FCP X Share performance - rMBP 2012 vs rMBP 2013

[DavidDoyle]

Having seen a recent comparison of Ivy Bridge vs Haswell rMBP's I'm wondering whether I should upgrade. Here's my current setup:

rMBP 2012 Ivy Bridge 2.7, 16 GB RAM, 768 GB SSD

I'm using FCPX to pull together multi-camera corporate shoots of around an hours length (2 cameras MXF format, 1 x H264) and edit down to 30 minute approx each.

I just tried to output to H264 in 1080p and 720p on a test 60 minute edit and the share time was around 5 hours each (fans pretty much full blast whole time but not maxing CPU's according to Activity Monitor)

Having seen this link I wonder if the later GPU in the 2013 rMBP will make a real difference in my workflow - I may be outputting to H264, or more likely output to Prores for a master file then use Compressor to create a number of different quality versions for different delivery methods.

Having seen the share time for 2 test films, would the rMBP 2013 make a sizeable percentage difference in share time? I also have a secondary i7 Mac Mini that can be used to offload some Compressor tasks.

The change would cost me around £1000 which is a lot cheaper than £4000 or so for a nMP.....

 

[adamneer]

as it is said near the bottom of the link you provided, you will not see nearly as much gain from the GPU during a final export as you will during real time playback. i really can't imagine you would gain enough render performance to justify the expense of an upgrade to just 1 year newer. What I did, was keep my 2012 rMBP and pick up a refurb 2012 maxed out iMac, since it has a MUCH higher clock speed, and the graphics card has twice the memory and roughly 8x the number of CUDA cores (which After Effects uses for ray tracing and Premiere uses for some basic timeline scrubbing i believe). Of course, the final benefit of the iMac over another rMBP is the ability to run 32gb RAM as opposed to 16. Not to mention the 4 USB ports instead of 2, and (if needed) the ethernet port. So now I've got a super powerful computer for my main work, and when i need to work remote, i still have a very fast laptop that I no longer have to connect and disconnect from my setup at home.

 

[adamneer]

forgot to mention that you would also definitely benefit from exporting to 1080 ProRes 422 and then using handbrake to encode your h.264 files. its always nice to have near lossless masters and Handbrake is made specifically for encoding anyway, whereas exporting directly to h.264 from your timeline can be more taxing on an NLE. I can't give you speed comparisons because I generally only work in short form, and without too much effects, Premiere will export to ProRes or H.264 in less than a minute for the most part. I do most of my work in After Effects, which has a completely different render structure, and I need to have full quality exports regardless because some renders end up being imported right back into the project for further compositing. And if I'm looking at a really long render (anything over 20-30 minutes) I want to make sure that if everything looks good, I don't have to waste my time going back in and rendering a master later. You never know what you might want to use later on in a project, and always having a ProRes 422 copy will save you time and maintain proper quality for reuse, particularly in the blacks, which h.264 likes to turn into lego blocks.

 

[DavidDoyle]

Thanks for the advice. At the moment portability and space are key for me - I have a single main machine I use for work (rMBP) and the Mac Mini for home. I do a bit of editing away from home then when I come home I plug in a thunderbolt display and love the flexibility this offers when editing in FCPX.

I'll do some tests to time an output of a typical edit out to Prores, then compress for the various formats I need. When I import my MXF content into FCPX the option to transcode to FCPX isn't offered as FCPX can handle it as well as Prores for editing (though I could transcode after import) - if I transcode after importing I guess this would then massively reduce the share time for a Prores master file later on.

I'm doing about 2-3 30 minute films per week into H264 and WMV formats, and as Compressor offers distributed rendering I'm going to give that a try and compare with Handbrake (where I'd just manually copy the master file to another host and divide the compression jobs manually)

If my work doesn't require portability in the future I'll consider an iMac, but at the moment the rMBP suits very well, hence questioning whether changing to the Haswell variant would produce a good % improvement in transcoding/sharing performance due to it's later GPU with 2 GB RAM as opposed to 1 GB on my Ivy Bridge machine, and the faster PCIe based storage as opposed to SATA.

Thanks.

 

[dollystereo]

you will not see almost any performance upgrade.
GeForce 650 vs 750 is almost the same card.
Consider a Mac Pro with an ATi card. (5770, 5780 or the sapphire 7950). Or if you have cash, the new mac pro.

 

[DavidDoyle]

you will not see almost any performance upgrade.
GeForce 650 vs 750 is almost the same card.

Ok but why is there a sizeable difference in the BruceX benchmark though?

Shows the 2012 rMBP to complete H.264 encode in 124 seconds, with 2013 rMBP completing in 88 seconds. It's a 40% reduction in time.

The other video tests also show improvement - Handbrake encode down from 46 seconds to 36 seconds (27% improvement) and 31% improvement in Motion FPS.

These performance increases are not just down to the Haswell CPU over the Ivy Bridge CPU, as the raw gains between the two are more like 12%

A Mac Pro would be great - when encoding I see all 4 cores being well worked and know that 6/8 cores would make a big difference - but portability is still key for me at the moment and I thought £1000 to achieve a 40% improvement in rendering performance is a pretty good return on investment.

 

[dollystereo]

That's just a synthetic benchmark. If you really want the new machine, go ahead. It's your cash.
Anyway, If I where you , I would wait for a good retina macbook pro with a ATI card, or at least a good Open CL card. Try to find the same benchmarks with Final Cut 10.1 and not 10.0.9

 

[2ms]

Hi, I'm not really into digital video, I was just wondering why you guys are recommending ATI cards. Does FCP use the compute capabilities of ATI better than it uses Cuda or something? Unfortunately, as someone who does a lot of ray tracing, I personally am stuck using Nvidia for Cuda because Nvidia got the market cornered with iRay.