lowering timeline res doesn't help playback fps?

Took another user's advice to lower the resolution of a busy timeline I'm working on, so changed from 4k to 1080p. But, I still get identical playback performance--ie fluctuating around 16-21 fps.

I also made sure the timeline quality was the same in the test and again got identical results.

Should it have done better?

My preferred methods for improving performance are:

1. Better hardware.
2. Better source material. (i.e. - NOT highly compressed delivery formats like H.264/5)
3. Proxies
4. Cache (or Optimized Media)
5. Render in Place (RiP)

Never been a fan of changing timeline resolution. I always keep that at the best I will need for delivery.

Should it have done better?

Yes. Look for a configuration error somewhere. Identical performance with greatly reduced playback demands doesn't add up.

As for the general practice, it makes far more sense to reduce timeline resolution for editing purposes, than creating proxies, buying new hardware, etc., etc. etc. Resolution independence is a prime feature of Resolve -- precisely so you *don't* have to engage in cumbersome workarounds.

If you only have video footage than lowering timeline resolution would help playback fps. If you have all sorts of other things happening, audio effects, transitions, fusion compositions, resolve fx etc, than it might not. It depends on what is actually being effected. Possible caching issues as well in case of cached effects and transitions etc. So make sure you are using up to date cache if you used caching mechanisms.

To test the concept of lowering timeline resolution you can always start a new project, and new timeline and test it on footage alone. One that won't play in real time of course.

Yes. Look for a configuration error somewhere. Identical performance with greatly reduced playback demands doesn't add up.

As for the general practice, it makes far more sense to reduce timeline resolution for editing purposes, than creating proxies, buying new hardware, etc., etc. etc. Resolution independence is a prime feature of Resolve -- precisely so you *don't* have to engage in cumbersome workarounds.

Thank you for this and glad to know I should be getting better results. I'm unfamiliar with "configuration errors" tho. Do you mean things like timeline quality (ie Full, Half, or Quarter)? I checked that a few times and the performance is the same on both 4k and 1080p timelines. By that I mean, using ProRes Proxy as the proxy files, I get around 19-21fps on both timelines regardless of whether playback is set to Full, Half, or Quarter--with the one exception of the 4k timeline at Full only plays back at 13fps.

Thoughts?

Timeline playback resolution is not lowering format of the timeline or rendering files to disk. It is lowering the image in the viewer, same as taking a screenshot every frame and resizing the image. Its meant to provide quick boost to playback speed at the expense of quality. Its good for doing type of work that is important to get timing correct, so you want real time playback, but not when quality is important. Timeline playback resolution is very differnt, in fact completely differnt to proxy media, optimized media or caching systems. All which can boost performance.

There is of course concept of resolution independence, meaning that you can lower working resolution in a timeline and proportionally scale many things, and when needed to back to original resolution of source media. This is also differnt than any of the mentioned concepts before. Its best to use this process when editing footage, since you can get boost in performance quickly and not lose anything. There are few caveats if different aspect ratios are used etc. but generally its a very underutilized method to boost playback performance.

I don't know which one of these you originally meant, but I'm pretty sure all of them can be used to boost playback performance. If you are not seeing any difference at all, I strongly suspect something is not being done correctly on your end.

Since you haven't mentioned what version of resolve, studio or free, version number, operating system of hardware you are using its hard to say anything else. But these methods all should if done correctly boost performance.

It would help to know what exactly is on your timeline and at what resolution you're monitoring the footage. The fact that you only get 21fps with Prores proxy suggests either your system is overwhelmed by what's on the timeline (you haven't said what's there) or there's some kind of configuration error (i..e, wrong settings). If you also reduced timeline playback quality to half or quarter and *still* see no improvement, something's gotta be wrong.

If you have 4K Prores proxy footage on the timeline with no fx applied, does it plays in real time with the timeline set to either 4K or HD? And what is your system?

Thank you to the past two responders and sorry for the confusion. To clarify, I'm talking about 2 different things:
1- taking a 4k timeline then changing its settings to be a 1080 timeline. Its output is also set to 1080, to the 2nd monitor I have it play on via Cleanview is 2k.
2- adjusting the playback resolution from Full, to Half, to Quarter.

I'm running Resolve Studio (ie the paid version) 19.1.3 on a M1 Pro MBP on Mac OS Sonoma. I know this is underpowered and am in the process of buying a Mac Studio in either the M2 Ultra, M3 Ultra, or M4 Max variety.

The footage is stored on and writes files to an OWC 4M2 Thunderbolt 3 NVMe Raid 0 device. It has 16 terabytes and runs pretty fast, so I wouldn't think that's the issue.

The footage is a mixed bag, but mostly 4k60. The timeline is 23.98 and the 60p is for better slomo. I don't think it matters for this test since I'm using the proxy files, but I have up to 8 GoPros, Pocket 4ks, Pocket 6ks, and some Sony and other makes.

It sounds like y'all are suggesting there's a bottleneck somewhere, so I'm guessing it's the M1 Pro MBP in some way--be it its GPU, CPU, some other hardware part, or a combo of parts.

Does this help give a clearer picture?

Does this help give a clearer picture?

It does. One thing that seems potentially bottleneck. Are you using processor heavy re timing filter, like optical flow and or speed warp? That could explain the demands on the processing.

The footage is stored on and writes files to an OWC 4M2 Thunderbolt 3 NVMe Raid 0 device. It has 16 terabytes and runs pretty fast, so I wouldn't think that's the issue.

Does this help give a clearer picture?

That would be my guess on the bottleneck. The thunderbolt. If you're trying to read multiple files at once it probably chokes. If you also have swap set to that external drive it will choke.

Try and copy a few test files on the internal drive to replicate your timeline situation and see if anything changes.

It does. One thing that seems potentially bottleneck. Are you using processor heavy re timing filter, like optical flow and or speed warp? That could explain the demands on the processing.

Thank you for that, but no, I'm not using any optical flow or speed wrap. Right now, the only filters are a little color correction. Many don't have any correction, but a few have one or two nodes (including the odd lut) from the Color page.

There is a speed change from 60p clips playing on a 24p timeline though. Could that be the culprit?

There is a speed change from 60p clips playing on a 24p timeline though. Could that be the culprit?

Possibly. Easy way to test. Make a 60fps timeline and put 60fps clips on the timeline with everything else being the same. See if its the same.

That would be my guess on the bottleneck. The thunderbolt. If you're trying to read multiple files at once it probably chokes. If you also have swap set to that external drive it will choke.

Try and copy a few test files on the internal drive to replicate your timeline situation and see if anything changes.

Really? Im surprised you would think so. I would do your test, but the files are too big to fit on my boot drive. However,below are the speed tests I just ran on the TH3 drive (OWC 4M2) that I'm currently using if you'd care to take a look:

4gbtest (read only) - results in MB/s
SEQ1M QD8 = 2834
SEQ1M QD1 = 2267
RND4K QD64 = 464
RND4D QD1 = 44

64gb test (read only) - results in MB/s
SEQ1M QD8 = 2655
SEQ1M QD1 = 2280
RND4K QD64 = 472
RND4D QD1 = 43

I'll also add that I have a colorist running an M2 Ultra Mac Studio who said he got perfect playback using the same make and model NVMe in a TB4 external drive. That enclosure's speed tests are a little better, but not much as far as I can see, to the OWC 4M2,I believe he also uses some kind of raid that has generally lower scores to mine.

Thoughts? Still think this is the bottleneck? Ideas of what else could be?

That would be my guess on the bottleneck. The thunderbolt. If you're trying to read multiple files at once it probably chokes. If you also have swap set to that external drive it will choke.

Try and copy a few test files on the internal drive to replicate your timeline situation and see if anything changes.

Really? Im surprised you would think so. I would do your test, but the files are too big to fit on my boot drive. However,below are the speed tests I just ran on the TH3 drive (OWC 4M2) that I'm currently using if you'd care to take a look:

4gbtest (read only) - results in MB/s
SEQ1M QD8 = 2834
SEQ1M QD1 = 2267
RND4K QD64 = 464
RND4D QD1 = 44

64gb test (read only) - results in MB/s
SEQ1M QD8 = 2655
SEQ1M QD1 = 2280
RND4K QD64 = 472
RND4D QD1 = 43

I'll also add that I have a colorist running an M2 Ultra Mac Studio who said he got perfect playback using the same make and model NVMe in a TB4 external drive. That enclosure's speed tests are a little better, but not much as far as I can see, to the OWC 4M2,I believe he also uses some kind of raid that has generally lower scores to mine.

Thoughts? Still think this is the bottleneck? Ideas of what else could be?

Yeah I still think its a bottleneck. There have been tests done on Macs where the swap has been set on external nvme drives and even the OS was run from external drives. You would think since its the same speed-ish or faster as some internal drives it would work the same. It does not, TB and USB have overheads, negotiation etc.

Here's the video, with the OWC TB4 external drive, and while its not the same as your use case, it signals that doing intensive bandwidth reads and writes to external drives (which swapping is) is not ideal.

Also TB4 is double the bandwidth which means less overhead.

While it might work with 1-2 video streams at once, doing like 10 reads like in your case it probably doesn't work that well...But I don't have a Mac to test out. However, it should behave somewhat the same as USB drives and I do have plenty of those, and they do start choking when you do multiple reads and writes at once and not one by one.

You could just move some files to the internal drive and see if it makes any difference. (cut some small parts to mimic your timeline).

Yeah I still think its a bottleneck. There have been tests done on Macs where the swap has been set on external nvme drives and even the OS was run from external drives. You would think since its the same speed-ish or faster as some internal drives it would work the same. It does not, TB and USB have overheads, negotiation etc.

Here's the video, with the OWC TB4 external drive, and while its not the same as your use case, it signals that doing intensive bandwidth reads and writes to external drives (which swapping is) is not ideal.

Also TB4 is double the bandwidth which means less overhead.

While it might work with 1-2 video streams at once, doing like 10 reads like in your case it probably doesn't work that well...But I don't have a Mac to test out. However, it should behave somewhat the same as USB drives and I do have plenty of those, and they do start choking when you do multiple reads and writes at once and not one by one.

You could just move some files to the internal drive and see if it makes any difference. (cut some small parts to mimic your timeline).

Thanks for this, but I think we can eliminate the SSDs as the bottleneck now as I tested the same drive on the M2 Ultra Mac Studio I'm trying out and there were no dropped frames. Played back perfectly even at full resolution on a 4k timeline, whereas the same project and drive on the M1 Pro MBP only got 13fps.

So, thoughts on where else the bottleneck might be or what else might be happening?

There is a speed change from 60p clips playing on a 24p timeline though. Could that be the culprit?

Possibly. Easy way to test. Make a 60fps timeline and put 60fps clips on the timeline with everything else being the same. See if its the same.

Tried what you said and got the following results
4k60 timeline
1/4 50fps
1/2 40fps
full 13fps

1080p60
1/4 50fps
1/2 50fps
full 40fps

So, does this tell you anything? I see some modest gains here, so I'd think the frame rate change on the fly taxes the performance a bit. I'd also think it suggests the external storage isn't the bottleneck, but I believe I've proven that it isn't as the same storage plays back a full on a 4k timeline w no dropped frames.

Hmm. what seems very strange to me is that you are not get 60 fps on 1080p at a quarter of resolution. That is very strange to me. It should run at full speed if you played it on a 20 year old calculator. So something else might happening. Setting or some artificial limitation. You are getting improved playback as expected, but I see no obvious reason from information provided by are you not getting full 60 fps playback that source media contains. 1080p 60fps should play at 1080p60 fps . And if you lower resolution by factor of four, your wrist watch should play it smoothly. So what is happening there I'm not sure.