Will there be a new Mac Pro in 2015?

[deconstruct60]

My guess is:

Q1-Q2 2016

Intel Xeon E5 v4 (Broadwell EP)
DDR4-2400 RAM (16GB min)
New PCI-e 3.0 SSD (256GB min, but cheaper 512GB-1TB options)
Double internal slot for SSD (1 left empty in the 256GB version)
6 Thunderbolt 3/USB 3.1 ports

Unless dumped the other GPU, the E5 v4 runs out of PCIe v3 lanes after get to two x4 PCIe v3 TB3 controllers. So no v3 SSD. No 3rd controller to get to 6 TB sockets either. Can't crank up both SSD and TB. It is going to be one or the other.

The C612 chipset supports 6 USB 3.0 sockets so maybe 6 of those, but probably just 4 of the new TB flavor.

Second SSD slot is doubtful since same x8 PCIe v2 limit the current Mac Pro has. Apple could put in a switch so the two x4 PCIe v2 SSDs can share the same x4 PCIe v2 allocation. That buys capacity by trading off speed. If try ti read or write at some time to both then get half the throughput. I suspect most folks are going to want top end speed from both. If they go to NVMe SSD that is even worse as the individual SSDs will be even closer to maxing out the x4 link all by themselves; let alone 2.

Apple could back away from the "More than two" TB socket path. One TB v3 controller. Some other Type C sockets that are just just DP+USB 3.0. Still have max 6 mainstream displays but it turned out folks didn't often attach 4+ TB devices to the systems. That would free up a x4 PCIe v3 link on CPU's lanes for ultra fast SSD connection. The other 2nd slot would be a just a 4x PCIe v2 SSD. Apple would have to get over their OCD symmetry syndrome though.

My secret hope is to see some "Jobs style" anticipated purchase of Skylake EP processors, like they did with the first Xeon (first Intel Mac Pro). But I think we need to see whether Intel wants to skip the Broadwell EP generation.

There was likely no "Jobs style" anticipated purchase ... in 2009.

Most reasonable large system makers that are building Xeon server class system gets get some "early" (as in before initial announcement) in some reasonable volume. That is so they can roll out "test"/"qualification" servers under NDA to their larger customers ( who typically have long purchase request timelines) and to ramp production on.

Apple merely jumped out the gate early with that initial drop. Relatively few beta units sent out and not a top 5 volume players so probably had enough with lower production targets to ramp to.

Apple didn't have special access they just ran out of the gate early whereas the other players we waiting to sync up with Intel's official announcement. It was more Apple pissing on Intel's launch marketing event than special access thing. It wasn't a norm. It still isn't one.

 

[deconstruct60]

CPUs in these days in terms of performance are more or less meaningless. Everything moves more and more on GPUs and that it the way it has to be.

Embarrassingly parallel work will move. There is more of that possible than software whose design origins date from >10-15 years ago would suggest. There is still and will be a subset of workloads that classic CPU cores will still run better. Graphics on something other than a graphics processor unit will be more of a odd duck going forward. The excuses of "needed access to main memory" , "needed more general CPU instructions for some limited sections of the code", etc. are being eliminated.

Apple invests in that very much, and Metal and Mac Pro are just simple emanations of it.

It isn't so much emanation as standardization. If most Pro Macs have two GPUs then their will be more software that makes use of that more commonly available resource. Similar if the standard libraries support GPGPU programming transparently it will be used more often.

Times when things were made by dual CPU and single GPU on Mac Platform are over. Simple as that.

CPUs aren't what they were. Lots of CPU designs are trending toward collapsing what was once two CPU packages into one package. If the transistor budget is twice as big now don't really need two packages. At that point get into how many is enough zone. 12, 16, 18? if need alot more than that the Perf/Watt of the narrowly focused GPGPUs is better.

Apple potentially waited for AMD to show the Fury GPUs, so they can update that computer relatively soon,

I guess since there was no Global Foundry switch and the rebadges are mostly the same clocked higher ... Fury is now back in the forefront as to the wait?

The Mac Pro seems to now suffer from tight coupling of too many things. If have to wait for the intersection of new CPU + GPU + Thunderbolt to get another system out then it turns into a periodic hobby business for Apple.

HBM GPU packages that lead to smaller, less complicated GPU boards are certainly on Apple's "like to have" list. Coupled with Apple moving the physical connectors off the board they have pretty much simplified what needs to be done to get to a custom design board at lower prices. Signals routed in and out of the board + power management ... and that is about it.


The staged launch of Fury strongly suggests that it is yield limited. I doubt Apple wants to jump into a yield limited component for the Mac Pro. That isn't a relatively soon indicator.

especially that even people already owning Mac Pro are eager to upgrade.

Not sure about that. A limited set. But most of the folks who bought one in 2013-2014 probably aren't moving. A large block of the 2012 folks are either. There are buckets of thread about how push a 2006-2008 up the hill another year or two ( even before the new Mac Pro form factor appeared).

I doubt they generate another 4 month demand backlog with this launch. If they go way past 2 years, maybe. Have to remember that in 2013 they had pragmatically killed off the whole EU market at for about 10-11 months of no "new from the store" Mac Pro ( whatever form factor). Few now are eager like that subset was back then.

And that it will have Haswell CPUs and best possible in that thermal envelope GPUs from AMD.

There is no major thermal envelope advantage to Xeon E5 v3 or v4 ( Haswell/Broadwell) package. There isn't likely going to be for v5 either. Intel is going to take the 140-160W budget and either crank higher core count and/or higher clock speed for the parts that Apple is likely to use.

The GPUs aren't that much better. Probably will take the same TDP budget and crank clocks/core count higher. ( Fury plows much of the power saved from dropping GDDR5 back into count and clock. )

The whole notion that Apple is waiting for way lower TDP parts to make it work is lacking substance.[/quote][/quote]

 

[deconstruct60]

I don't see how Apple could fit two water block inside the nMP...

Apple doesn't necessarily need a water block. Water is more effective Thermal collector transmitter than air but Apple isn't using direct air blowers on the current GPUs. They are using metal/Aluminum. Aluminum is better than water. The thermal core is both larger and have more surface area than any of the nominal metal blocks being used on card precisely because it does not have to fit in a slot (or even double slot ).


The second aspect of water is to move the heat to the edge (away from the other heat generating components ) and then expel it. Again the thermal core already largely does this now. It is moved in and that central core is all expelled out just about all of the top edge. (largley no other connectors soaking up exhaust space. A little wifi antenna space but that's it. ).

AMD also is using the water block to make it smaller and quieter than their previous top end single GPU designs.

[ at 1:37 into this video they say the size is one of the selling points]

AMD is trying to put this into mini-ITX chassis. The Nano is easier to plug into common small ITX chassis (as don't need to find a place to stick the fan on a edge mounting. ), but a small chassis that can take edge cooler blower will work.

Nvidia's stuff in the same performance class are likely going to be physically bigger cards. AMD has about a year window here where they will be able to fit into smaller space. Lots of folks have big boxes with slots where not as much of a differential but AMD doesn't have to win everywhere. They just need to sell enough to be more profitable. [ even inside of larger box the shorter card show enable somewhat better/cleaner routing of cabling. ]

The embedded GPUs in the Mac Pro are already compacted. A Fury Package model would make it easier to build custom boards from at least the electrical layout perspective. Need to properly couple it to the thermal core. That isn't a major disconnect though.

 

[tuxon86]

Apple doesn't necessarily need a water block. Water is more effective Thermal collector transmitter than air but Apple isn't using direct air blowers on the current GPUs. They are using metal/Aluminum. Aluminum is better than water. The thermal core is both larger and have more surface area than any of the nominal metal blocks being used on card precisely because it does not have to fit in a slot (or even double slot ).


The second aspect of water is to move the heat to the edge (away from the other heat generating components ) and then expel it. Again the thermal core already largely does this now. It is moved in and that central core is all expelled out just about all of the top edge. (largley no other connectors soaking up exhaust space. A little wifi antenna space but that's it. ).

AMD also is using the water block to make it smaller and quieter than their previous top end single GPU designs.

[ at 1:37 into this video they say the size is one of the selling points]

AMD is trying to put this into mini-ITX chassis. The Nano is easier to plug into common small ITX chassis (as don't need to find a place to stick the fan on a edge mounting. ), but a small chassis that can take edge cooler blower will work.

Nvidia's stuff in the same performance class are likely going to be physically bigger cards. AMD has about a year window here where they will be able to fit into smaller space. Lots of folks have big boxes with slots where not as much of a differential but AMD doesn't have to win everywhere. They just need to sell enough to be more profitable. [ even inside of larger box the shorter card show enable somewhat better/cleaner routing of cabling. ]

The embedded GPUs in the Mac Pro are already compacted. A Fury Package model would make it easier to build custom boards from at least the electrical layout perspective. Need to properly couple it to the thermal core. That isn't a major disconnect though.

Time will tell...

 

[AidenShaw]

Apple doesn't necessarily need a water block. Water is more effective Thermal collector transmitter than air but Apple isn't using direct air blowers on the current GPUs. They are using metal/Aluminum. Aluminum is better than water. The thermal core is both larger and have more surface area than any of the nominal metal blocks being used on card precisely because it does not have to fit in a slot (or even double slot ).

Unfortunately, the Apple Tube has no flexibility - from day one it has needed to throttle components to stay within a very restricted power and thermal envelope. The "form over function" design has left no leeway for a new version with even modestly larger TDP requirements - it's a "dead end" design.

AMD also is using the water block to make it smaller and quieter than their previous top end single GPU designs.

[ at 1:37 into this video they say the size is one of the selling points]

That video makes me feel very sorry for ATI - if letting the user choose whether the load meter LEDs are blue or red is one of the big ticket items for their flagship GPU....

ps:
I really don't like the new reply editing options.... No formatting or WSIWYG editing? How 1980's.

 

[flat five]

ps:
I really don't like the new reply editing options.... No formatting or WSIWYG editing? How 1980's.

click the B and type
or I and B
or just I
or U
etc..

(or i'm probably misinterpreting what you're saying)

 

[AidenShaw]

click the B and type
or I and B
or just I
or U
etc..

(or i'm probably misinterpreting what you're saying)

I don't see any "I" or "B" or.... Just a simple text widget with no buttons or options.

From my normal account, the "tool bar" is empty (no buttons) - from another system it had the formatting icons. Anyone know which security setting might be the reason?

 

[flat five]

I don't see any "I" or "B" or.... Just a simple text widget with no buttons or options.

hmm.. not sure then. i get this:

using Safari.app

 

[AidenShaw]

- I see that from another system at work, but from my main workstation I see this:

Damn, can't include an image since the toolbar is blank!

(edited from the work server, showing a screen capture from my workstation)

 

[flat five]

- I see that from another system at work, but from my main workstation I see this:

Damn, can't include an image since the toolbar is blank!

View attachment 562820 (edited from the work server, showing a screen capture from my workstation)

that's odd.
no idea what you should try in order to get the toolbar.
maybe a mod can assist?
good luck

 

[avkdm]

- I see that from another system at work, but from my main workstation I see this:

Damn, can't include an image since the toolbar is blank!

View attachment 562820 (edited from the work server, showing a screen capture from my workstation)

Go to preferences in your profile page and select "use the rich text editor to create and edit messages"

 

[AidenShaw]

Go to preferences in your profile page and select "use the rich text editor to create and edit messages"

It's already set to rtf....

 

[avkdm]

It's already set to rtf....

Have you got flash switched off? Otherwise install Java For Mac and see if that helps.

 

[koyoot]

http://wccftech.com/fast-amd-radeon-r9-nano-find/

If this turns out to be true it is just astonishing. 4096 GCN cores with clock rate of around 950 MHz gives 7.7 TFLOPs of compute power.

I will not be shocked if next Mac Pro will simply double the performance of current D700, then .

 

[AidenShaw]

Have you got flash switched off? Otherwise install Java For Mac and see if that helps.

The buttons are there after all - if I hover I see the alt text and they work.

The button images aren't loading....

 

[lowendlinux]

http://wccftech.com/fast-amd-radeon-r9-nano-find/

If this turns out to be true it is just astonishing. 4096 GCN cores with clock rate of around 950 MHz gives 7.7 TFLOPs of compute power.

I will not be shocked if next Mac Pro will simply double the performance of current D700, then .

I hope so I really want AMD to knock one out of the park actually I want them to knock two out.

 

[avkdm]

I hope so I really want AMD to knock one out of the park actually I want them to knock two out.

Me too, always love the underdog. I bought an R9 290 reference for this reason (price too!)Trouble is it was power hungry and noisy, I new this before purchase and didnt think it would bother me, but it obviously did in the end so I went with a 970 even though the card was rock solid.. Anyhow lets hope the nano turns out to be the one. If its as good as its made out to be Nvidia will finally have some competition in the low power/quiet/high performance stakes.

 

[AidenShaw]

Me too, always love the underdog. I bought an R9 290 reference for this reason (price too!)Trouble is it was power hungry and noisy, I new this before purchase and didnt think it would bother me, but it obviously did in the end so I went with a 970 even though the card was rock solid.. Anyhow lets hope the nano turns out to be the one. If its as good as its made out to be Nvidia will finally have some competition in the low power/quiet/high performance stakes.

From the Windows side, few people expect that ATI will survive the next 12 months. Nvidia is so far ahead on the high end with relatively low power GPUs, and Intel has taken over the low end with embedded GPUs - there's no place left for ATI.

 

[avkdm]

From the Windows side, few people expect that ATI will survive the next 12 months. Nvidia is so far ahead on the high end with relatively low power GPUs, and Intel has taken over the low end with embedded GPUs - there's no place left for ATI.

I hope you are incorrect otherwise Nvidia prices will go through the roof!

 

[deconstruct60]

Unfortunately, the Apple Tube has no flexibility - from day one it has needed to throttle components to stay within a very restricted power and thermal envelope. The "form over function" design has left no leeway for a new version with even modestly larger TDP requirements - it's a "dead end" design.

It is a dead end if there are never any future improvement in components, but that isn't going to happen.

They could keep the same basic design and increase the diameter and height from 6.6" d , 9.9" h to 7.6" d , 10.9" h and not give up on the approach. The larger diameter gives increased air flow (with no necessary noise increase) and the thermal core would grow in two dimensions (i.e., bigger capacity). Could probably get a 500-550W power supply in there and dissipate that envelope just as "almost" silently.

IMHO, I do think they cut it too small. There seems to be a bit of an expectation that the "Compute" GPU will go under utilized a significant fraction of the time. That is somewhat at odds with making it a standard component. The software to drive more CPU+GPU balances loads is coming. I think the original one was weighted toward software that that predates the shift (if they collected data with mostly that software, then drifting into that sub-optimal space isn't surprising. ). Additionally, the "power distribution" load requirements for USB 3.1 and TBv3 aren't going to help either ( won't have 100W on every port (or any port ) , but going to get more power hungry stuff connected. )


It doesn't scale up comfortably to 900-1000W desktop, but they aren't shooting for that kind of range. "Fast at any thermal cost" isn't the point. Not going to overclock CPU or GPU so stuff like water chillers aren't as necessary. There is enough power to get work done for most folks.

The internal component mismatch are a bigger flaw right now. 10+ SATA ports on the chipset where there are zero SATA devices is a huge waste. Pragmatically, Apple could stick a standard M.2 SATA socket on the 2nd GPU. That's more a flaw of Apple's OCD design (maximum symmetry, no mix-and-match) than anything connected with the core trends the Mac Pro is now following. The 3rd party market for M.2 SATA SSDs would be simple to allow. What Apple needs (and are way ahead of the curve) is more PCIe lanes (or faster ones on chipset). But that is just one indicator of "too early" the 2013 design is/was. The components still need to catch up a bit more so than power.


I doubt Apple does down this path but could stick a "screw low noise, ramp the fan high" mode if feed in power over a Type C connector.

That video makes me feel very sorry for ATI - if letting the user choose whether the load meter LEDs are blue or red is one of the big ticket items for their flagship GPU....

Errr, go to any gamer oriented system build site. There are lights on cases, "display show case" transparent sides, glowing things, and "Pimp my ride" options abound. AMD knows who they are selling to with this specific card and market. The FirePro versions will be more boring looking with alot less disco when they show up in 6-9 months. (they might skip FirePro versions until get to next generation HBM2 models ).

 

[deconstruct60]

Time will tell...

Thermodynamics and aerodynamics probably aren't going to change much over time. The assertion that there is a technical requirement for water chillers aren't going to shift in time from an engineering stand point.

Will the Mac Pro be able to overclock as high as the water chiller on Fury X? No. One article I saw said it was rate 500-600W. That's more than the whole MP power supply output. If not that much power going in there won't be that much power going out. Having a Fury X GPU present doesn't mean you have to overclock it.

Water chillers are just as much to get around dealing with arbitrary components and case combinations as they are about total thermal transfer capacity. That's not going to change over time either in classic PC "box with slots" designs either. It is part of the inherent constraints of that design approach.

 

[Stacc]

From the Windows side, few people expect that ATI will survive the next 12 months. Nvidia is so far ahead on the high end with relatively low power GPUs, and Intel has taken over the low end with embedded GPUs - there's no place left for ATI.

This is crazy talk. ATI is no longer a company...

Additionally, the "power distribution" load requirements for USB 3.1 and TBv3 aren't going to help either ( won't have 100W on every port (or any port ) , but going to get more power hungry stuff connected. )

Just a note on this, the 100 W of power over thunderbolt 3 is optional. This was designed with charging laptops in mind, so there isn't really a need to charge a macbook pro from a mac pro. I am sure the intended use is charging a laptop from a monitor or docking station.

 

[koyoot]

Thermodynamics and aerodynamics probably aren't going to change much over time. The assertion that there is a technical requirement for water chillers aren't going to shift in time from an engineering stand point.

Will the Mac Pro be able to overclock as high as the water chiller on Fury X? No. One article I saw said it was rate 500-600W. That's more than the whole MP power supply output. If not that much power going in there won't be that much power going out. Having a Fury X GPU present doesn't mean you have to overclock it.

Water chillers are just as much to get around dealing with arbitrary components and case combinations as they are about total thermal transfer capacity. That's not going to change over time either in classic PC "box with slots" designs either. It is part of the inherent constraints of that design approach.

First, if we are talking Fury X: Max power for the GPU is rated at 375W. Nominal TDP with stock cklock is 275W and we can expect the power draw to be less.

Now take this. Fury Nano is rumored to be 4096 GCN core GPU with around 7.8 TFLOPs of compute power at 175W TDP, which makes it run at around 950 MHz.
Making it running at 850 MHz would make it able to get to current Mac Pro design. At a 125W of TDP. And into Retina iMac.

AidenShaw, I really think you should ask people more educated than your sources, because AMD will survive not only next 12 months but couple more. The last bits about splitting AMD into two companys were spilled by, well... Nvidia, which is scared of next year node and their APUs.

Imagine this. AMD gives an APU performing like Intel Haswell core i5 with 2048 GCN cores and 2 GB of HBM, and possibility of adding DDR4. THATS what Nvidia is scared. And that is not fantasy. Yep. Haswell i5 coupled with Radeon R9 280X at a price of... 150$? Make it even 200$ and you have bargain of the century. That is reality of the next year.

 

[avkdm]

The last AMD CPU I purchased was a Socket 939 4200+, my very first build a Duron but I digress. The 4200 I believe was one of the last competitive chips AMD had that matched Intel. What makes you think this will all magically turn around next year. They havent been competitive on the CPU front for 7-8 years in my opinion. I think they should stick to discrete GPUs. This is where their future lies. To put it bluntly APU = HTPC/Laptop and thats about it - lets not kid ourselves.
I think AMD saw the writing on the wall themselves when they decided to acquire ATI a few years back.

 

[tuxon86]

Thermodynamics and aerodynamics probably aren't going to change much over time. The assertion that there is a technical requirement for water chillers aren't going to shift in time from an engineering stand point.

Will the Mac Pro be able to overclock as high as the water chiller on Fury X? No. One article I saw said it was rate 500-600W. That's more than the whole MP power supply output. If not that much power going in there won't be that much power going out. Having a Fury X GPU present doesn't mean you have to overclock it.

Water chillers are just as much to get around dealing with arbitrary components and case combinations as they are about total thermal transfer capacity. That's not going to change over time either in classic PC "box with slots" designs either. It is part of the inherent constraints of that design approach.

What I meant is that we aren't sure Apple will use the Fury based GPU yet, ergo Time Will Tell...