Early Speed Tets of MacPro 2013

[Walter Soyka]

[Ronny Courtens] “we agree parallel computing is the way to go. Obviously that’s also the way Apple is going with this machine. That’s why I resent the title “speed tests of the new MacPro” while this is merely a benchmark test of the supposed CPU, not of the complete system.”

We agree there.

I’ll add that for most of us here, any synthetic benchmark like GeekBench is useless. Comparing GeekBench numbers doesn’t necessarily translate into noticeable performance gains with a specific application like FCPX.

Editorial applications in particular may be difficult to benchmark with a single score because NLEs can bang on different subsystems (storage for file I/O, CPU for compression/decompression, GPU for effects). Different editors may have totally different needs from each of these subsystems, so trying to meaningfully reduce the performance profile to a single number is probably futile.

On the other hand, for some specific users, these numbers can be a meaningful indicator of performance. Many 3D renderers are (at least for the time being) CPU-based, and a GeekBench CPU score may be a valid tool for comparing systems. (Personally, I’d prefer CINEBENCH, as it’s a real-world rendering benchmark, not a purely synthetic computational test, but that’s a separate matter.)

This Mac Pro cannot be the absolute speed champ. On CPU-specific tasks, it’ll get clobbered by PC systems which have two processors where this has one. On GPU-specific tasks, it’ll get clobbered by PC systems which have four GPUs where this has two. On storage-specific tasks, it’s matched or exceeded by PCs equipped with similar flash PCIe storage systems.

That’s not to say it won’t be a good machine overall, or that it won’t be competitive in terms of performance/cost — especially on price competitiveness, that remains to be seen. It will lag serious performance PCs by design, but it will be the fastest Mac ever and I think we can very reasonably expect this machine will be the fastest (supported) FCPX/M5 system.

Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events

[Ronny Courtens]

Well, I think we entirely agree on every point. After all it’s not about having the fastest machine in benchmarks, it’s about having the best equipment you can get for your particular workflow. And I have no doubt that FCPX and the coming family of X pro apps will truly fly on this Mac.

[Chris Harlan]

[Walter Soyka] “That’s not to say it won’t be a good machine overall, or that it won’t be competitive in terms of performance/cost — especially on price competitiveness, that remains to be seen.”

This is what I really wonder about this machine and its ultimate viability. I find it attractive, and for the right price I’d buy one. It strikes me, though, that its design might put it in a no man’s land of not quite enough power but a little too expensive to be of any interest to anyone other than an FCPX user. I really am curious about how they’ll price it. It will be quite the balancing act.

[Paul Dickin]

Hi
Does this reason for only one CPU stand up to scrutiny?

Quote:
I suspect the real reason is that Apple still {can’t be bothered|can’t figure out how} to add NUMA (non-uniform memory access) support to OS X. The existing 2-socket Mac Pros have been crippled by the interleaved-mode NUMA. (which, interestingly enough, is even hardcoded in the firmware, so when running OSes that support NUMA on a Mac Pro, you can’t benefit from that support, presumably so the benchmarks don’t shine a bad light on OS X – of course it’s blindingly obvious when compared to a same-spec workstation from HP, Dell, etc.)

…with PCIe 3, the controller for the bus is now in the CPU, which means that if you have two GPUs, you’ll start to see GPU affinity problems as well.
I’ve seen this with linux and nvidia cards, with two CPUs we would get a maximum CPU->GPU transfer of around 7-8gigs, take out a CPU and it’d bump up to 10-13.
https://news.ycombinator.com/item?id=5923300

Quote:
This gets worse under the i7/newer Xeon architecture as each PCIe card is also associated with a different CPU. So if you have a program on the wrong CPU trying to talk to the GPU, you’ll see diminished performance. One commentator noted an almost 30% speed loss in GPU performance with a dual CPU machine vs. a single CPU machine.
…this might provide some insight on why Apple might be willing to sacrifice dual CPU configurations if it could net them at least better GPU and memory performance.
Apple could have possibly recoded Mach to support a more robust NUMA implementation, but think of the risk that could introduce, and the small number of dual processor Mac users out there. Probably wasn’t worth the stability risk.
https://forums.macrumors.com/showthread.php?t=1601942