Reader Comments on my “Why I Probably Will Never Buy a Mac Pro Again” post: More CPU Cores and Dual GPUs vs Single GPU (and performance with PhaseONE CaptureOne Pro)
Reader comments on Why I Probably Will Never Buy a Mac Pro Again.
Chris R writes on performance with PhaseONE CaptureOne Pro:
I’ve just read your MacPro article with interest and can really see your point of view, however as you well know software that does not make use of the cores or system really let a system down in speed.
But what I wanted to mention in particular is that the Phase1 software really is quick for processing images, again as you well know Phase1 relies mainly on the GPU to process the images and indeed on my 2013 6-core MacPro can batch a 100 images from my Canon 5D Mark IV in 1 min 10 sec which is really important when working at speed shooting product shots onsite with catalogue clients.
Although the new iMacs are really quick compared to the 2013 MacPro, are they as fast at batching images with only one GPU compared the the MacPro’s two cards? Maybe you could do a quick comparative test at some point if you have already done one recently then I must have missed it.
Also, beautiful as they 5K monitor is on the iMac, editing all day using a glossy reflective screen can result in headaches and eye strain so a matte monitor is better in this regard, I guess partly that’s why you have two monitors.
DIGLLOYD: I try to avoid “points of view” when discussing performance, leaning heavily on test results from real-world tests—facts. But I do have to rely on a point of view in choosing the real-world tasks most relevant to myself as well as choosing others that I believe will have general relevance. For example, 3D gaming performance is at the bottom of my priority list (zero relevance); Photoshop performance is at the top. Then there is the whole workflow context, such as in a studio, where particular software like PhaseOne may be essential.
I’ve said for years that I would never edit on an iMac 5K display, which is why I run the NEC PA302W as my editing/proofing display to this day, including taking on the hassle of securing/stowing one in my Mercedes Sprinter photography adventure van for use in the field. Prints are difficult to match against the super-contrasty and saturated transmissive iMac 5K display, but there is also the pixel density challenge. That said, the iMac 5K is the best (most enjoyable) display available at any price for viewing images, with a free computer included. It is stunning for viewing images.
- 2.5K or 4K or 5K Display for Image Editing and Viewing?
- Can a 2016 MacBook Pro support an 8K display?.
- iMac 5K (Late 2015): Sheer Viewing Pleasure in the Fastest Mac Available
- iMac 5K for Stunning Black and White Images
- What’s the Best Way to Enjoy Images at their Finest?
- Too-High Pixel Density on 5K and 8K Displays Impedes Image Assessment
Performance—dual GPUs and more CPUs are not necessarily faster
I’ve written for years on how all that matters to performance is the performance for one’s own workflow, and also the poor state of CPU and GPU utilization with most programs.
Benchmarks and general tests often correlate only moderately to specific workflow tasks. That is why I test performance with things like raw file import and export, and real-world tasks of my own. The 2017 iMac 5K is very impressive given its single GPU and 4-core CPU; it holds its own and wins sometimes too.
A very small sampler of many years of study:
- Adobe Lightroom Performance: Inefficiencies Could be Fixed
- 2017 iMac 5K: Lightroom Import and Export
- Which is faster for RAW File Conversion, CPU or GPU? (see results, read my emphasis of what I’ve said just above)
- 2013 Mac Pro: CaptureOne Pro: CPU vs GPU on D800E Raw to JPEG
Phase One CaptureOne Pro
100 images in 70 seconds doesn’t mean anything without comparative figures on other Macs, though Chris seems to find that performance very acceptable.
I presume that PhaseOne has improved performance over the past several years, but I no longer test with CaptureOne PRO due to difficulty in getting PhaseOne to provide licenses for testing several machines at a time. I grew tired of asking every release and wasting time re-explaining and finally being told “no”. I certainly am not going to buy CaptureOne Pro just for a few days of tests, and repeat that ad-nauseum. C1 Pro is a very fine raw converter, but it has no place in my workflow for several reasons.
Update: reader Ken Scott of Digital Transitions sent me a 3-machine license for CaptureOne Pro. Thanks Ken!!! I’ll plan on some C1 Pro tests for the iMac Pro vs 2017 iMac 5K.
C1Pro had plenty of growing pains with GPU support, as did and do (to this day) Photoshop and Lightroom. Back in 2014, my 6-way test showed disappointing difference between 4 core and 8 core machines with PhaseOne CaptureOnePro. As for dual GPU and double the cores vs iMac 5K, back in late 2014 I wrote this about the test results:
In a stunning repudiation of Apple’s dual GPU strategy for the 2013 Mac Pro, the iMac 5K beats out the Mac Pro with D700 GPUs on not just Unigine Valley, but also PhaseOne CaptureOne Pro...
The 2014 iMac 5K is no speed demon compared to the 2017 iMac 5K, which is now considerably faster. And the above test used the fastest 2013 Mac Pro that could not even be purchased from Apple (custom CPU upgrade), and it is an 8 core machine, 2 more cores than the 6 core that Chris references. It is still my workhorse machine, though not for long, because by 2018 I will be transitioning to either an iMac Pro or 2017 iMac 5K.
My key point here is that assuming much of anything about performance based on CPU core count or single vs dual GPUs is an error; assumptions have a way of being proven wrong. Whatever the workflow task, it has to be proven and that proof frequently has only moderate correlation with the performance of other tasks or software, and sometimes is wildly divergent. As a sort of dirty secret, run-speed is often wildly at odds with marketing claims of GPU speed.