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

Is the new Mac Pro a Failure for traditional Mac Creative and Professional customers


  • Total voters
    417
Status
Not open for further replies.
And Apple doesn't care at all about CUDA. None of their software uses it, and it's not even bundled with OS X. It's a third party install.

Not to mention, Apple's not going to use cards that put their own software at a disadvantage because of low OpenCL performance.

Their own software (which i only could think of ) is finalcut? What about the other tons of big players who use OpcenCL and Cuda?

Nvidia "low" opencl performance? i have seen a lot of benchmarks from all kind of programs, where Nvidia cards score quite similar and in some cases even outperform in OpenCL. The performance statement is not a verry strong one and a reason to defend the ATI only option.
 
Their own software (which i only could think of ) is finalcut? What about the other tons of big players who use OpcenCL and Cuda?

Nvidia "low" opencl performance? i have seen a lot of benchmarks from all kind of programs, where Nvidia cards score quite similar and in some cases even outperform in OpenCL. The performance statement is not a verry strong one and a reason to defend the ATI only option.
http://cdn.sweclockers.com/artikel/diagram/10130?key=6e3c39edaaa5f182742dac364baf5b54
http://cdn.sweclockers.com/artikel/diagram/10101?key=a1b152dcfc8131bcace16043c8d6f7f8
http://cdn.sweclockers.com/artikel/diagram/10100?key=08ab0a1bc1a005ddd068932b10ac531c
 
I don't care to. I have had enough bad experiences with AMD including the 2011 MacBook Pro 15" AMD GPU burning up. They are inferior and lower quality.

The nMP is all about FCP X and OpenCL. Whether you are prepared to look at and accept the evidence or not, when the nMP was being developed, then AMD was superior to the Nvidia offering of OpenCL performance. This being what Apple what was looking for when they selected the AMD.
 
  • Like
Reactions: linuxcooldude
I dont see the problem for Apple to give their user a choice for AMD or Nvidia, like many other workstation suppliers does like Dell and HP. Now they kill a whole range of users. Well, basically they say; take it or leave it. So, there i go :(
This. I have not had a problem yet with nVidia on anything. I'm not saying they are perfect but in 15 years I have had a GeForce let me down.
 
The nMP is all about FCP X and OpenCL. Whether you are prepared to look at and accept the evidence or not, when the nMP was being developed, then AMD was superior to the Nvidia offering of OpenCL performance. This being what Apple what was looking for when they selected the AMD.
I don't care. Maybe it isn't rational but I don't care. I stand by nVidia for the same reason I stand by Intel as my choice of processor manufacturers.

Even if the nVidia variant cost more I would buy it given the choice.
 
I got about 20 Dell laptops with the 8600M GPU that had the solder problem. Only two failed (one was my personal system that died about an hour out of SFO on a three week trip to Hong Kong, Thailand and Indonesia - fortunately I had brought along a ThinkPad as a backup). Probably because the Dells ran much cooler than the Apple laptops the problem was less prevalent.

The 8600M was a real screwup that affected everyone (Apple, Dell, IBM, HP,....).
 
You will not find anything for yourself on Apple platform, for a long time.
Welcome to the thread where I voted the nMP a failure. The cMP had the option and all it needed was an updated chipset/CPU/GPU and some thunderbolt ports for good measure. Apple dropped the ball.

My "Mac Pro" is a self built Core i7 5930K with 32GB DDR4 and 2 GTX 970 cards in SLI running Windows 10 Pro. I'm taken care of for quite a long time.

In the mean time I have my Mac Mini and MacBook. We do have a 27" Retina iMac in the house that has an AMD graphics chip in it. We'll see how it does. It's not my daily machine and I almost never use it but if it can keep from frying in 4 years I might rethink my position.
 
Their own software (which i only could think of ) is finalcut? What about the other tons of big players who use OpcenCL and Cuda?

Nvidia "low" opencl performance? i have seen a lot of benchmarks from all kind of programs, where Nvidia cards score quite similar and in some cases even outperform in OpenCL. The performance statement is not a verry strong one and a reason to defend the ATI only option.

I haven't seen any benchmarks in OS X where any Nvidia card outperforms a D700 in OpenCL.

If you've got one feel free to post it.
 
Currently AMD is a better choice for Apple than Nvidia. Not only is CUDA proprietary but most of their other technology as well, such as Gameworks and PhysX. AMD is more on open sourcing their technology which supports OpenCL much better than Nvidia, which is probably why its Apples current card of choice.
 
I haven't seen any benchmarks in OS X where any Nvidia card outperforms a D700 in OpenCL.

If you've got one feel free to post it.
How could a nVidia card beat it? Everything but a Mac Pro uses laptop chips or the built in integrated graphics. There has literally not been a single dual nVidia graphics solution available to the Apple OS X platform.

Here is a link bench marking the Mac Pro with two D700 cards in it. This is the link to my machine running 3D Mark. Read it and weep. 12770 vs 15847. I'm just guessing that the score is for theFireStrike 1.1 test. AMD better you crack me up.
 
Perhaps you've missed noticing that many around here think that Nvidia is a better choice for them, and think to hell with what's "better for Apple".

Most companies try to satisfy their users, not dictate to them.

Because the software applications that their using prefer Nvidia/CUDA. So the users are stuck using that brand of video card. Not really Apples problem. Instead of applications that could use both CUDA or OpenCL. Or rather make better use of OpenCL.

How could a nVidia card beat it? Everything but a Mac Pro uses laptop chips or the built in integrated graphics. There has literally not been a single dual nVidia graphics solution available to the Apple OS X platform.

Here is a link bench marking the Mac Pro with two D700 cards in it. This is the link to my machine running 3D Mark. Read it and weep. 12770 vs 15847. I'm just guessing that the score is for theFireStrike 1.1 test. AMD better you crack me up.

Gaming video benchmarks. Not sure of anybody buying a Mac Pro to play games on and not designed for them. Video games are not really pro applications.
 
I found a Apple Mac Pro with a 3.5GHz 6 for Xeon and dual D700 cards. It seems the Mac Pro does even worse then I thought. A direct comparison is here.
 
Gaming video benchmarks. Not sure of anybody buying a Mac Pro to play games on and not designed for them. Video games are not really pro applications.
People tested them. It's still a good indicator of real world performance that can be expected from these machines. But go ahead and dismiss it because you still think OpenCL is going to matter in the long run.

Edit:

I found a OpenCL benchmark that is available on OS X and Windows. I'm loading it onto my machine now. I'll let you know what the Open CL score is and see if anyone has done it with a nMP. https://compubench.com/result.jsp?benchmark=compu15d

Edit Edit:

Attached is a screen shot of the results of the Open CL benchmark on my machine and this link is the Mac Pro 2013.
 

Attachments

  • OpenCL.png
    OpenCL.png
    700.4 KB · Views: 158
Last edited:
Because the software applications that their using prefer Nvidia/CUDA. So the users are stuck using that brand of video card. Not really Apples problem. Instead of applications that could use both CUDA or OpenCL. Or rather make better use of OpenCL.
Wow. You just don't understand the difference between meeting customers' needs and furthering Apple's agenda. And only with an Apple desktop are you "stuck" with a particular video card....

Sooner or later if you continue to tell people that "you're holding it wrong" they're going to say "goodbye - I'm not holding it any more".
 
Last edited:
  • Like
Reactions: tuxon86 and Daisy81
Attached is a screen shot of the results of the Open CL benchmark on my machine and this link is the Mac Pro 2013.

Gotta run it under OS X. OS X and Windows drivers is Apples and Oranges. The OpenCL drivers Nvidia writes for OS X is what Apple is considering. Apple isn't building these machines for running Windows. That's why I specifically said "OS X" when I mentioned benchmarks.

(I looked it up for you. It's a mixed bag, but on a few key tests like video composition, AMD takes the win in a huge way: )
https://compubench.com/compare.jsp?benchmark=compu15d&did1=27734468&os1=OS X&api1=cl&hwtype1=dGPU&hwname1=AMD+Radeon+HD+-+FirePro+D700+Compute+Engine&did2=22333225&os2=OS X&api2=cl&hwtype2=dGPU&hwname2=NVIDIA+GeForce+GTX+970

The interesting thing is on stuff that is useful in games, like particles, Nvidia does better. But in stuff useful for pro apps they totally fall flat. It's like Nvidia has optimized directly for games at the cost of anything else.
 
Last edited:
Gotta run it under OS X. OS X and Windows drivers is Apples and Oranges. The OpenCL drivers Nvidia writes for OS X is what Apple is considering. Apple isn't building these machines for running Windows. That's why I specifically said "OS X" when I mentioned benchmarks.

(I looked it up for you. It's a mixed bag, but on a few key tests like video composition, AMD takes the win in a huge way: )
https://compubench.com/compare.jsp?benchmark=compu15d&did1=27734468&os1=OS X&api1=cl&hwtype1=dGPU&hwname1=AMD+Radeon+HD+-+FirePro+D700+Compute+Engine&did2=22333225&os2=OS X&api2=cl&hwtype2=dGPU&hwname2=NVIDIA+GeForce+GTX+970
Top of the line D700 cards vs mid range cards in SLI. That is nothing to boast about. Keep in mind my graphics cards in total cost ~$600.00.

Plus to compare my results to the OS X results.

Face Detection: Mac Pro: 70.387 mPixels/s Mine: 122.2 mPixels/s
TV-L1 Optical Flow: Mac Pro: 16.514 mPixels/s Mine: 23.241 mPixels/s
Ocean Surface Simulation: Mac Pro: 1277.726 Frames/s Mine: 1548.1 Frames/s
Particle Simulation - 64k: Mac Pro: 430.375 mInteraction/s Mine: 1272.2 mInteraction/s
T-Rex: Mac Pro: 6.427 Frames/s Mine: 8.5835 Frames/s
Video Composition: Mac Pro: 91.549 Frames/s Mine: 99.822 Frames/S
Bitcoin Mining: Mac Pro: 421.183 mHash/s Mine: 474.42 mHash/s

So there you have it. nVidia with a Core i7 and two mid range cards beats the dated 2013 Mac Pro with a old CPU and junk D700 cards in every single test on what is supposed to be Apple's strong suit. Then on the other tech solution the results are more like a slaughter.

Edit:

It's important to note I am not bashing Apple. Instead I am frustrated the they have an awesome platform but they are always crippling their machines and ripping people off with old chips.
 
Last edited:
  • Like
Reactions: tuxon86
Top of the line D700 cards vs mid range cards in SLI. That is nothing to boast about. Keep in mind my graphics cards in total cost ~$600.00.

I can bring it up to a 980 and Nvidia still gets destroyed in benchmarks skewed towards pro apps:
https://compubench.com/compare.jsp?benchmark=compu15d&did1=27734468&os1=OS X&api1=cl&hwtype1=dGPU&hwname1=AMD+Radeon+HD+-+FirePro+D700+Compute+Engine&did2=22357886&os2=OS X&api2=cl&hwtype2=dGPU&hwname2=NVIDIA+GeForce+GTX+980

Nvidia cards just aren't great for pro applications. That's the problem Apple has with them. Nvidia tries to hide the performance issues by using CUDA, which can't be benched against an AMD card. It's going to be embarrassing for Nvidia if AMD's new CUDA transcoder runs CUDA faster on AMD cards vs. Nvidia cards, but I think it's probably going to happen.

As far as I can tell, this is just a single D700. This is your dual cards against a single D700, and your dual cards are just barely staying ahead of a single AMD card.
 
I can bring it up to a 980 and Nvidia still gets destroyed in benchmarks skewed towards pro apps:
https://compubench.com/compare.jsp?benchmark=compu15d&did1=27734468&os1=OS X&api1=cl&hwtype1=dGPU&hwname1=AMD+Radeon+HD+-+FirePro+D700+Compute+Engine&did2=22357886&os2=OS X&api2=cl&hwtype2=dGPU&hwname2=NVIDIA+GeForce+GTX+980

Nvidia cards just aren't great for pro applications. That's the problem Apple has with them. Nvidia tries to hide the performance issues by using CUDA, which can't be benched against an AMD card. It's going to be embarrassing for Nvidia if AMD's new CUDA transcoder runs CUDA faster on AMD cards vs. Nvidia cards, but I think it's probably going to happen.

As far as I can tell, this is just a single D700. This is your dual cards against a single D700, and your dual cards are just barely staying ahead of a single AMD card.
Dude go back and read my edited post. You compared your Mac Pro to a lesser machine then mine. I compared the Mac Pro results to the results in my screen shot I got and my machine beat the Mac Pro in every test.

I realize you made the nVidia system run OS X but seeing as there is no current hardware on the Apple front that is in the same league your test is invalid and proof that Apple has lost touch with reality.

Don't give me any of that BS about only one card in the Apple system being used. This bench mark is not holding anything back on either machine. You don't get to choose to throttle the test. There are no Mac Pros with one D700.
 
Last edited:
I realize you made the nVidia system run OS X but seeing as there is no current hardware on the Apple front that is in the same league your test is invalid and proof that Apple has lost touch with reality.

There is, it's in the results database. Machine with a Geforce 980 running OS X. D700 still beats it. I attached the link.

Two 980s vs two D700s don't make a difference. The D700s will just win by double. Single, dual, doesn't matter as long as the comparison is always single vs single or dual vs dual.

You should know very well that OS X only supports one card. You tried to pull a fast one with a dual card windows benchmark to try to make up for the bad performance. If you want to compare against two D700s running under Windows, I'm sure I can dig that up, but all the numbers I've provided so far are single card.
 
There is, it's in the results database. Machine with a Geforce 980 running OS X. D700 still beats it. I attached the link.

Two 980s vs two D700s don't make a difference. The D700s will just win by double. Single, dual, doesn't matter as long as the comparison is always single vs single or dual vs dual.

You should know very well that OS X only supports one card. You tried to pull a fast one with a dual card windows benchmark to try to make up for the bad performance. If you want to compare against two D700s running under Windows, I'm sure I can dig that up, but all the numbers I've provided so far are single card.

It doenst matter how much the d700 destroys any other card in opencl if my tools of need doesnt care so much about opencl and lean on Cuda accelarators. Adobe is well known to Apple and there love/hate relationship is quite hostile. In the history of creators Apple was your choice of system, but now it seems like they swim in arrogance or something. I cant see any reason why Apple ignore a whole range of (long history) customers and kill their benifits for their tools. If AMD is yor best choice for your tools u chooce AMD. If you have more benifits with a Nvidia card because the Cuda accelarators Speed up your proces, you choose Nvidia.

If Adobe moves over to opencl only, it would be much easier. But thats not the reality, is it? The reality is that Apple doenst give people a little bit more GPU choice. HP does, DELL does, or build it yourself and it will fly. The only downside is leaving OSX. But Windows10 is a great platform to. So, there we are. I voted the nMP as an failure. I am in the run for a new heavy system that will eat my 1200dpi 4k matte paintings in Photoshop. Missing more CPU options (non xeons like i7 6 and 8-cores), DDR4 and AMD / Nvidia GPU options. This system is for sure an FCP monster. But i dont know many FCP users in my network where most of them go for the Adobe suit.

I rest my case. I simply have to move to the other side.

(ps, i have seen some comments on other forums that Adobe has dropped CUDA in Photoshop. But I havent found any articles that supports this nor that it will drop CUDA in the future. So if anyone have seen such articles, let me know)
 
Last edited:
  • Like
Reactions: tuxon86 and Daisy81
Status
Not open for further replies.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.