Which Will Encode Faster 2016MBP15/eGPU Pro or 2018 MBP15 with Vega 20 dGPU

[0388279]

HI All,

I currently own a 2016 MBP15 (16 GB/2.9 GHz i7/460). Encoding 10 minute videos 1080P, 60 FPS, HEVC 10 bit with FPCX/Compressor can take up to 60 minutes. I am considering the following two options to reduce encoding time:

1. Blackmagic eGPU Pro with the Vega 56 graphics card to use with my 2016MBP15, or
2. 2018 MBP (32 GB/2.9 GHz i9/Vega 20). No eGPU.

Which of these two combinations will do a faster job of encoding videos using FPCX/Compressor?

Thank you in advance.

Don Barar

 

[Trebuin]

Check your CPU use & GPU use % using something like istat menus. A lot of encoding is CPU oriented. GPU can help when you are rendering specific elements & doctoring film in a specific way. If you have 100% CPU use, then the faster CPU & more cores wins. If your GPU use is like mine (5% used) then CPU is the way to go. If your GPU uses 100% at times & the CPU is at 100%, the then GPU will help, but it is a tossup. If the CPU is not at 100% & the GPU is hitting 100%, then a fast video card would be the way to go.

 

[0388279]

Hi T:

To monitor CPU/GPU traffic I used the activity monitor. See attached screen shot. Looks like about 80-90% CPU and maybe only about 15% dGPU.

I would have thought the dGPU would have carried more of the load. Looks like the faster CPU is the way to go.

Don Barar

 

[Trebuin]

Hi T:

Thank you for your input. To monitor CPU/GPU traffic I used the activity monitor. See attached screen shot. Looks like about 80-90% CPU and maybe only about 15% GPU.

I wonder if a more powerful dGPU or eGPU would change this distribution?

Don Barar

Looks like what I saw in the past running some rendering. GPU didn't help at all...I think a better CPU might help, but given that you're not even seeing 100%, you might be fine with your current setup. There seems to be limited work for your GPU in this case.

 

[0388279]

Looks like what I saw in the past running some rendering. GPU didn't help at all...I think a better CPU might help, but given that you're not even seeing 100%, you might be fine with your current setup. There seems to be limited work for your GPU in this case.

In my videos there are videos of me, modest effects on me (color effect, motion unsharp mask effect, Hawaiki Keyer), video screen casts, and motion animations. Using FCPX/Compressor, I wonder what aspects of a video are rendered in the CPU? GPU?

Don

 

[Trebuin]

In my videos there are videos of me, modest effects on me (color effect, motion unsharp mask effect, Hawaiki Keyer), video screen casts, and motion animations. Using FCPX/Compressor, I wonder what aspects of a video are rendered in the CPU? GPU?

Don

Compression is run on the CPU exclusively. Most rendering does that. Any polygons (& effects that utilize them) can run through the GPU. The more complex the polygon, the more GPU power is required.

 

[kcrossley]

Great question. How does memory factor into all of this? For example, will a 2018 MacBook Pro with 32GB of memory outperform the same MacBook Pro with 16GB of memory when encoding a video?

 

[leman]

Doesn’t the 2018 Mac use the T2 chip fir the encoding anyway?

 

[lJoSquaredl]

Hi T:

To monitor CPU/GPU traffic I used the activity monitor. See attached screen shot. Looks like about 80-90% CPU and maybe only about 15% dGPU.

I would have thought the dGPU would have carried more of the load. Looks like the faster CPU is the way to go.

Don Barar

When I tested some stuff in FCPX it seemed like the CPU was used more for transcoding and rendering while in the edit, but as soon as export time came my GPU was maxed to like 100% and my CPU was way down compared to before. Seems like FCPX leverages them both at different times or things from what I remember.