Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Except!

The benchmarks look kind of...underwhelming.

Except geekbench only benches CPU performance, not GPU.

The question is if they ran just the freebie version of GB it's not going to show you the full performance of the machine.

UPDATE:
Yep. They ran Geekbench 2.4.3 Tryout for Mac OS X x86 (32-bit). It's not the true performance.

The paid version is true 64-bit and will show you true performance, perhaps 2000 higher or more.

And geekbench still doesn't show you the video performance. What about testing it with Cinebench?
 
Last edited:
CPU's not improved in speed? IBM must be creating Scotch Mist then. The POWER7 is up to 4.25GHZwith 12 cores on a single CPU. The POWER8 about to be talked about and released will be even better.

Remember why Apple dumped the POWER chips? They didn't hit the GHz demanded by Jobs.

Do you have any idea how much power those dual FirePro GPU's draw? Didn't thing so.

The current Mac Pro has a 1000W PS, how much do you think the new one will need?

Each GPU draw between 250 and 375. So you're going to need a a similar supply in the new little box.

A bit off on the power draw argument. W9000 runs a 100watt(idle) to ~275watt(FULL BALLS OUT 100% LOAD).

The new MacPro probably ships with 750watt psu since its now a closed system. Not to mention 1000watt psu in that size of case would be an extraordinary feat of physics and engineering.
 
I think there will be a surge of iMac sales just after the Pro's pricing is announced....
A few months back a Mac Pro 1,1 could be had at 2GHz for about $500, 2.66GHz models for $700-$800... now there's no 2GHz models up for sale, the 2.66 models seem to have crept up in price... maybe people are getting their hands on the old shiny cheesegraters while they still can.
 
Exactly why did people expect a 12 core machine to be that much faster than previous 12 core machines? These are Intel CPUs and for the most part you can kind of predict how much faster a new apple computer is going to be based on the specs of the Intel CPUs. Since there are very few benchmarks out there on these new XEON processors there was a bit of mystery but still I would be shocked if they were more than 30% faster. No CPU between generations is ever faster than that unless it adds more cores. CPUs get more efficient but they don't perform miracles.

In terms of cost look up the price of Xeon processors and be prepared to be blown away. That 12 core CPU will likely be at least $2,000.00 on its own. Each of those GPUs could also be north of $1,000.00.
 
Whatever.

When the new Mac Pro finally ships, I am confident that the flagship Mac Pro (with the 12-cores and all of the bells and whistles) will Geekbench in the 35000-40000 range.

If this 23000+ was a real benchmark, I am not going to trip over a run on a prototype device running a beta OS. If it is not a real benchmark, then this entire thread is a bunch of people blowing hot air.

Apple are attempting to change the paradigm once again. We will see if they are successful. I will be bemused if future workstations by other companies resemble the Mac Pro five years from now.
 
Just read a overview of Thunderbolt 2 on AnandTech. TB 2 is still being run from PCIe 2.0 controllers, so that pretty much means the new MacPro will be using the PCIe 3.0 for the GPUs (at least the way I look at it).

TB 2 does not increase the overall bandwidth required, so that means each TB 2 port takes 4 PCIe 2.0 lanes. TB2 just combines the 4 bi-directional 10Gbps lanes to 2 20Gbps lanes.

I have no idea how this is going to work on the new Mac Pro. Does the chipset have 24 2.0 lanes?

-mark
 
Whatever.

When the new Mac Pro finally ships, I am confident that the flagship Mac Pro (with the 12-cores and all of the bells and whistles) will Geekbench in the 35000-40000 range.

If this 23000+ was a real benchmark, I am not going to trip over a run on a prototype device running a beta OS. If it is not a real benchmark, then this entire thread is a bunch of people blowing hot air.

Apple are attempting to change the paradigm once again. We will see if they are successful. I will be bemused if future workstations by other companies resemble the Mac Pro five years from now.

Incredible.

I saluted Cupertino while reading this, I was that moved.

And if the 23,000 turns out to be what it looks like, what a 12 core running at 2.7 Ghz can do, what are you going to do then?
 
http://www.promise.com/storage/raid_series.aspx?region=en-global&m=192&rsn1=40&rsn3=49
http://www.attotech.com/products/product.php?scat=29&prod=100&sku=TLFC-1082-D00
These will probably get the support. Before you balk at the price it is really only 150-200 more than a proper PCI card. Probably for the TB controller price.

Experience has shown me that once you start going into Fibre channel price really isn't a concern.

As I said, the real concern because adding another party to the mix. It's not something I am pressed about, but those that have to troubleshoot this stuff want as little parties involved as possible, or at best want to deal with one company to handle those issues.

And again, Avid will undoubtedly be certifying cards/boxes such as those (which I am already well aware of) to be used with ISIS systems.

For right now, and for what I have been told, Avid doesn't support those systems. They may shortly given the current state of affairs.

You don't throw random parts into a bottom of the barrel ISIS system which would run upwards of $110,000 for the GigE chassis, then call Avid's abysmal tech support to help you troubleshoot your issues. The first thing they will tell you is to remove any uncertified parts. If they are nice, they will ask you why you didn't get cert'ed parts and direct you to a vendor that can help. If it's the usual day, they will close your ticket and hand up.

----------

The paid version is true 64-bit and will show you true performance, perhaps 2000 higher or more.

Agreed, but that's still not very much all things considered. The HP Z820 I linked earlier may have gotten 40,000+, compared to the MacPro's 23,000, due to it's dual socket nature, but that's what many are looking for in a workstation.

Not to mention that the HP Z820 was using yesteryear's Xeon chips.

No, it has 40 3.0 lanes. Plus an additional 8 2.0 lanes through the SouthBridge.

And if it were a dual socket system, you'd get another 40 3.0 lanes from the second socket.

The Apple Itube - too little, too late.

http://ark.intel.com/products/64622...E5-4650-20M-Cache-2_70-GHz-8_00-GTs-Intel-QPI

Just fast enough to be impressive for iMac and Macbook Pro users.

I am really hoping Apple puts out a MacPro . . . . PRO. Twice the size with twice the CPU power, RAM, space for internal expansion, and actual PCIe slots.

Whatever.....

Interestingly enough, the G3 was the first real breakaway from the traditional workstation. The PowerMac G5 put it back in line with the rest of the industry albeit with half of the internal components. Apple was really taking pages from HP, Dell, IBM (at the time) and other case makers that had long been tool-less towers.

I have a feeling that Apple won't be changing anything other than the way Apple users think of workstation. By that, I really mean there will be a bunch of folks that really don't know what's out there in terms of technology being smug about a machine that barely scratches the surface of what a modestly priced workstation can do.

Kind of how we were in the G4 laptops days.
 
Incredible.

I saluted Cupertino while reading this, I was that moved.

And if the 23,000 turns out to be what it looks like, what a 12 core running at 2.7 Ghz can do, what are you going to do then?

Then Cupertino will have some 'splainin' to do.

They will have to convince their targeted audience that the Mac Pro can do what it is intended to do, or at least provide a roadmap on how they are going to get it right.
 
Yep. They ran Geekbench 2.4.3 Tryout for Mac OS X x86 (32-bit). It's not the true performance.

The paid version is true 64-bit and will show you true performance, perhaps 2000 higher or more.


That's not the point, because every other computer's score also goes higher by about 2000 points in 64bit. The whole point is a comparative one, that is that the new Mac Pro isn't terribly faster than what we can already get today.

Sure the GPU will provide additional power, but that is a non-starter since everyone has access to additional GPU's if they want to, just buy yourself a couple of Nvidia Titans (each with 6gb of VRAM) and there you go!
 
Cool. Networked with my 2010 2 2.93 12core, plus GPU speed, my new setup will SCREAM.

:apple:
 
quibble in the pronunciation of ignorance

Here's what's up -- lack of significant advancements in CPU technology is a real restriction. There just isn't much to do other than add more cores, and all those cores are hot, which then comes at the expense of clock speed. It's a tricky balance.

In light of that, Apple's decided that the big leap forward in the new generation of Mac Pros will come from the GPU side. Geekbench is a CPU-only test, so you're not seeing the important part. The important part is that on a GPU test it would obliterate any out of the box mac ever, probably by an order of magnitude. From here on out, it's all eggs in the GPU basket.

If the software creators don't take advantage of the dual workstation GPUs, it won't do anyone any good. If they do, then it'll make this new Mac Pro wipe the floor with the old model in real world use (not Geekbench).

Of course, you can always add GPUs to the old model. But it's the change in philosophy that I'm commenting on.

The main thing which many wrong approach here is very simple; childrens optimism by technological evolution should leave a long time ago. We are talking about a single processor versus two, and regardless of their up to 400 MHz lower clock wins two of its collective opponents. Someone here is obviously pretty bad mathematician when receives the difference of 1000 between the numbers 23901 and 21980. Therefore, the qualitative difference of almost 10% think that in this wider scheme of things gives a different approach. A processor with 12 cores compared to 2 * 6 cores clocked first, by none so small as some people explain here, even in view of the great number of cores 2.7 GHz is high, with the previous processors with 3.07ghz frequency.
 
No, it has 40 3.0 lanes. Plus an additional 8 2.0 lanes through the SouthBridge.

And if it were a dual socket system, you'd get another 40 3.0 lanes from the second socket.

The Apple Itube - too little, too late.

http://ark.intel.com/products/64622...E5-4650-20M-Cache-2_70-GHz-8_00-GTs-Intel-QPI

Something is going to be ham-stringed then. I'll bet MacVidCards is right in that Apple is only using 8 3.0 lanes per GPU. Hmmm. Should be interesting when this thing is released. I'm still going to reserve final judgement until then.

-mark
 
So here is a rough comparison. Instead of looking at the average MacPro 2012 score I found one that is very similar to the new MacPro that was benchmarked

http://browser.primatelabs.com/geekbench2/2014412

Notice:
- same amount of RAM
- Same version of Geekbench (2.4.3 tryout 32-bit)

Score 22577.

So the new MP, in its current state, outperforms the current MP by 5.9%. Not very impressive. We will see what optimizations can do between now and the end of the year when the retail version is available.

One other item to remember is that in order to get close to the same predicted spec of the benchmarked new MP using the current components, you are looking at a $8,999 workstation (Dual X5675, Dual GPU (5770 in this case), SSD (512GB in this case)).

Given that the current top end Xeon that would be used in a MP (Xeon E5-2470) retail for ~$1570-1900 (without the Apple mark-up), and that the W9000 GPUs described in the keynote run about $3000 in retail (non-Apple proprietary configuration), this machine seems like it will be more expensive than the top of the line MP of today.

Guesstimated prices with Apple Markup
12-core processor - $3000
Proprietary W9000 - $3000 each
512GB M.2 PCIe SSD - $750
64GB RAM $2000
Other components - $1250

Total = $13,000

Even if you use a comparable graphics card to the Radeon 7770 (FirePro W5000) the price only comes down to about $8500. Then you need to add in your external thunderbolt enclosures and devices as necessary.

Seems like the new MP is an expensive proposition.

GL
 
In all due respects the software companies have had more than enough time - Snow Leopard was released in 2009 and here we are in 2013 with vendors who still haven't cottoned onto using OpenCL, OpenGL, GCD, let alone planned ahead to migrate from OpenGL 'Compatibility' to 'Core' given that OS X Lion was released in July 2011 and not a single attempt by vendors to have some sort of strategy moving forward. If developers cannot see what the over all trend is by listening to what is being spoken about at the WWDC then I really have to ask why developers are paid so much in the first place given that the only way Apple could be more explicitly would be if they wrote the code for you.

IMHO the biggest failing of Apple is their reliance on third parties when they should be using their cash and buying out the big names, firing their management who made the stupid decisions that held back product development for OS X and push forward with products that are optimised for OS X rather than it being an after thought.

I agree that the big name software devs should be coding their apps for optimisation on OSX. OpenCL/GL have been out for some time (from what I remember at least 6 years) and are truly what makes FCPX smooth as butter on a Mac.

I am no dev, and don't know the story behind their inability to adopt those platforms...

Oh, it is quote obvious: financial Bean-Counters. What they'll do is mandate a policy of having One Standardized Code Base for the core application, minimizing the tailoring between Windows & OS X. This policy saves them money (fewer programmers) and thus optimizes their profits.

That the customer doesn't get the biggest bang for their buck isn't their priority: profits are, and they can get away with it because neither OS is sronger than the other, plus the accountants will tell them that much most of their (non-pirated) consumer base is functionally "locked in", so the way to get even the stingiest ones to fork over the bucks for an upgrade is to sunset eligibility pricing ... ie, "Upgrade pricing for Acrobat XI ends on May 30th for Acrobat 8 users" (etc).

I still see the MacPro as being a niche product. Something that won't replace a facilities' big iron systems. Something that sits in a client area to show off. Or again, something that an iMac user or Macbook Pro user can upgrade to.

I suspect that Apple thinks that they can upsell to Prosumers who are using iMacs .. but what they probably are forgetting is that this customer base probably was on the PowerMac/MP previously, but "moved downscale" in a matter of speaking becuase the iMac was a good enough performer and a better bang for the buck. As such, a reversal to an upward migration isn't likely unless the Tube Mac Pro comes in at "dirt cheap" (my SWAG is $1500 base) by Apple standards.

Apple is essentially pulling themselves out of the enterprise workstation game, which in all due respects it wasn't gaining any traction in.

True, and much of that is because of Apple's own policies - - poor stratetic communitions and "whim" based product support = unacceptable continuity.

At the same time, some of the 3rd Party vendors haven't been a friend to compensate/offset: it wasn't until CS5 that Mac-Photoshop finally went 64-bit and while Adobe will try to fingerpoint at the 64-bit Carbon/Cocoa issue, it is obvious that what really happened there is that Apple EOL'ed Carbon to force Adobe to finally modernize.

Same here. I REALLY thought it was going to be the same size, with 3.5" HDD caddies that pivoted out from the sides and such, a true tool less system like the ones we get from HP.

Once i saw the comparison I was PISSED.

Same here.

SMBs, boutique shops, indie production houses etc. will be this machines main focus. Oh, and only those indie production houses that don't plan on using shared storage systems for video post production.

Anyone whose use case exploits internal storage capabilities are going to be forced into externals, and using TB will result in a big hit on their operating budgets. Sure, one can use USB3...and many will...but that consumer decision will simply perpetuate TB being overly expensive, just as has been the history of Firewire.


I'm sorry for throwing this pretty naiv question into the round, but can someone tell me what if apple made it possible to adequately connect 2 or more of these machines via TB2 to a grid system or the like and a well written drive utilizes the additional resources??
A quick pro and con list would help me immense for better understanding.

That's an idea that has been thrown around, and the intuitively obvious approach to impliment it would have been to use XGrid.

However, XGrid was discontinued by Apple last year (July 2012, when Mt Lion launched) with no hints of any possible replacement.

My personal guess would be to look into the thick technical weeds on what's in Grand Central Dispatch (GCD) today, to see if perhaps they've snuck something into there...

-hh
 
Hi, welcome to the real world. This product is made by Apple. It will cost closer to $2,999.

The high end retina cost more. I'm guessing 4k to 5,5k depending on the VRAM of the GPU. SSD will have limited options and RAM will be overpried a lot.

So
4k for 512 SSD, 16GB RAM, 4GB VRAM (2GB on both)
5.5k for 512 SSD, 32GB RAM, 6GB VRAM
7k for 1TB SSD 64GB RAM, 8 GB VRAM
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.