Final Cut Pro X: 2013 MBP vs 2016 MBP

[ACepero]

I've had a late 2013 build of a 15" MBP for about three years now. It's got the following specs:

2.3ghz
16GB
512GB
NVIDIA GPU

I use Final Cut Pro on the MBP to import, edit, and process 4K videos on Youtube. I noticed that the MBP was taking a while to upload videos and there was serious lag.

I decided to get a fully loaded 2016 MBP:
2.9ghz
16GB
1TB
Radeon 460

To make sure I didn't just jump the gun on upgrading for the sake of upgrading, I set up both MBP's side by side to "write" the same video - both plugged in, both using the same wifi connection, and with no other programs running.

The late 2013 build was processing the video much faster than the newer MBP.

How is that even possible with a slower CPU and a slower GPU?

 

[Sanpete]

Is your 2016 still indexing in the background? The 2016 should be quite a bit faster.

 

[ACepero]

Is your 2016 still indexing in the background? The 2016 should be quite a bit faster.

That's a good question. I'll have to look into that. I just brought home the computer last night, opened it up for the first time, downloaded FCPX and then did my side by side test.

Common sense tells me that the 2016 tMBP has to be faster in general, and more specifcally, with respect to graphics intensive applications and processes than my 2013 rMBP with half the GPU. But, the 2013 rMBP absolutely smoked the 2016 tMBP. I stopped the side by side test when the 2013 was at 66% and the 2016 was still on 50%.

 

[Sanpete]

Use it for a couple days and try again. By then you'll have a good idea if it's running properly.

 

[ACepero]

What can I use - benchmark wise - to ascertain performance between two MBP's? Is there a website I can visit on both units that will measure GPU performance in addition to casually using it with FCPX's workflows?

 

[SteveJUAE]

What can I use - benchmark wise - to ascertain performance between two MBP's? Is there a website I can visit on both units that will measure GPU performance in addition to casually using it with FCPX's workflows?

Running some synthetic benchmarks may make you feel better but your real life workflow is surely more important

 

[ACepero]

I ran Cinebench on both machines.

The 2016 tbMBP performed twice as better in the OpenCL test and slighter better on the CPU test as compared to the 2013 rMPB. So, I turned off Spotlight on both machines to prevent indexing and tried my Final Cut Pro X workflow again.

And once again the 2013 rMBP was significantly faster in writing the video than the 2016 tbMBP. I'm confused - I really want to like the new machine, but its like a kid that scores really well in exams but can't translate the success in the classroom to real life.

I'm going to exchange the machine and hope that I got a dud. Lather, rinse, and repeat and see what happens. :/

 

[Sanpete]

Not likely to be a hardware issue, given the high OpenCL score, so unless it's easy for you to get a replacement, I'd search a bit more for what's behind this odd result first. In real life the 2016 is quite a bit faster than the 2013 in FCPX.

I have no ideas about what the problem is, though.

 

[ACepero]

Not likely to be a hardware issue, given the high OpenCL score, so unless it's easy for you to get a replacement, I'd search a bit more for what's behind this odd result first. In real life the 2016 is quite a bit faster than the 2013 in FCPX.

I have no ideas about what the problem is, though.

Easy is a relative term - I'm still under the 14 day return window. I picked up the machine this past Sunday. Been working late so I've only gotten to test it over the past two nights. I'll keep digging though....

 

[Boneheadxan]

Why not try a new project on the 2016 machine and then test that against the old machine?
I have both these machines as well, but I haven't gotten around to moving my sessions (LogicX) to the new machine.

 

[ACepero]

Why not try a new project on the 2016 machine and then test that against the old machine?
I have both these machines as well, but I haven't gotten around to moving my sessions (LogicX) to the new machine.

That's essentially what I did. I created new libraries and imported the same set of videos to both machines. Added a generic title to the beginning of each video and initiated the process of exporting the video to Youtube in 4K and in "best quality" mode.

I thought that maybe it was due to the fact that I am on beta version of Sierra on the old machine - so I just loaded that onto the new machine. Still the same.

 

[ACepero]

I did some reading and decided to try another test - instead of a workflow that uploads straight to Youtube, I tried to strictly render a small clip with 3D effects.

The results were finally what was expected - this time the 2016 tMBP destroyed the 2013 MBP in rendering the video. Not sure what the difference is between what I was doing before and this because I'm still learning my way around FCPX - but I was satisfied with the results.

I'm going to do some more comparisons over the weekend.

 

[tCC_]

I did some reading and decided to try another test - instead of a workflow that uploads straight to Youtube, I tried to strictly render a small clip with 3D effects.

The results were finally what was expected - this time the 2016 tMBP destroyed the 2013 MBP in rendering the video. Not sure what the difference is between what I was doing before and this because I'm still learning my way around FCPX - but I was satisfied with the results.

I'm going to do some more comparisons over the weekend.

Isn't it that editing video without effects only using cpu and not the gpu ?
While using effects like u just did enabled the gpu and thats why it's faster ?

For rendering that is.

 

[ACepero]

I checked the fcp.co site and tried out the Bruce X benchmark.

The tbMBP averaged 19.40 seconds over three attempts while the 2013 MBP averaged over double that. I think that answers any questions I had about the performance of the machine.

 

[jackoatmon]

there's some explanation for this

the 2016 is sooooooooooooooooo much faster in FCPX than previous macs

it's kinda astonishing actually

maybe like people said still indexing or something, or perhaps there's a problem with the machine

 

[Sanpete]

there's some explanation for this

the 2016 is sooooooooooooooooo much faster in FCPX than previous macs

it's kinda astonishing actually

maybe like people said still indexing or something, or perhaps there's a problem with the machine

@tCC_ has the explanation above. Some FCPX operations rely only on the CPU, some also draw on the dGPU. The former are sometimes slower on the new machines, the latter much faster.

 

[diamond.g]

For the CPU difference you could run Intel's Power Gadget on both systems to see if it is throttling the clock. The newer system should have a higher Turbo but it may not be able to hold it for as long.

 

[jackoatmon]

@tCC_ has the explanation above. Some FCPX operations rely only on the CPU, some also draw on the dGPU. The former are sometimes slower on the new machines, the latter much faster.

No he's definitely wrong. It's much, much faster on all operations.

 

[Exile714]

No he's definitely wrong. It's much, much faster on all operations.

No, he's not "wrong" because he's trying to explain a direct observation he's had. In the situation he first described, the 2016 was indeed slower. This isn't one of those "I hate everything Apple made since Steve Jobs" discussions, and the guy seems very willing to work through this issue with everyone here.

Anyway, another explanation might lie in the weird wireless (WiFi and Bluetooth) issues the 2016 MBP seems to have. I don't know how FCPX works with render and upload to YouTube, but *some* people have seen WiFi performance drop when the USB-C ports are seeing high usage, so maybe there's an issue with the bus?

 

[Sanpete]

No he's definitely wrong. It's much, much faster on all operations.

There's ample evidence that the older MBP quad cores are faster in some operations than the current ones.

 

[samcraig]

What has been proven is something people need to pay attention to closely. Mileage may vary. It really depends on what you are going to use the device for. Better specs may not equal better performance if you're not engaging in activity that leverages the newer/better tech.

It's not just FCPx - but with everything.

 

[diamond.g]

Using Intel's Power Gadget I have noticed that my 2.7 2016 MBP doesn't appear to sustain over 35W power usage. I ran a GeekBench 4 test. Can anyone using FCPx run the tests but also have the Power Gadget running to determine if they see power usage over 35W. What is odd, with a lower power usage, we should see the max speed stepping. Can anyone confirm?

 

[Sanpete]

Using Intel's Power Gadget I have noticed that my 2.7 2016 MBP doesn't appear to sustain over 35W power usage. I ran a GeekBench 4 test. Can anyone using FCPx run the tests but also have the Power Gadget running to determine if they see power usage over 35W. What is odd, with a lower power usage, we should see the max speed stepping. Can anyone confirm?

Notebook Check got the 2.6 15" up to 89.5 W. That was on an earlier version of the OS, but I doubt it's changed that much. Are you testing on battery?

 

[diamond.g]

Notebook Check got the 2.6 15" up to 89.5 W. That was on an earlier version of the OS, but I doubt it's changed that much. Are you testing on battery?

On Mains. Just set it up and my photo library download and spotlight are happening at the same time.

The power gadget only looks at the Intel side of things. The concern is that Apple may be using the cTDP which may come with a lower top end sustainable speed than the 2013-2015 models. Which could explain why CPU performance is lagging in some tests.

 

[Sanpete]

The concern is that Apple may be using the cTDP which may come with a lower top end sustainable speed than the 2013-2015 models. Which could explain why CPU performance is lagging in some tests.

There has been some evidence that the turbo, which was already spec'd to be more limited than before, may have been further curtailed, though this hasn't been confirmed by a lot of evidence yet.