M2 Mac Mini is failing to play back mutliple clips of 6K ProRes unless I select Prefer Performance mode in FCPX

darkhorse69 wrote:

Yeah they are the tests I did that prompted me to post this question. 24p 5888x3312 ProRes RAW HQ. Data rate is about 180MB/s. The SSD speed is 1.5GB/s.

The effective bandwidth of the internal SSD for the base model is likely 60 to 80 percent of the 1.5 GB/s (in real-world situations). I estimate the target data rate to be more like 300 to 318 MB/s for the 6K ProRes RAW HQ clips. This would suggest 3 streams might be okay but at 4 you would see some significant slowdowns with playback set to Higher Quality (which is what your reporting below confirms).

o effects. 1 clip would play fine, two clips would not play without frame drops (warn on frame drop checked inside FCPX). RAM is less than half full and green for two clips. GPU history graph is totally maxed out.

Three clips would occasionally put the RAM in yellow and would stutter. Four or more clips put the RAM in red and also stutter.

I have a feeling even though RAM is green there may still be swap going on, and the SSD is slowing down while reading the data while also trying to be used as swap. However with the GPU history graph fully maxed out that indicates that even with more RAM there would still be a bottleneck.

My thinking is an M2 Pro or even higher is required and also 16GB RAM, perhaps also the 512GB SSD for swap storage speed.

I have returned the Mac Mini. I am going to see how Premiere Pro goes on my PC.

Probably best to have returned the model you had.

With your additional reporting, it really seems that your bottleneck was RAM. The unified RAM is used to drive any connected displays, act as swap/virtual memory, along with servicing the CPU, GPU, neural engine, and other media engines. That's a lot of the system HW using 8 GB of RAM.

I would expect you'd be happy with an M2 Pro Mac mini with 32 GB RAM and preferably 2 TB of SSD storage. Beside the much faster SSD in the larger storage configurationns, the M2 Pro models have significantly faster memory, as well. Optionally, you could go for the upgraded CPU/GPU option. If more storage is needed, a fast RAID of SSDs connected via Thunderbolt would likely work well. It's probably a good idea to get the most you can afford, as little of it, if any, is upgradeable. It very well may be that a Mac Studio or the forthcoming Apple silicon Mac Pro would be a better choice(?).

---

Besides the HW issue, there is consideration of the best, or most efficient use, of FCP (or any other NLE) when editing. When doing most editing, there is no real need to use super high-bandwidth media. That's why there are proxies. When it comes time to add effects and do color work, then it would be preferable to use higher-quality media (when super responsive playback and UI interaction isn't so much of an issue). There are lots of switches to flip to get FCP performing better when needed (turning off background rendering is one thing). Judicious choices when preparing media and how one uses FCP can make a huge difference in the user's experience. Now, if you have a huge budget and want to edit huge frame sizes in high bandwidth codecs, you can do it up to a point using a Mac (not with a mini, probably more so on the PC side where you can custom build massive setups with many GPUs).

The Mac mini M2 with 8GB RAM and 256 GB SSD is the low-end model and its SSD (along with the 512 GB model) is much slower than the 1 TB or 2 TB models. It only has one SSD chip. The others have two and are in a RAID-0 configuration, which makes those SSDs quite a bit faster than what you have.

If you are trying to play back the media on the internal SSD, you don't have much room to work with, either, especially with 6K ProRes RAW HQ footage. ProRes RAW HQ footage at 6K has a target data rate similar to ProRes 4444 at 6K (see https://www.apple.com/final-cut-pro/docs/Apple_ProRes_RAW.pdf and the Target Data Rates table at the end of https://www.apple.com/final-cut-pro/docs/Apple_ProRes.pdf).

Like most marketing hype, what you quoted probably best applies to one of the higher-end configurations of the Mac mini M2. And to be clear, it is likely you've hit a performance bottleneck, unless you have some other clearcut issue. Also, there is an issue with macOS Ventura 13.2 that is causing slowdowns and other potential issues. Activity Monitor has had some issues in recent macOS releases, so it may not be indicating things as well as it should. So, besides any obvious issue with SSD performance and your footage, it may be a bit difficult to pin down other possible causes for your bottleneck (besides the GPU)...

Please play each of those clips in Quicktime Player, do CMD+I to "Get Info", then spin down the disclosure triangle show "video details". Take a screen cap of all that -- inc'l general info -- and post it here. We need to verify the resolution, frame rate and data rate.

In FCP, create a new test project using "automatic settings", then add those two clips to the timeline. Do not add any effects except to rescale them to 50% and reposition so you can see them both in the viewer. Will that play back smoothly at 1x normal speed in quality mode? We need to ensure it's a playback issue vs an effects issue.

Just to be clear for future people reading this, your title said the M2 Mini could not play smoothly multiple clips of 6K ProRes unless the viewer is in "better performance". In fact your clips were 6k ProRes RAW HQ. That was not mentioned in the Apple statement you cited, nor did Apple reference a given playback mode.

"Regular" ProRes is a different codec with different decoding requirements than ProRes RAW HQ. The computational burden of decoding and playing back regular ProRes is quite low. It is likely an entry-level M2 Mac Mini could play back two streams of 8k/24 ProRes 422.

By contrast the computational burden of decoding and playing back ProRes RAW is higher. In general it is quite efficient relative to certain difficult variants such as 4k/24 Sony XAVC-S, but it's less efficient than regular ProRes. ProRes RAW is true Bayer-format RAW and requires demosaicing by the playback computer. ProRes RAW HQ has an even higher data rate but is not needed for most applications.

6k material has about 2x the pixels per frame as 4k, so in general the compute and I/O burden of playback is about 2x higher than 4k, assuming both use the same codec and bitrate. FCP 10.6.5 on my old 2017 iMac 27 can play three streams of 4k/24 ProRes RAW without any dropped frames with the viewer in high quality mode. In better performance mode it can play at least 10 streams of 4k/24 ProRes RAW without any dropped frames.

My contrast my M2 Pro Mac Mini can play nine streams of 4k/24 ProRes RAW without any dropped frames in high quality mode. In better performance mode it can play so many streams I didn't have time to test that fully.

I think most serious professionals who edit multiple streams of 6k ProRes RAW HQ would not expect an entry-level M2 machine to handle that perfectly unless the viewer was set to "better performance". Most people in that position would get at least an M2 Pro machine.

The 4070Ti GPU costs more than the entire base M2 machine and it consumes up to 285 watts. By contrast the entire M2 Mac Mini machine consumes a peak of 50 watts.

You previously asked how could you know what Mac machine configuration could handle what type of workflow without trying them out. We collectively have lots of machines and some of us would be willing to run tests for you.

I have an M1 Ultra Mac studio, M1 Max MacBook Pro 16, M2 Pro Mac Mini, 2017 iMac 27, and 2019 MacBook Pro 16. I regularly edit multiple streams of 4k ProRes RAW, 6k BRAW and 8k REDRAW using FCP. If you have a specific FCP workflow you want tested, let me know.

8 GB of RAM, since it's "shared memory" is a bit low, too. RAM in the "green" is memory pressure, which is not quite the same as usage. Two 6K ProRes RAW HQ streams are a lot of pixels to push around. Since Activity Monitor isn't currently as useful as it was, direct comparison with a more feature-laden M2 mini might be the best way to test things. Evaluating performance of Apple silicon Macs is not easy, due to the deep integration of the various components...

(Sorry for the extra, slightly redundant info. Ran out of time to edit things in previous post.)

Please post an EtreCheck report https://etrecheck.com giving the application Full Disk Access in Privacy System Preferences. Copy the report and use the Add Text button to paste the result into reply box. 

Yeah, I hear you. You are right about an M1 Max ( or Ultra). M2 Macs are marginally faster overall than similar M1 models (same thing happens with every new generation of HW).

Any RAW codec is intended to provide a raw format from which you can "develop" your source/camera media with more options than with a more constrained codec.

If the magnetic timeline and the power of using metadata isn't compelling enough reason to want to use FCP, then other options may be better (and sometimes cheaper on other HW). Those two reasons are enough for me for most projects, as I end up saving at least 50% of the overall project editing time when I use FCP versus AVID Media Composer or DaVinci Resolve Studio. Coming from a time when a simple Gaussian blur or sharpening on a photo image in Photoshop used to take 20 or 30 minutes, I have no problems with more recent systems. But, yes, it's nice to have a responsive computer. To each, his/her own... ;-)

Since the silicon Macs appeared on the scene, there have been issues with Activity Monitor. It doesn't show everything that is going on. The memory pressure (green, yellow, red) values you report likely don't show the other "unified" uses of RAM (so the RAM may, in fact, be effectively fully utilized before seeing yellow or red in the memory pressure graph). I'm running Monterey on a 2019 Mac Pro and Activity monitor doesn't even show the GPU activity/history (but iStat Menus does).

I'm not sure if you implied it, or not, but HEVC isn't a good archive format. It's a super highly-compressed lossy format only really suitable for "efficient" delivery of content. Marketers routinely lie via omission. "And both chips..." says nothing about the actual configuration but I'd bet a dollar that the benchmarks and performance stats that are used come from the highest-end configuration (unless specifically broken down by configuration, which is rarely done). Never rely on marketing. Third-party websites that do testing can be helpful when trying to learn about real-world performance...

Do you have an Apple Store near you? You can test drive the machines (call first/make an appointment to make sure they have them in stock) to see which performs the best.

Also, please follow Joema's instructions for a test and the request for more info to help rule out other issues.

I'm glad that you found a working solution, but to be fair to the M2 mac, you are not exactly comparing "Apples to Apples", because you are not working with the ProRes RAW in your pc. Instead you are using a much lower bandwidth format.

I bet that if you had done a similar conversion and worked with much less demanding media on the M2, it (even with the meager 8GB of RAM) would have cut through it just as well.