single socket MacPro.
From what I gathered looking at the data sheets, No.
In this case, the DDR3 is attached directly to the CPU's memory controller. No more going through the chipset. Yay! LGA 1366 = 1366 pins, and not all are even used (remaining are reserved for future functions).
So how would you connect 1440 pins to a component that doesn't have that many?
Simple. It doesn't require 1440.
I figured 240-Pins for each DIMM.
Just what are you guys going to do with such powerful machines?
Research parallel universes?
"Research"? No...
Create.
As for the pins needed for RAM...
While DDR3 modules use 240 pins per module, not all 240 need to be connected directly to the memory controller. Many are ground and voltage pins, which can go directly to the power and ground circuitry in the motherboard (and can, indeed, be shared among channels,) it is only the address and data pins that *MUST* go to the memory controller. (Although for proper signal propagation, it's good form to include the signal pins plus some ground.)
In addition, each channel must go the memory controller, not each module. So if we have three modules per channel, three channels per processor, we may have nine modules total, but only three sets of address/data need to go to the proc. If you ever look at the traces on a motherboard, you'll see that the traces for a second SDRAM module go directly from the first one. SDRAM (including DDR, DDR2, DDR3, etc,) is a 'serial' architecture. Back in the PC-100 memory days, you often saw four memory slots. These boards only had a single 'channel' of RAM, so only one set of pins went to the memory controller, the additional slots just added 'on the end', as it were. It's also why you see slots for a single channel bunched together. The wiring is simple. Once you have one slot wired, you just extend the lines a few millimeters longer to get to a second slot. A second *CHANNEL* can be wired to go somewhere else on the board; but you don't want to have multiple slots for a single channel be far away.
Disclaimer again: I'm not a motherboard or memory engineer; this is all based on experience gained from years of doing support on these things.
So even though one might conclude that you need 720 pins JUST for the three memory channels, significantly fewer are actually used for it. (I don't have the exact number off the top of my head; and it's possible that I couldn't provide it without breaking NDA, so I'm not going to search.)
Just what are you guys going to do with such powerful machines?
Research parallel universes?
What are we going to do with them?
Accurately guess what the next Mac Pro after Gainestown will be...
Anyone else want to chime in here???
https://forums.macrumors.com/showthread.php?p=7125714&posted=1#post7125714
https://forums.macrumors.com/showthread.php?p=7125714&posted=1#post7125714
All pins would have to go through the connector though, if Apple chooses to use risers again.
BTW, it wouldn't break the NDA. JEDEC (4.20.19 - 240-Pin PC3-6400/PC3-8500/PC3-10600/PC3-12800 DDR3 SDRAM
Unbuffered DIMM Design Specification), Rev. 1.0, October 2008) spec is freely available for DDR3.
Pin assignments are pgs. 11 - 12 (Table 3.0.3).
I can definitely wait for Westmere, I have my current Mac Pro and planning to get a 17 inch MacBook Pro soon when i get my tax returned and i've already started to save up for Westmere Mac Pro, 18 months is plenty of time for me to put more money away for it but don't get me wrong Gainestown is going to be awesome and I can wait to see everyone that is waiting for it to show it off for us and good luck to you too I know most of you guys here have been waiting for a long time.
All pins would have to go through the connector though, if Apple chooses to use risers again.
BTW, it wouldn't break the NDA. JEDEC (4.20.19 - 240-Pin PC3-6400/PC3-8500/PC3-10600/PC3-12800 DDR3 SDRAM
Unbuffered DIMM Design Specification), Rev. 1.0, October 2008) spec is freely available for DDR3.
Pin assignments are pgs. 11 - 12 (Table 3.0.3).
Thank you, yes.
So one thing is that all of the Vdd and Vss pins (for all modules) could share many fewer connections to a riser card. No need to have all those separate pins for the same power distribution. You could get away with a few higher-amperage 'large pins', which would take up significantly less room. I count 70 pins per channel between those two. So we could replace 210 of the 720 pins with two Vdd and Vss 'large pins'. And I haven't even dug through to see how many more pins could be duplicated between channels.
Let's assume that we can't actually drop a full 210 off, and say we have 550 pins per triple-channel riser. Perfectly doable.
have these mockups been put up yet?
february 4th, on gizmodo...
http://i.gizmodo.com/5146440/would-you-be-happy-with-this-as-your-next-mac-pro
found on my nightly google news search of intel, gainestown, macpro, and combinations thereof.
february 4th, on gizmodo...
http://i.gizmodo.com/5146440/would-you-be-happy-with-this-as-your-next-mac-pro
found on my nightly google news search of intel, gainestown, macpro, and combinations thereof.
have these mockups been put up yet?
february 4th, on gizmodo...
http://i.gizmodo.com/5146440/would-you-be-happy-with-this-as-your-next-mac-pro
found on my nightly google news search of intel, gainestown, macpro, and combinations thereof.
The only difference from the current Mac Pro seems to be...a black Apple-logo.
That's pretty funny. Just how big do you think the molecules in air are exactly? Hole size isn't the issue.....it's the ratio of open area to closed area.
S-
- Status