2018 MacBook Pro purchase

We have to replace the current MacBook Pro (late 2013) we’ve been using with mimoLIve and we’re hopeful for some advice in selecting features for the 2018 model. In particular, we’re confused as to which graphics card to select. There are 3 options:

Radeon Pro 560X with 4GB of GDDR5 memory (standard)

Radeon Pro Vega 16 with 4GB of HBM2 memory (+$250)

Radeon Pro Vega 20 with 4GB of HBM2 memory (+$350)

Should we be more concerned with the cpu?

2.6GHz i7 (standard)
2.9GHz i9 ($300)

Or the memory?

16GB (standard)
32GB ($400)

I would guess the advice would be to max out in all areas, but I don’t want to spend the money if it’s not necessary. Any guidance or shared experiences with the 2018 MacBook Pro would be welcome.


@alternativeplan Happy New Year! We’re looking forward to hearing your feedback on mimoLive in 2019!

It remains very difficult to make a definitive recommendation as to your MacBook Pro and the “max out” option is certainly the easy way out.

mimoLive normally doesn’t tax the GPU very much except for decoding some types of video sources. The biggest performance bottleneck is the video memory and I suspect that the HBM2 memory is a slight advantage over the GDDR memory, but I would prefer more memory over faster.

I would generally recommend to max out the CPU in relations to your budget, because this keeps the value of the machine high for longer and provides head room for future uses. Also, if you’re using video sources that need to be decompressed in software such as NDI or mimoCall, a faster processor will help.

We’ve rarely seen mimoLive use more than about 4GB of memory, but since you can’t easily upgrade the memory later on, go for the biggest option.

One thing to keep in mind is that you need to watch for thermal throttling which we have seen in some MacBook Pro configurations more severely than others. Make sure that you’re keeping an eye on environmental temperatures and direct sun light. Use the Intel Power Gadget to monitor this “feature” of the MacBook Pro.

I hope this helps!

Best regards,


yeah, that’s unfortunately the worst thing about getting a new MBP, none of those are up-gradable in the future, including the SSD. Now, with the advent of Thunderbolt 3, this is someone mitigated as you can add on really fast drives and eGPUs for more storage and graphics power, but I don’t know what ML supports on the eGPU boxes. @“Oliver (Boinx)” could you let us know what the eGPU support is for ML, if any? Thanks.

@kmac1036 mimoLive is currently using Quartz Composer for rendering the graphics. This is currently not supported by Apple on eGPUs. It might be supported in a future OS update. Also, long term we will move to technology based on Metal which is supported by eGPUs.

That said, the benefit of using an eGPU for mimoLive compared to the internal GPU options wouldn’t be that great currently anyways.

ah ok, thanks for the info.

Speaking of the Thunderbolt 3 ports, may we assume this will increase the number cameras we can input? I know one port must be used for power. Is there an adapter that provides power and a pass-through port that will permit us to use 4 cameras at a time?

@alternativeplan which cameras and how are you hooking them up (converter boxes, etc)? There are a couple of TB3 docs that have enough power to run the MBP and allow passthrough. Elgato just announced a new one at CES, I think OWC/macsales.com has one & I think Belkin, if I remember correctly.

@alternativeplan We’ve been using 8 HD SDI cameras with the Blackmagic Design Quad card housed in a Sonnet PCIe Thunderbolt 3 enclosure.

Thank you for your responses. @kmac1036 , Thank you for your suggestion for TB3 docs. We will research. Currently we’re using blackmagic HDMI to SDI converters, to go in via thunderbolt 2. As there are only 2 thunderbolt ports, we’ve been limited to two cameras. We have used an iPhone as a third cam thru USB, but only as a lockoff on the scoreboard. I suppose we’ll have to add a thunderbolt 2 to TB3 adapter to the string. Here’s an image of the flow: http://altplaninc.com/images/stream_workflow.png

@alternativeplan The TB2 to TB3 adaptors can do what you want, but the MBP doesn’t have enough power on the TB3 ports to power more than 3 adaptors.

@alternativeplan yeah, you might be better served with getting what @“Oliver (Boinx)” mentioned - native TB3 expansion chassis & multi channel capture card.

I couple things I noticed over the weekend doing work on my setup:

Oliver is correct, the VRAM is really important. I have 2 HDMI captures & an iPhone X plugged via USB, with a 2 up preview (had no more than 2 sources live) 2 mics (via Scarlett 2i2) and capturing audio output via HDMI (laptop), it was idling at 950MB VRAM. I also noticed via Activity Monitor that ML seems to only use 1 thread per core (early 2011 17" MBP, 2.2Ghz i7). My capture and output is 720p30 fps. I tried going to 60 fps and it couldn’t keep up, it was dropping frames.

Hi I’m wondering how this will affect the situation with ML? I have bought a Blackmagic eGPU Pro - thinking this would offload the GPU requirement of a multicamera live show - but as far as I can tell from this thread and the ML memory usage display - ML is using the Macbook Pro internal GPU and not touching the eGPU.

Should I send the eGPU back? Or is there a software update in the pipeline that would drop Quartz Composer in favour of Metal that would make actual use of the eGPU?

Quartz Composer is now officially depreciated. Although it is supported in macOS Catalina, Apple does not guarantee it will be in versions of macOS after Catalina. Some powerful Final Cut plugins depend on quartz compositions generated using Quartz Composer.

I’ve managed to get the eGPU to display in ML - the important thing to do is launch ML on the external monitor being rendered by the eGPU - and have no windows of ML on the Macbook Pro - now I get the full 8GB o fmemory displayed - interestingly as you can see from the screenshot - the free memory is tiny - and that is puzzling as when using the internal GPU it was a fraction of the 4GB being used?

1 Like

Oh yeah! so glad you tested this! Progress on eGPU is a great thing!

Like @Oliver_Boinx indicated that while ML can use the eGPU (I have it working using about 2.5GB of the 7.98 GB available GPU memory) - I still get quite a lot of CPU usage required for ML. This makes the Macbook fan create quite a lot of noise - though the Macbook copes well with 4 camera inputs and quite a lot of compositing.

My main concern right now is what the roadmap is regarding to Quartz Composer being deprecated

Interestingly I seem to use LESS cpu with a quieter fan and have the a=same level of frame performance when not using the Blackmagic eGPU - and simply using the internal MacBookPro GPUs. The AMD 4GB Radeon Pro 560x is fairly maxed out as you can see - beut everything seems to work a bit better.

With the eGPU plugged in - and all the cameras going in direct to the Macbook Pro - but with the HDMI display using the eGPU - I get more or less the same CPU use as not using the eGPU at all - but the GPU work is more nicely spread across GPU’s - possible benefits?

The hard work is getting all the data from the interface (decoding it) and getting it onto the graphics card. This is reflected in the CPU usage. You may get better results by using different input devices. For example, the Blackmagic SDI input devices use a lot less CPU than any USB3 capture device.

Deprecated means that there will be no further improvements, but it may well be available for some years to come. QuickTime was deprecated for more than 5 years before Apple finally removed it.

On the other hand, we’re already working on a drop-in replacement for QC based on Metal which should be ready in time. For macOS Catalina, mimoLive continues to work.

1 Like

Thanks for this - I’ve suspected that for a while. The issue for me is looking at price for a multi-camera set up - we need to set up many small “home” studios - so the ability to use cheap and ideally compact cameras looks essential. That points to either usb webcams or cheap cameras with HDMI out.

I’d like to know some detail regarding what settings to use in terms of camera output, frame rate and compression. I noted before that using MPEG compression on camera input rather than the RGB raw input (being vague here) resulted in lower CPU usage.