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.
People Who Complain About the New Mac Pro Misunderstand It
- Thread starter Radiating
- Start date
- Sort by reaction score
Usually the difference in "doability" of a job comes down to RAM. The CPU just makes something go faster or slower, but if the job can't fit in the RAM and your computer starts swapping, its time to give up. Besides a second CPU, the 2 socket systems usually come with 2x or 4x the DIMM slots and motherboards that support 2-4x the RAM.
So, there are plenty of people that can't get a job done with most, if not all, single socket systems, but can with most dual socket systems. But it isn't really because of the CPU.
Again that's not true. I've used thunderbolt 1 for exactly that purpose with the most powerful Nvidia video card that you can currently buy, the Titan and saw slowdowns that are barley worth mentioning over thunderbolt 1. Anandtech has an article detailing how most PCIe cards under utilize their bandwidth by an order of magnitude.
The problem is that thunderbolt graphics drivers are not available over OS X, but in Windows 8 Bootcamp there are no issues, on the MacBook Air 2012, the iMac 2012, and the 2010 cMBP, which I've tested and ran benchmarks on.
Last edited:
Depends on the project.
For rendering we primarily use Maxwell and V-Ray, but also Octane and Arnold depending on the shot. C4D once in awhile. Lots of sims in Maya, Realflow, Houdini and others. Comps are done in Nuke or AE. Most of our footage is shot 5k on Epic so transcoding too.
We used to have everyone on 2010 12 core 2.93 Ghz Mac Pros with a few Windows boxes sprinkled in for particular tasks. Some senior artists were moved to Windows about a year ago on dual E5-2687w machines when It appeared Apple was going to skip that generation - that gave them a 60-70% performance boost. Held off on moving everyone. We've been predominantly Apple for a long time and were hoping for an update.
Our farm is all Linux so that's no problem.
Without a dual CPU model we'll likely move seniors to dual E5-2687w v2 and hand the previous ones down to juniors.
You haven't seen one person unable to do their job because of a computer that isn't out yet. On the contrary I haven't seen one person able to do their job using a computer that isn't out yet. . .
I'm not sure which part you're referring to. I've rendered plenty.
----------
I'm not sure if you're agreeing with me, but you're making the same point.
----------
You mean single processor systems that exist currently? I wouldn't necessarily lump in the new Mac Pro with "current" single processor systems, as it's surely as fast or faster than anything currently available. But it's not out yet, so we don't really know exactly. Just educated guesses.
I took it to be a challenge that you have never sat at your computer and watched a render take place, instead of wandering off to do something else.
Personally, I would be surprised if people made a habit of not watching their render at least just at the start, to make sure they haven't done something stupid. It would seem silly to let your 5 hour render go unchecked without at least watching it for a few minutes to check you didn't forget to apply a texture or something...
I'm not sure you understood my post. A 10% faster CPU isn't going to make something doable on a computer or not. It will just make it 10% faster.
RAM is really what impacts "doableness". Usually CPUs are not the bottleneck in the computer for wether or not you can do something. If the job is so big that CPU time is really the issue, you have clusters with 1000s of CPUs for that. Doubling or even quadrupling the core count isn't going to take something from taking too much time to doable when you need 100s of thousands of CPU-hours.
This isn't an educated guess, its fact stemming from the technological progression of computing over the half decade or so. The improvements in the Mac Pro will be incrimental over SP workstations with Sandy Bridge being sold today.
I get that you're mad because your business will suffer from Apple's decision, but that's no reason to baselessly bash the new MacPro with a constant hourly stream of insults.
Nobody ever said that 12 PCIe lanes was enough to handle all expansion for every user that exists. What was said is that we are getting 3 times as many expansion lanes as the last model, which is a real benefit.
The new Mac Pro has equal or better potential for expandability on paper in every way compared to the old model. 128GB of ram is not less than 128GB of ram, and arguing about ridiculously stratospheric amount of ram to try to bash a product is crazy, nobody who would actually use this computer would ever make arguments like you do.
The new Mac Pro makes for an amazing computer when it comes to a mid-level or entry level workstation, but it's not going to be competetive at the high end in a few months and I don't think Apple is trying to be competetive with the Z820 and others. They're trying to go every so slightly down market to apeal to a broader base, independent film makers small businesses etc. The machine practically screams of being targeted to these folks.
Selling 200,000 $3000 workstations is a lot better than selling 20,000 $6,000 workstations.
Nobody ever said that 12 PCIe lanes was enough to handle all expansion for every user that exists. What was said is that we are getting 3 times as many expansion lanes as the last model, which is a real benefit.
The new Mac Pro has equal or better potential for expandability on paper in every way compared to the old model. 128GB of ram is not less than 128GB of ram, and arguing about ridiculously stratospheric amount of ram to try to bash a product is crazy, nobody who would actually use this computer would ever make arguments like you do.
The new Mac Pro makes for an amazing computer when it comes to a mid-level or entry level workstation, but it's not going to be competetive at the high end in a few months and I don't think Apple is trying to be competetive with the Z820 and others. They're trying to go every so slightly down market to apeal to a broader base, independent film makers small businesses etc. The machine practically screams of being targeted to these folks.
Selling 200,000 $3000 workstations is a lot better than selling 20,000 $6,000 workstations.
Last edited by a moderator:
I think you're incorrect. I think a single CPU new Mac Pro may outperform a previous gen dual CPU. We'll have to wait and see though.
----------
Everyone makes mistakes, but eventually you have to not repeat them. I'm sure everyone has screwed up an overnight render before. It happens. But yes, I've sat and watched the render bar waiting for it to move. Luckily technology is improving and renders are not as painful most of the time. Of course YMMV depending on what you do.
Well when you throw in 2 GPUs that cost a couple thousand bucks each, and run tasks that can use them, sure.
But the gain of a Ivy Bridge E5-2697 over the top of the line Mac Pro in the previous generation in purely CPU bound tasks is going to be small.
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"
If true this would be great, but it's safe to say that many of us are (extremely) skeptical about this.
Can you post Unigine and Cinebench benchmark results for Titan vs Titan-Thunderbolt to prove it performs without any hiccups over TB?
The math would not seem to bear this out.
I wouldn't hold your breath.
For some reason he is incapable of admitting that 8 RAM slots in old Pro is more expandability than the 4 slots in new one.
Realities of math just enrage him, won't just admit an obvious truth.
Yes the iApproach to sell more MP 6.1
So how does this work? Do they lose the "Pros" when they "go ever so slightly down market?" OK, let's say they lose a % of the Pros. Let's say the $3000 is more than a % of the iMac upgraders want to spend. Let's say the lower end model is a better price and attracts "new users" but the specs are laughable base on that price.
Does this bring Apple back to where they are now if they sell 200,000 MP 6.1's? They will claim the numbers are not there. I understand the 200,000 is just a number thrown out there. What amount of MP 6.1's will needed to be sold at what price to make it a success?
I can see the Apple press now "Apple regains place in desktop market with the new MP 6.1!" But the actual number sold is of course confidential!
I just tried to approximate the choking down that TB does by moving Titan to a 4x PCIE lane in 5,1.
Uningine Valley scores took an immediate 10% hit. I tried a variety of resolutions and it was pretty much always 10% or thereabouts.
I imagine TB will also lose some more due to overhead though the Math Major may have more input. I am losing the Titan tomorrow so further tests will be with GTX780.
10% is hardly devastating. Unfortunate, but still, it's a Titan. 90% of a Titan is still playing any modern game at absolute max settings comfortably. And still likely better than a HD7970, or W9000.
So as far as I'm concerned, you've just proved that TB2 is a comparable, albeit expensive for the moment, alternative to an internal PCIe connection.
1. No such thing as TB for GPU cards in OSX, ie, what I just did doesn't prove much.
2. I ran 1 test. Doesn't represent all uses, and certainly not GPGPU.
But interesting nonetheless
It's interesting. Sadly it's not a path to upgrading a Mac Pro's graphics.
2xTitans would be faster than the internal 2xFirePros in the 2013 Mac Pro, AFAICT. However, if TB2 is already throttling a single Titan by roughly 10%, then trying to perform a Mac Pro video upgrade in the future probably won't be feasible because 2xTitans over TB2 (or whatever cards appears in 2014/2015) would get throttled a lot. Heck, TB2 can't even keep up with one external Titan, and just one Titan is not an upgrade over 2xFirePros.
Seems like internal upgrades are the upgrade path of choice for the Mac Pro, if it's even possible (which Apple hasn't committed to, at least not yet).
So, if Apple sticks with this form factor, we're left with upgrading our Mac Pros every 2-3 years if we want better graphics.
Bottom line: I still don't see how Apple did us a solid with this new approach.
(EDIT: On the bright side, it should be possible to upgrade MacBook Pro or MacBook Air graphics with an external GPU, assuming Apple ever releases Thunderbolt graphics driver software).
Last edited:
The Future of MacPro: Open boxes and PSUs on your desk
eGPU Thunderbolt tests
http://www.tomshardware.com/reviews/pci-express-graphics-thunderbolt,3263-8.html
eGPU Thunderbolt tests
http://www.tomshardware.com/reviews/pci-express-graphics-thunderbolt,3263-8.html
Last edited:
Thanks for the thread. I liked the breakdown you did, but the conclusions aren't going to wash.
The MacPro design, as it is, seems to be a right step in the wrong direction. Apple is condensing the dimensions and putting capable expansion outside the box: boffo. They have, however, neutered the brilliance of the Pro by going to a single processor. There is no single processor machine that beats a dual when it comes to user-active performance. A single processor is fine, if the machine is sitting there doing its thing with no one using it.
Apple should expeditiously produce a dual CPU model. If Intel and Apple can't meet on making such capabilities in processors or computer design, Apple should start making their own CPUs again and make proper high-end Pros which are extremely powerful, even if extremely expensive.
Unfortunately, you're failing to comprehend "capability" with other metrics, such as "value" and "utility".
And even for capability, there's no doubt that one can hang all sorts of stuff onto the tube Mac Pro. However the same was true of the old one ... It was also capable of external expansion - plus it also had internal expansion capability, and because of Apple's employment of industry standards, it wasn't stupidly expensive to do this, which made it a good value.
Strike one.
For a lot of the small operators who utilize a MP, it isn't a big Enterprise setting with a million dollar fast network storage downstairs or budgets that can drop $10K for RAM, etc. as such, a portion of the customer base is being shown the door by Apple...much like those that were abandoned with the Xserve dead ending. If it is a big enough segment that the new Mac Pro will (or won't) be a marketplace failure is a TBD... And what makes this an even bigger question is what about the Prosumer who would tend to buy this product because of a high-end hobby? He's probably not to be happy either because this new Mac Pro Is not a business tax deduction (yes, I've heard this as a higher system expenses rationalization attempt) nor does he have fibre channel storage down the hall...or other Enterprise iron of that ilk.
Strike two.
All I know is that for my IT needs, I'm glad I got a 2012...even though it lacked contemporary features: it still serves my specufic use case requirements better than the Tube ... cost NO object ... and it does it for less (bonus!). As such, I'm lucky in that I can skip this generation.
In looking forward, time will tell for just how well this product holds up. I do see that for the customer segment who pushes on GPU performance that there's huge questions on if the Tube's GPU cards will be upgradable such as in 2015, When there's new technology but it may not be supported by Apple: Will this machine become obsolete in only 2 years Because it simply can't be upgraded for those things that are important for its particular customer segment? No I'm not talking about throwing another thunderbolt external appointment upgrading the internal cards. Eventually gonna become old tech.
This may be game over.
-hh
----------
You really need to first figure out if the "business" is really a profitable one, or if it is more of a Labor of Love where most of his expenses are being recouped.
-hh
Not yet. Nothing stopping Apple bringing out drivers for them.
Indeed. However that one test is in line with many other tests that others have done on TB GPUs. And most of them show only small performance losses when using TB over PCIE.
As far as GPGPU goes, it will heavily depend on the software. Some software will run at full speed if it's not limited by bandwidth, other software might run at 1/4 speed. There's no real way to know how any particular person's tasks will perform, without trying those exact same tasks.
Benchmarking against games is easy; there's not that many super-demanding titles, and each one only has a few graphics settings to tweak. Comparing that to the limitless number of different GPGPU tasks one could be required to perform, and it quickly becomes clear that trying to answer "Is a TB external GPU good enough for CUDA/OpenCL?" becomes impossible.
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...