Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Cost may not be a main concern, but consider if you wrote in a line item for a grant to buy a Titan (which is a decent entry-level scientific computing card), whether or not its expensive, the price has just changed,

Not really. If the grant mentioned a specific Titan and a specific Mac Pro to get into, then the current Mac Pro still exists. And will exist as a refurb /used after the next one comes out. Every single current Mac Pro currently in the sales channel isn't going to wink out of existence in the next 5-6 months.

If the grant was just for the card then more likely already have the current Mac Pro.

So costs .... the same.


For grants not already in the hopper going through approval processes it is precisely the wrong time to include the Mac Pro at all, Titan or not, into a grant process. The device cost is unknown. The performance on grounded real world benchmarks is largely unknown. The depth and scope of the version 1 bugs/glitches is unknown .


In 8-12 months the cost factors on Titan class performance probably would have shifted, so the next cost for that level of performance is unknown too.
 
...
2x LGA 2011 sockets
8x RAM slots
3x PCI-E 3.0 x16, 1-2 x8's
USB3
4-6 3.5/2.5" Sata III bays
Thunderbolt - sure why not
Similar case to current Mac Pro.

5 slots probably would not fly at all even if they had stayed with the rectangle with slots basic design.

While the dual package allow you to provision that many lanes, a single package version would not. The Thunderbolt solution would soak up another 12-20 lanes ( x8-16 for embedded GPU and x4 for TB controller).
Once start to push to divide the single and dual package models into less symmetry Apple would likely pick one and go with it.

As long as the dual package models are piggybacking on the single package volume they were viable. When they both were going down in volume.

The HP z820 or Dell T7600 class boxes Apple is clearly moving away from. Even if had committed to rectangular design it was going to be much more of a disconnect to make the single/dual subsets work with the same general box. Especially if throw the Thunderbolt curve ball in. [ Not that many other PC vendors are picking up TB quickly, but some of the last will probably be the large box workstation crowd. ]


This design, we'll probably get 1 or 2, for ingest, and depending on how things go might keep one for Smoke. So if we figure $6K per box that is 150k lost sales.

$150K / Mac unit sales revenue for a quarter = what percentage?

Frankly Apple killed off far more than 20-200 units sales of Mac Pros just with their Osborne effect announcements (and EU sales ban ).

It is not a short term game. Apple is not primarily after the $6-9K boxes. How many $2k-4K boxes is a bigger issue.
 
No reason we can't use two titans connected separately on two different TB ports. And I'm sure no-one has yet tried SLI on two external GPUs on two different TB ports. There should be plenty of potential for beefing up the new Mac Pro's graphics capabilities, and/or GPGPU. IF one is willing to spend that much money...
That's a good point that I hadn't thought of. I think my brain shied away from the idea because:

1) It means having two external Thunderbolt expansion boxes, two TB cables, possibly two external power bricks, etc -- thus turning your desk into a very spidery mess of cables, boxes, etc.

2) it would also cost several thousand dollars to buy two TB expansion boxes (NOT cheap), two Titans, two TB cables, etc. -- doubling the original investment in your computer.

But you're right, it probably could be done. By the time you buy all that, it might be cheaper to just wait and upgrade to a newer Mac Pro though. (Or switch to a PC, which already supports two internal Titan cards, heresy I know. :))
 
People costs are almost always the most expensive thing in any business. And I don't know if you were exactly using that fact to justify increasing hardware costs or not, but I would disagree if you were. Just because you spend 10x as much (or 100x) on item A over item B, that doesn't mean an additional X% of the cost of item B is trivial or inconsiquential. And you mention the very reason that is, opportunity cost. If you can get basically the same machine for $1000 less, that is $1000 you can put somewhere else to help you make more money.

The point is the hardware cost is the least of practically all costs. I touched on this in a previous post regarding in our industry, high end audio, crew, logistics, software and training costs are far more than the hardware. Additionally, the computer even at five grand is the least of the hardware costs for apps that may need that much horsepower.

The grand savings you mention doesn't take into account any other cost arbitrage/benefits of spending that grand. For us it's a considerable reduction in the price of proprietary hardware in DSP cards and interfaces and migrating equipment that can be more easily used with other software packages. There are also efficiencies in configuration and system scaling that weren't possible prior. We are doing this with current MPs, MBP and Minis. This isn't speculation or theory, it's a proven business/technology model that we've been using for the last several years. We're at an large entertainment brand with a significant international footprint. That said, what works for us isn't necessarily going to work for other disciplines and probably not everyone else in audio. However, at the pointy end of the spectrum you'll find that the methods and practices we are using are quite common in our field.

The handwringing and consternation in the thread is overblown as I'd bet no one commenting as used one and we don't yet know the cost or how they will really perform. When it comes out check it out and see if it fits the intended app. If not move on to a tool that will.
 
So we're back to saying that 4 >8 again. I really thought we were making progress.

Oh well.

This is why I keep posting, people posting nonsense like this.

4 isn't greater than or even equal to 8, stop trying to find a way that it is.

Why not just say "In many ways I think it is more expandable, except of course for RAM where it holds half as much"

MacVidCards, your repeated personal attacks and fixation on argumentative nonsense indicate a profound disinterest in being impartial and rational about your views. You are someone who clearly starts with a viewpoint and then tries to prove it instead of the other way around.

Regarding the 4 vs 8 slots in terms of expandabity, you are trying to compare Apples to Oranges.

The new Mac Pro has 4 32GB max ram slots, the old one has ZERO 32gb max ram slots. Read that a few times if you don't get it.

You are doing the equivalent of saying "You have 4 apples, and I have 8 oranges therefore you have half as many oranges as I do"

That is ridiculous. A better comparison would be "You have 8 pounds of fruit, and I have 8 pounds of fruit therefore we have the same amount of fruit". Or "The old Mac Pro can have 128GB of RAM and the new one can have 128GB of RAM therefore both can have the same volume of memory"

"Expandability" is how much room you have. You're basically trying to say "I have eight 5 gallon buckets and you have four 10 gallon buckets therefore you have less room for stuff than I do because 8 > 4!". That is a silly premise.
 
Last edited:
No misunderstanding. This is going to be a flop. You could write a book on why this may not be the case, that will not change the fact that this will be a flop.


I think there's a fundamental misunderstanding about most of the compalints about the new Mac Pro.

Before we begin though let's address the cost, thunderbolt expansion is expensive because it's new not because it's expensive to make. I own an Sonnet Echo Express Pro, that cost me upwards of $700. The parts that go into it are marked up over 800% and the market will tank in price very quickly considering you can buy the pcie to thunderbolt controllers for $80 right now, but Intel won't let consumers buy them, only manufacturers.

The first thing to understand about Apple's New Mac Pro is that it had to have a single processor, the chipset only supported a single processor at this time with thunderbolt, so if Apple wanted thunderbolt they had to go with a single processor. However you can have up to 6 external Xeon Phi co-processors, and the single processor has as many cores and is faster than the dual processors it replaces.

Second, Apple had to go with Firepros instead of Nvidia cards, AMD is far less interested in marking up their workstation cards, and sure you don't get CUDA core support, but if you did then you'd be paying $2000 more for the machine. Instead now you can have both the internal GPUs and up to SIX external ones. I've attatched a GeForce Titan with a PCIe to Thunderbolt adapter and VisionTek 3.5" 450w stand alone power supply to practically every Mac that Apple makes that has thunderbolt and it works extremely well. If you're actualy running applications that need CUDA cores, having 6 external cards sold at market price is a huge blessing, over 2 internal ones that are sold at a mark up.

Third, external PCI cards, external hard drives, and external optical drives are the way of the future. Why?

Flexibility:

Image

The last generation Mac Pro had to make an educated guess on how much EMPTY space you needed, and when you were done using that empty space, there was NOTHING that you could do to expand it without an engineering degree and a manufacturing plant. Apple had to guess that you needed 4 PCIe slots, if you needed 5, well too bad, and if you didn't need those slots you were wasting space.

The new Mac Pro does away with the guessing game and gives you the option of nearly unlimited expandability. You can have up to 60 PCIe slots now, which is 15 times as many, and up to 300 hard drive arrays or optical drives, which is 75 times, and 150 times more respectivly. And guess what, the Mac Pro is smaller. How is that not revolutionary?


Fourth, the biggest issues of all that has come up. Clutter. Cables, Mess, Junk. Hoestly I think this is a very short sighted issue. If it REALLY bothers you that much, then just buy a recycled Mac Pro case off eBay for $100 and some zip ties you can have all the internal parts you want. However, I haven't heard of a Mac Mini or an iMac user complaining, and guess what? Companies will come up with neat little solutions to package all of your external devices around the new Mac Pro, and here's the thing, if you ever decide that you need even more room, you won't be SOL, you'll simply buy a new dock or stand for your Mac Pro with more room.


The new Mac Pro is about flexibility and scalability, unlike the last version which was one size fits all that didn't really do a good job at fitting anyone except the exact person who needs 4 hard drives, 2 optical drives and 4 PCI slots exactly.

Image


The Mac Pro is the future of computing.

Finally, the biggest issue of all, performance the new Mac Pro can meet the needs of literally anyone who owned the last generation Mac Pro, every single performance and expandability criteria is higher. Everything you could do with the last model, you can do equally or better with the new one.

Image

You'll notice that not only are all performance metrics improved across the board, all of your expandability capabilities are improved across the board. Instead of having 2 PCIe 2.0 x16 slots used normally for the graphics cards, you have 2 PCIe 3.0 x16 slots used normally for graphics. Instead of 4 PCIe 2.0 lanes for expandability (shared between slots 3 & 4) you have 3 PCIe 2.0 lanes shared between 6 thunderbolt cables. That's 3 times the expandability before we consider daisy chaining, and you can use full x16 cards at x4 speeds just like before. On top of that there are several Mini PCIe connectors internally, not to mention USB 3.0.
 
.....
But what really amazes me is this type of design has be bandied about on this and other forums I peruse (I have just be lurking about), and it seemed to me most people expressed that they really didn't like the idea of this type of design -- yet this is exactly what Apple produced. You would think that they would use forums like this to do market research :rolleyes:…

These forums are probably the last place do market research on feature selection. Research into what FUD is likely going to be thrown at a new design but it is generally a poor sampling when consists of threads of running commentary.

Tapping into deployed configuration sampling. Would so more relevant information. For something like

http://update.omnigroup.com/

Hardware -> Memory

or

https://forums.macrumors.com/poll.php?do=showresults&pollid=10203


for stats of what a broad sampling of what folks actually use. There are often occurrences of "the sky is going to fall if I can't install 98GB of RAM" and then when look at the number of folks even over 32GB, it isn't that large. Apple probably even has probably better numbers if just track configs attached to bug reports.


I'm sorry, my experience has not be positive with TB expansion -- that's not to say my experiences aren't unique.

Thunderbolt has be grossly oversold with some people. It is has be badly mismarketed ( undersold or just plain badly targeted) with other groups.

Apple's typical Marketing Hyperbole doesn't really help ( Thunderbolt is fastest I/O technology there is ? Not even close. )

The Mac Pro has an abnormally high number of Thunderbolt ports probably because some of the intended use will not be as a Thunderbolt port (i.e., to propagate a TB daisy chain. )


I'm glad you said "on paper". In theory this is a great design. In every tech adventure I've been on it has never turned out as good as it looked "on paper" -- never!

But that cuts both ways. There are folks saying Thunderbolt and focus on GPGPU for number crunching can't work "on paper" also. There are tons of "paper" corner cases that folks can chase after. If those corner case correlations to the full spectrum of deployment is low it doesn't really have much impact.
 
MacVidCards, your repeated personal attacks and fixation on argumentative nonsense indicate a profound disinterest in being impartial and rational about your views. You are someone who clearly starts with a viewpoint and then tries to prove it instead of the other way around.

Regarding the 4 vs 8 slots in terms of expandabity, you are trying to compare Apples to Oranges.

The new Mac Pro has 4 32GB max ram slots, the old one has ZERO 32gb max ram slots. Read that a few times if you don't get it.

You are doing the equivalent of saying "You have 4 apples, and I have 8 oranges therefore you have half as many oranges as I do"

That is ridiculous. A better comparison would be "You have 8 pounds of fruit, and I have 8 pounds of fruit therefore we have the same amount of fruit". Or "The old Mac Pro can have 128GB of RAM and the new one can have 128GB of RAM therefore both can have the same volume of memory"

"Expandability" is how much room you have. You're basically trying to say "I have eight 5 gallon buckets and you have four 10 gallon buckets therefore you have less room for stuff than I do because 8 > 4!". That is a silly premise.

So even if your comment about the memory limits of the existing Mac Pro was correct, that would make your original thesis incorrect.

"...all of your expandability capabilities are improved across the board."

Last I checked 128 was not more than 128.
 
No misunderstanding. This is going to be a flop. You could write a book on why this may not be the case, that will not change the fact that this will be a flop.

Without any concrete pricing it is extremely premature to declare it a flop.

There are lots of complaints that really boil down to just being about costs. The ones about sunk costs aren't really going to change the results over the short term. However, the ones about new costs are really largely incomplete with the Mac Pro system price locked down. Folks who want it to fail snag the high estimates and those in zealotry to defend and spin it are on the opposite extreme end of the spectrum. Neither one is probably correct.

It is certainly the case that a fraction of former Mac Pro buyers are not going to buy it. But frankly those folks couldn't push the current form factor into the "interesting enough growth" to keep Apple interested in continuous Mac Pro development. So it is highly likely they aren't the critical factor with the new design either.
 
The new Mac Pro has 4 32GB max ram slots, the old one has ZERO 32gb max ram slots. Read that a few times if you don't get it.

Who makes 32GB single DIMM memory modules? ZERO me thinks. I could be wrong, but couldn't find any.
They just recently got 16GB variants.
In a few years they may make some priced out of this world. The bad taste for me if there is any is the small factor almost guarantees added cost due to the limited (more limited than 2010 anyway) slot options. Both memory and the anemic TB offerings. Which in no way compete with the actual PCI offerings. 100 to 1.
Change will likely happen and things will get created. My hats off to you super-fans that beta test all these "future boxes".
 
So even if your comment about the memory limits of the existing Mac Pro was correct, that would make your original thesis incorrect.

"...all of your expandability capabilities are improved across the board."

Last I checked 128 was not more than 128.

Very true :) But 1866 Mhz is more than 1333 Mhz, which means that the capabilities of this area of expansion have improved. There's a reason I worded it this specific way because there are a few cases where the new Mac Pro has the same room for expansion as the old Mac Pro but there is a spec bump, which is important to consider. Every aspect related to expansion is either faster or gives you more room. And every core performance criteria is boosted.
 
Very true :)Every aspect related to expansion is either faster or gives you more room. And every core performance criteria is boosted.

It is amazing what 3-4 years of other hard work makes...
It does not 'give' you more room. It forces you outside. It is a pure defection argument.
It is faster expansion but it is in a pretty useless form right now. Even 2 real form PCI slots could have stopped this entire thread from existing.
I don't understand the complete haters and I don't understand the complete lovers of this nMP. It needs to exist 1st. It needs to be priced 2nd. And Intel and Apple need to push TB hard. I am glad something was announced. Even if it is hard to swallow.
 
Last edited:
Who makes 32GB single DIMM memory modules? ZERO me thinks. I could be wrong, but couldn't find any.

Not sure where you were looking.

http://www.crucial.com/store/mpartspecs.aspx?mtbpoid=6768B191A5CA7304



They just recently got 16GB variants.

Depends upon what recently is. There has been 16GB for a while. That is what has been used to push the current Mac Pro to 96GB ( and 128GB if run windows) limits.

If they is Mac Pro and relative to the 2009-2012 time frame then recently.


In a few years they may make some priced out of this world.

may not take as long as you think with DDR 4 coming with Xeon E5 v3 ( Haswell ) late next year. These DDR 3 will be mainstream and the higher markup modules will be the DDR 4 ( which actually have higher pressing need for higher density... so the volume produced is going to go up. )


The bad taste for me if there is any is the small factor almost guarantees added cost due to the limited (more limited than 2010 anyway) slot options.

There isn't going to be multi-level banks of cheaper ( generation ago ) RAM in upcoming E5 v3 boxes from competitors either.


Change will likely happen and things will get created. My hats off to you super-fans that beta test all these "future boxes".

Not just super fans. Some folks have already moved bulk storage outside the box because for the sizes required it would not fit in a single box anyway.
 
Who makes 32GB single DIMM memory modules? ZERO me thinks. I could be wrong, but couldn't find any.
They just recently got 16GB variants.

There are several manufacturers. You're probably thinking of SODIMM memory, the old and new Mac Pro use 240 pin memory, and 32GB modules just came out for it. Currently as 32GB modules are a new product prices are pretty high (twice as expensive per GB) but prices should come down in the near future.

In a few years they may make some priced out of this world. The bad taste for me if there is any is the small factor almost guarantees added cost due to the limited (more limited than 2010 anyway) slot options. Both memory and the anemic TB offerings. Which in no way compete with the actual PCI offerings. 100 to 1.
Change will likely happen and things will get created. My hats off to you super-fans that beta test all these "future boxes".

Actually 4x16GB kits are cheaper than 8x8GB kits currently so there is no added costs if you're not going to be using the new 32GB dimms.

Other than that thunderbolt offerings can compete with PCIe offerings because Thunderbolt literally is a PCI Express cable and converting from thunderbolt to a standard PCI express slot is incredibly trivial. One of the first Thunderbolt products you could buy was a box to convert between Thunderbolt and PCIe x4 (there are x16 boxes available now too). You can literally put any PCIe card you could put in the previous Mac Pro in the new one and then some. All you need is a PCIe box to connect to.

These boxes are incredibly overpriced at this time, but the parts that go into them are also very cheap - complete thunderbolt to PCIe x16 controllers run $80 to aftermarket manufacturers so once the smallest amount of competition comes in to play you will see these boxes selling for peanuts.

So you can use all of your previous PCI cards at the same speeds you used to use them as, and you can have a ton more in the new machine.


Except CPU.

And some RAID scenarios. (The flash storage is about 40% as fast as what we currently run internal.)

That's a reasonable point but I don't think the CPU isn't something you expand it's something you replace. The storage expansion has improved though as you can now run several 2600 MB/s multi terabyte drives over thunderbolt, either through a few RAID arrays or by using high end PCIe drive like the OCZ Z-Drive in PCIe boxes.
 
Except CPU.

Eh? The Xeon E5 v2 1600 series offerings are not a major improvement over the Xeon 3500/3600 in the current offering?

Across the majority of the upcoming line up change that isn't really going to hold up.


And some RAID scenarios. (The flash storage is about 40% as fast as what we currently run internal.)

Run internal with re-routed sata/sas connections to a PCI-e RAID card?
 
I have a Titan in a PC. ..Cost is a concern to me. ..My PC costs less than $3,000.
What will be the price of a nMP running a Titan?

Cost may not be a main concern, but consider if you wrote in a line item for a grant to buy a Titan (which is a decent entry-level scientific computing card), whether or not its expensive, the price has just changed, and its nigh impossible to come up with the additional money.

Beyond that, just because I'm spending a thousand dollars on a card doesn't mean I'm not going to be annoyed as spending a few hundred more.

I refer you to my point:

I never said it as the best possible scenario, but it IS possible. and it IS practical.

If you can't justify the extra cost of a TB enclosure after buying an extremely expensive GPU (one that does NOT have a very good performance vs cost ratio), then build a hackintosh for it, and forget the new Mac Pro. The fact that it is an expensive option does not change the fact that it IS still an option.

I run X-Plane. ..10% difference in frame rate could be the difference between the plane flying or the plane crashing into the ground.

Lower the graphics settings. Or build a hackintosh.
 
Eh? The Xeon E5 v2 1600 series offerings are not a major improvement over the Xeon 3500/3600 in the current offering?

Across the majority of the upcoming line up change that isn't really going to hold up.

Run internal with re-routed sata/sas connections to a PCI-e RAID card?


Referring to 12 core models.

My current internal RAID runs 3 GB/s.

Actually like the new machine for what it is, but disappointed they are no longer going to make high end machines. They used to care about performance.
 
Thanks all but "overpriced" seems to be the word de jour. I am sure it will come down. That is at least one constant.
I even checked Crucial for 32GB DIMMS. :eek:
$1000.00? Yeah. I like the kind for $40.00.
I was not thinking SO-DIMMS I really couldn't find any in the 10 mins I looked. Oh well, I failed Google today. Feeling like a Dinosaur because it seems like 8GB just came out.

Yup. The "box" with PCI slots and the extra cost is my worry. But again, prevalence can drop prices.
I might just start having faith in this little Final Cut X - Xbox. Maybe Apple will make a Logic Pro Xbox too:D
I will be getting 10-20 on launch so I will find out soon enough. Complete with an assload of converters.
 
The storage expansion has improved though as you can now run several 2600 MB/s multi terabyte drives over thunderbolt, either through a few RAID arrays or by using high end PCIe drive like the OCZ Z-Drive in PCIe boxes.
But 20Gbps is only 2500MB/s, which is only achieved by running two TB cables today. I'm wondering why you think you can run "several 2600 MB/s" data streams today over TB1.
 
Not just super fans. Some folks have already moved bulk storage outside the box because for the sizes required it would not fit in a single box anyway.

Not the storage outside the box. The Thunderbolt only part. I have fairly large scale Xsan deployments. I get that part just fine. It is the consumer nature of the TB controller and the need to introduce more failure points and chatter with adapters that is a little off-putting. I am just cranky. I will deal with it. Probably.
 
But 20Gbps is only 2500MB/s, which is only achieved by running two TB cables today.

It does not. Thunder ports are not additive. The Thunderbolt backbone network of which two host ports represent either side of the backbone is limited to 20Gb/s aggregated for data and video bandwidth total. If talking just PCI-e data then it is 10Gb/s on the back bone. That's it. It is less than the x4 PCI-e bandwidth from the controller to the host.

So there is no way, no how 2500MB/s is coming without splitting the load onto mulitple TB networks leveraging multiple TB controllers.


Even TB v2 is probably not going to do 2500MB/s. You'll be closer to the x4 PCI-e v2.0 therotical maximum. It likely isn't going to be able to grab all 20GB/s . The removal of the border is more so to all 4K video traffic to be a roadhog, not necessarily to hand all of the bandwidth over to PCI-e data traffic.
 
MacVidCards, your repeated personal attacks and fixation on argumentative nonsense indicate a profound disinterest in being impartial and rational about your views. You are someone who clearly starts with a viewpoint and then tries to prove it instead of the other way around.

Regarding the 4 vs 8 slots in terms of expandabity, you are trying to compare Apples to Oranges.

The new Mac Pro has 4 32GB max ram slots, the old one has ZERO 32gb max ram slots. Read that a few times if you don't get it.


Read this a few times and see if you get it:

http://www.kingston.com/us/business/server_solutions/best_practices/maximizing_memory/

Says the Xeon 5600 supports 32GB DIMMS. The memory controller for X5600 chips is on CPU. Last question remaining is for someone to try.

So it is just as likely that a 2012 Mac Pro can use 32GB DIMMS as a 2013.

So we are back to 8 > 4 until someone says they've tried a 32GB DIMM that Kingston says will work and it doesn't.
 
This thread reminds me of when Apple announced they were moving to Intel. People were pissed. They wanted a 3Ghz G6, but it never came. It's probably going to turn out fine. If you've never had over 10 cables coming out of the back of your Mac Pro, you're probably using it wrong.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.