Could the 16" MBP with 8GB GPU compete with iMac 8-core for video editing?

[johngwheeler]

An earlier test by the "Max Tech" YouTube channel compared the previous 8-core MBP to an identically priced 8-core iMac for video editing tasks, and found that in most cases the iMac was significantly faster. The exception was transcoding tasks for H.265/HEVC and H.264 that used the on-board T2 chip on the MBP.

Much of the difference was attributed to better thermal management on the iMac and better GPU options.

I'm wondering if the improvements in these areas on the new 16" MBP would make the MBP performance closer to the current iMac than previously?

I like the idea of have a portable editing solution, but don't want sacrifice too much performance compared to a desktop.

No doubt, someone will be testing this over the next few weeks, but I'm interested in hearing what improvements you would expect to see.

Thanks!

 

[Zwhaler]

Of course the new 16" MacBook Pro is closer to the iMac than previously, but by how much is a different question. For editing, I think the two will perform pretty similarly for 1080p and light 4k content, but heavy 4k editing the iMac will be noticably faster due to the improved GPU, particularly the Vega 48. The thermals and sustained performance differences will be the most pronounced when rendering, during which time the iMac will certainly outperform the MacBook Pro.

 

[danwells]

There's no way a CPU set to draw 45W will meet the sustained performance of one that gets 95W. All else being equal (and these are very similar CPUs, so all else is pretty close to equal), the iMac will run closer to max clock more of the time. Note that, while the two chips have the same maximum turbo, the base clock speed on the iMac is 50% higher (3.6 gHz vs 2.4 gHz).. My best guess is that the iMac will average 25% or so faster in pure sustained CPU?

Assuming the iMac doesn't have a Vega 48, the two GPUs should be pretty similar. I couldn't find them benchmarked directly against each other - but they both benchmark into exactly the same group of competitors. If the iMac DOES have a Vega 48, that's about 150% the speed of the Navi 5500M...

RAM and general architecture should be quite similar - assuming the two machines have the same amount of RAM (or else both have more than enough for the task at hand). One big wild card is the storage. Many, even most iMacs ship with Fusion Drives (pure SSD storage is available, but I wonder what percentage of iMac owners spring for the upgrade). The MacBook Pro always has a superfast SSD.

If the iMac has the i9, enough memory, the Vega 48 and a SSD, iMac by as much as 50%? At least 25%... Remember this is ONLY when it's actively rendering or exporting - both machines will be fast enough that you'll be placing edits in real time unless your video is in some absurd format that uses a huge amount of CPU or GPU. Also remember that certain transcodes are speeded up by the T2 (which the iMac lacks unless it's an iMac Pro)

If it's missing only the Vega, it matters a lot how GPU-bound the task is. If it's CPU performance, the answer is more or less as above - if it's pure GPU, they should be very close.

If the poor iMac has a Fusion Drive (or 8 GB of RAM), the MacBook Pro is almost certainly faster unless the iMac has the OS, editor and footage on a fast (USB 3.1 or Thunderbolt 3, NVMe) external SSD - video files are huge, and will probably wind up on the hard drive portion of the Fusion Drive.

 

[johngwheeler]

There's no way a CPU set to draw 45W will meet the sustained performance of one that gets 95W. All else being equal (and these are very similar CPUs, so all else is pretty close to equal), the iMac will run closer to max clock more of the time. Note that, while the two chips have the same maximum turbo, the base clock speed on the iMac is 50% higher (3.6 gHz vs 2.4 gHz).. My best guess is that the iMac will average 25% or so faster in pure sustained CPU?

Assuming the iMac doesn't have a Vega 48, the two GPUs should be pretty similar. I couldn't find them benchmarked directly against each other - but they both benchmark into exactly the same group of competitors. If the iMac DOES have a Vega 48, that's about 150% the speed of the Navi 5500M...

RAM and general architecture should be quite similar - assuming the two machines have the same amount of RAM (or else both have more than enough for the task at hand). One big wild card is the storage. Many, even most iMacs ship with Fusion Drives (pure SSD storage is available, but I wonder what percentage of iMac owners spring for the upgrade). The MacBook Pro always has a superfast SSD.

If the iMac has the i9, enough memory, the Vega 48 and a SSD, iMac by as much as 50%? At least 25%... Remember this is ONLY when it's actively rendering or exporting - both machines will be fast enough that you'll be placing edits in real time unless your video is in some absurd format that uses a huge amount of CPU or GPU. Also remember that certain transcodes are speeded up by the T2 (which the iMac lacks unless it's an iMac Pro)

If it's missing only the Vega, it matters a lot how GPU-bound the task is. If it's CPU performance, the answer is more or less as above - if it's pure GPU, they should be very close.

If the poor iMac has a Fusion Drive (or 8 GB of RAM), the MacBook Pro is almost certainly faster unless the iMac has the OS, editor and footage on a fast (USB 3.1 or Thunderbolt 3, NVMe) external SSD - video files are huge, and will probably wind up on the hard drive portion of the Fusion Drive.

Many thanks for your very complete reply!

 

[gtg465x]

I doubt it, unless you're working with H.265 or something that's accelerated by the T2 chip. The 580X and Vega 48 in the iMac 5K are especially strong compute GPUs, which is what you want for video editing tasks. The new AMD 5000 series are gaming optimized cards and aren't that great for compute tasks. Just look at these Geekbench results that measure compute...

Geekbench Metal

Pro Vega 48 - 46951 (79% faster than Pro 5500M)
Pro 580X - 36611 (40% faster than Pro 5500M)
Pro 5500M - 26221
Pro 5300M - 22635

Geekbench OpenCL

Pro Vega 48 - 47539 (71% faster than Pro 5500M)
Pro 580X - 39204 (41% faster than Pro 5500M)
Pro 5500M - 27830
Pro 5300M - 23203

However, in terms of gaming, these new cards are pretty good. The 5300M and 5500M appear to slightly beat out the 580X in gaming benchmarks...

Unigine Heaven (OpenGL, Ultra quality, Extreme tessellation, 1600x900 8xAA) / Extreme Preset

2019 iMac 5K, i9 9900K, Pro Vega 48 - 71.3 fps (33% faster than MBP 16" with 5500M)
2019 MBP 16", i9 9880H, Pro 5500M 4 GB - 53.3 fps
2019 MBP 16", i7 9750H, Pro 5300M 4 GB - 47.7 fps
2019 iMac 5K, i9 9900K, Pro 580X - 45.6 fps (14% slower than MBP 16" with 5500M)

 

[Spectrum]

I doubt it, unless you're working with H.265 or something that's accelerated by the T2 chip. The 580X and Vega 48 in the iMac 5K are especially strong compute GPUs, which is what you want for video editing tasks. The new AMD 5000 series are gaming optimized cards and aren't that great for compute tasks. Just look at these Geekbench results that measure compute...

Geekbench Metal

Pro Vega 48 - 46951 (79% faster than Pro 5500M)
Pro 580X - 36611 (40% faster than Pro 5500M)
Pro 5500M - 26221
Pro 5300M - 22635

Geekbench OpenCL

Pro Vega 48 - 47539 (71% faster than Pro 5500M)
Pro 580X - 39204 (41% faster than Pro 5500M)
Pro 5500M - 27830
Pro 5300M - 23203

However, in terms of gaming, these new cards are pretty good. The 5300M and 5500M appear to slightly beat out the 580X in gaming benchmarks...

Unigine Heaven (OpenGL, Ultra quality, Extreme tessellation, 1600x900 8xAA) / Extreme Preset

2019 iMac 5K, i9 9900K, Pro Vega 48 - 71.3 fps (33% faster than MBP 16" with 5500M)
2019 MBP 16", i9 9880H, Pro 5500M 4 GB - 53.3 fps
2019 MBP 16", i7 9750H, Pro 5300M 4 GB - 47.7 fps
2019 iMac 5K, i9 9900K, Pro 580X - 45.6 fps (14% slower than MBP 16" with 5500M)

Amazing that the new mobile GPU matches the iMac 580X.
I really wish they could have fitted one of these mobile GPUs into the Mac mini...or a small, simple, $200 eGPU containing this mobile GPU and a couple of downstream HDMI or TB3 ports would be great.

The external BlackMagic eGPU is ridiculously over-sized, over-priced and under-specced when the internal eGPU of the new MBPro performs better...

 

[bsbeamer]

If your applications can take advantage of multiple GPU solutions, I'd take a look at eGPU like Sonnet eGFX Breakaway Box 650 OC (GPU-650OC-TB3) or Razer Core X. Sonnet is a closer overall "partner" with Apple and they are frequently updating their GPU compatibility list. Unfortunately, Razer is not updating their list. Likely have the same compatibility, but it's difficult to "guarantee".

Sonnet GPU compatibility:

https://www.sonnettech.com/support/downloads/manuals/Compatibility_Graphics_Cards.pdf

Might be a compromise for a middle of the road solution. Would really suggest against the Blackmagic eGPUs right now. The GPUs themselves cannot be swapped out on those. For the price, seems like a poor "investment" right now. Sonnet works with current GPUs like 5500/5700 series or Vega 64. Radeon VII should also be compatible eventually when macOS supports it.

Not a lot of reports with MBP 16" and eGPU yet. Hopefully in the coming 2-3 weeks when more of the BTO configurations are in hands of users.

 

[Spectrum]

If your applications can take advantage of multiple GPU solutions, I'd take a look at eGPU like Sonnet eGFX Breakaway Box 650 OC (GPU-650OC-TB3) or Razer Core X. Sonnet is a closer overall "partner" with Apple and they are frequently updating their GPU compatibility list. Unfortunately, Razer is not updating their list. Likely have the same compatibility, but it's difficult to "guarantee".

Sonnet GPU compatibility:

https://www.sonnettech.com/support/downloads/manuals/Compatibility_Graphics_Cards.pdf

Might be a compromise for a middle of the road solution. Would really suggest against the Blackmagic eGPUs right now. The GPUs themselves cannot be swapped out on those. For the price, seems like a poor "investment" right now. Sonnet works with current GPUs like 5500/5700 series or Vega 64. Radeon VII should also be compatible eventually when macOS supports it.

Not a lot of reports with MBP 16" and eGPU yet. Hopefully in the coming 2-3 weeks when more of the BTO configurations are in hands of users.

Thanks for the ideas. My point really is, why the need for a big external eGPU, when top performance can be had with a modern mobile GPU like that in the new MBPro?

 

[bsbeamer]

"NEED" is not there when 5500M style options exist, but there is "only" one GPU built into the MBP16. Multiple GPUs accelerate many video post-production workflows, especially those that can tap into Metal acceleration. Everything from scrubbing timelines, to encode and decode (with the right codecs). Adobe has been really tapping into Metal lately, so more Metal processing power just makes everything faster.

The "big external eGPU" is basically a power horse you can connect to when at a desk station (likely with additional external monitors) for a desktop-like environment to accelerate your work. Then not using it when mobile, in the field, etc.

Take a look at compute scores for AMD GPUs and you'll see where the benefits start to come into play. There are better options than this, but decent example for now. (Personally think Geekbench is overrated and results are not real-world transferrable, but another topic for another day.)

Metal Benchmarks - Geekbench

Something like a Vega64 (57941 score) is much "better" than an RX580 (33992 score) or 5500M (26395 score).