Imac vs Mac pro for Capture One

[george2787]

I read that a 2013 mac pro "flies" with the dual d700, one is kind of expecting that given the price but then I saw this benchmark:

http://macperformanceguide.com/iMac5K_2014-CaptureOnePro-raw-to-JPEG.html

I wasn't expecting a mobile GPU to beat 2 workstation class working together, and the SSD in the mac pro is faster... the only thing that may explain this behavior is that the imac has better driver support in yosemite (doesn't specify what system the mac pro is running) and c1 8 is not running at full potential in the mac pro.

Any insights or test will be appreciated

[deejjjaaaa]

I read that a 2013 mac pro "flies" with the dual d700, one is kind of expecting that given the price but then I saw this benchmark:

http://macperformanceguide.com/iMac5K_2014-CaptureOnePro-raw-to-JPEG.html

I wasn't expecting a mobile GPU to beat 2 workstation class working together, and the SSD in the mac pro is faster... the only thing that may explain this behavior is that the imac has better driver support in yosemite (doesn't specify what system the mac pro is running) and c1 8 is not running at full potential in the mac pro.

Any insights or test will be appreciated

just a small note, it is not one mobile GPU - you forgot that iMac has intel CPU i7-4790K with embedded GPU there... so you have 2 GPUs in both machines and real life, not synthetic GPU tests (where also nothing is rendered on screen as it is a batch processing)

[colinm]

I wasn't expecting a mobile GPU to beat 2 workstation class working together, and the SSD in the mac pro is faster... the only thing that may explain this behavior is that the imac has better driver support in yosemite (doesn't specify what system the mac pro is running) and c1 8 is not running at full potential in the mac pro.

The Mac Pro has a Xeon. The Xeon line moves slowly by design—the consumer-targeted Core line (and the iMac which uses it) has seen repeated speed bumps and updates. Right now, the Retina iMac has faster cores and more of them, yielding excellent results in CPU-bound highly-parallel tasks.

This situation is normal and will repeat itself continuously as the two processor lines and Macintosh lines play a game of leapfrog. The Mac Pro will be faster in some workloads, then the iMac and portables will be faster in some workloads, ad infinitum.

Similarly, the D700 is a beast but growing dated for some lighter-weight usages. The iMac, in comparison, has a nearly brand new GPU that excels at simple tasks and light loads.

Tick tock, ping pong, back and forth. 

[deejjjaaaa]

also the dGPU in that iMac is pretty much on par with one D700

R9 M295X = 2048:128:32 = 256bit bus = 750:800:1375 MHz
D700 = 2048:128:32 = 384bit bus = 650:850:1370 Mhz

[deejjjaaaa]

the Retina iMac has faster cores and more of them

iMac has i7-4790K = 4 cores/8 threads and that MacPro has Xeon E5-1680v2 ( http://ark.intel.com/products/77912/Intel-Xeon-Processor-E5-1680-v2-25M-Cache-3_00-GHz ) = 8 cores/16 threads = 3Ghz base/3.9Ghz turbo
and iMac has faster cores only if you run a single core - engage all 4 vs all 8 and your max Ghz will drop and then Xeon will pull ahead summarily

[lkuhlmann]

Performance of Capture One is more than the speed of the GPU. The pure processing speed in the 5K iMac and the Mac Pro will favor the 2x D700. As mentioned, the 6 core Xeon used by Lloyd is good, but so it the new Intel in the iMac. The M295X GPU is brand new, and essentially a mobile version of the HD7950 and also close in spec to the D700.

In our tests the 7950 beats a single D700, but since Capture One 8 supports all available GPU's the Mac Pro generally will beat the M295X.
The benchmark of Lloyd processes Nikon 36MPixel files to jpg files. Not the only measure of performance. Processing of Nikon files also include read, decode, "GPU process", encode and writing to file. In all a mix of CPU, disk IO and GPU tasks.

The iMac apparently holds an edge in this test, but I would think the Mac Pro will get ahead in processing of multiple IQ280 files.
-Lionel

[deejjjaaaa]

As mentioned, the 6 core Xeon used by Lloyd is good

8 core, not 6 core

[Hägar the horrible]

So why is there a difference to his former Mac Pro test? 

[Wayne Fox]

click through all the tests ... overall the MacPro is the winner.  Anything real intensive it easily wins, such as the large photoshop suite on this test http://macperformanceguide.com/iMac5K_2014-photoshop-benchmarks.html

But the higher clock speed definitely benefits some operations, and some operations do not thread very well, so more cores doesn’t help.  Here the iMac has and advantage, and overall performance of the new iMac is great.

I  agree with the observation of diglloyd of Apple not offering a single graphic card option is pretty lame.  The machine is obviously intended for 4K video work, and so they pushed the extra card and probably compromised things that might benefit other power users like photographers.

But the display is not wide gamut, so while the high resolution is nice, to me that doesn’t outweigh the benefits of more gamut. I run dual NEC 301w’s which the MacPro handles no problem.

The other big factor to me is ram, I think the iMac is limited to 32gig, I have 128gig in my MacPro and the OS never pages to virtual memory, and photoshop rarely uses the scratch disk.  I have 10 to 20 apps open at a time, including C1, LR, PS.

as to why C1 lags in this test, seems he should wait till there is a yosemite version of C1 and then run the test again.  I think C1 8 has some issues with openCL and yosemite. 

[deejjjaaaa]

click through all the tests ... overall the MacPro is the winner.  Anything real intensive it easily wins, such as the large photoshop suite on this test http://macperformanceguide.com/iMac5K_2014-photoshop-benchmarks.html

yes, seriously only in one test



"... with a file size that requires about 56GB of memory usage in Photoshop, far beyond what most users are likely to encounter. ..." :-)... I mean you buy MacPro only when you really, really need one.

[george2787]

My point is... a couple years ago you "needed" a mac pro because you could fit in work drives... now it's all thunderbolt and the imac can provide the same speed if you don't need 6 ports. Then was the output stability of firewire connection... which is also gone and if sou want to tether a digital back you're using the same thunderbolt to firewire adapter AFAIK.

Still if you need more than 32 GB of RAM or serious multicore power then the mac pro is the way to go.

So I saw the imac as a compromise solution for photographers that don't need that much, but to my surprise an imac can perform better than a mac pro in many areas that are within the normal usage for photo work for a fraction of the price. Don't get me wrong, I find that great in my situation because I can get a fast and relatively portable machine for tethering and still have cash left for a decent monitor for critical color work. 

[Manoli]

My point is...

All good points.

I think the old maxim still holds, you want to maximise speed ? (a) Max out RAM (b) use the fastest disks possible, today that means SSD's and TB2, preferably in a RAID-0 configuration (a striped volume) as a minimum - before you get into the more esoteric configs.  Buy a mac-pro by all means, just buy it for the right reasons – speed differences alone won't be one.

For photographers, the MPG tests all indicate that differences in processing speed will be outweighed by the above. For video/motion (which is really what the mac pro is aimed at) , see this comment on MPG (extract below)

quote
Transcoding video with Quicktime player is a poor test of performance for many reasons, the largest being that, unlike most pro video apps (Avid Media Composer aside), it does not leverage GPU acceleration at all, and it's implementation of multi-threading is notoriously poor ...

Really, the best test if you want to appeal to pro users is to transcode from 4k RED Camera or Arri Alexa raw files into an intermediate codec such as Apple ProRes 4444 or Avid DNxHD 345x using an application like Davinci Reslove.

If you're not trying to appeal to a pro user, then do the same test to ProRes 422 HQ using Final Cut Pro X or Compressor, which is a common workflow for the DSLR and Go Pro crowd.
unquote

Edit:
In this test, a more typical Ps CC one,  the speed difference is even less than 15% - negligible.
 

[Manoli]

OK

Qué ?

This is a huge crop that approximates 4-500% ...

(a) I don't see what relevance your post has to do with this thread, which is hardware centric
(b) Your screen grab is meant to illustrate what ?
(c) I've got no idea what file you think you've uploaded, but on my system it's showing as 'untitled.pages'

Not sure how to turn that back now... don't think I can...

re-process in C1 with the correct profile

Used screen grab to avoid any clues... at least I think so

export a 'proper' file from lightroom, use LR/Mogrify 2 - include and exclude EXIF data as you wish.

you tell me... remember this is "nose-bleed" level crop

tell you what ?

[allegretto]

I see... yes.

I meant to post to the thread about C18.0 & Yosemite

Well I can fix that. We'll just do that.

As far as re importing properly... did you click on the images?

OK, I'll move

BYW; I think  you spilled a little bit of water over there...

Qué ?

(a) I don't see what relevance your post has to do with this thread, which is hardware centric
(b) Your screen grab is meant to illustrate what ?
(c) I've got no idea what file you think you've uploaded, but on my system it's showing as 'untitled.pages'

re-process in C1 with the correct profile

export a 'proper' file from lightroom, use LR/Mogrify 2 - include and exclude EXIF data as you wish.

tell you what ?