Copy link to clipboard
Copied
Hi,
I have a projects containing over 2000 clips (basically it's the result of a 2 hour video chopped up into 2000 cuts). The playhead gets laggy when editing and deleting clips on the timeline.
I noticed it doesn't happen when I have let's say 1000 clips from a chopped up video.
So I think it might be a hardware bottleneck somewhere.
The videos are 1080p 30fps
I have:
6 cores /12 threads intel cpu
32gb of ram
GTX 970 gpu
The source files are read from a HDD. I tried reading them from an SSD to see if that improved the playback but it doesn't.
Premiere is up to date.
CUDA accelaration is enabled on Premiere.
What do you think is the most likely cause of this performance issue with projects with many cuts?
CPU, RAM or GPU?
I can up my RAM to 64 GB which is the max my coard accepts.
Thanks.
Copy link to clipboard
Copied
Your machine is not very powerfull at least a quite old gpu.
I would shop up the timeline is several sequences or projects.
Copy link to clipboard
Copied
Copy link to clipboard
Copied
What do you think?
I could double my Ram cheaply but getting a modern CPU and GPU would be a
significantly bigger investment just to fix this particular problem.
Thank you
Copy link to clipboard
Copied
A bit of background knowledge, that is very useful in figuring out a lot of how Premiere projets behave.
That sequence? That exists only as metadata ... a text file, essentially. That lists the time code of things that happen on it, what the bits of media are, any effects used, all the details of what it takes to gather the media, apply effects, and then play that 'sequence'.
So it's not replaying video, it's rebuilding the sequence from that text file every time you play it back. Having to grab all those bits ... you see?
Next, the CPU is then told to grab bits from various video and/or audio files from whereever they are on the computer, and add just those bits, and at the right time.
Plus the effects are recomputed and added to the image during playback.
Yea, it takes a solid machine to handle this!
But note at the top, those text files that everything following is based on? Premiere needs to do a ton of RAM and cache work. So if you're low on RAM, and these days, 32GB is minimalist in real terms, Premiere is going to have issues with long complex-cut sequences.
You can cover part of that if you have a lot of very fast cache drive space available. As in like I do, a second Nvme SSD drive on the motherboard, that is totally and only video post cache files.
If your cache drive is the same as the OS/Programs, that's ... problematic. If there's anything else Premiere uses, such as the original media or project files, on that same drive as the cache files, that's a potential blockage.
And as always, the sustained speed of drives and their connections in reality, not manufacturer specs, is incredibly important.
So that's some things to check.
I'm pretty much with Ann though ... that GPU is simply not usable anymore, Nvidia abandoned those so Premiere can't really use it for much. You don't have many cores on the CPU either. Yea, you need to run fairly simple sequences.
Do this in sections. Then add them all together simply to export. Or ... do an export to say ProRes422 of each sub-sequence, reimport them, join them together, and make your final export.
Copy link to clipboard
Copied
Dont use chatgpt its unreliable.
Copy link to clipboard
Copied
Now that is a fantastic answer, Neil!
Thank you so much for taking the time to really cover all the possible components that might be affecting my performance.
I'll give you some context on workflow set up.
I use:
- Separate internal Nvme drive for OS/Software including Premiere
- Separate internal SATA SSD for cache/previews
- External HDD for project source files
- Internal SSD for export files
Basically the chopped up sequence with over 2000 cuts from a software that removes the silences and spits out an XML file with all the corresponding cuts to finetune and import into Premiere. This is the file that creates the sequence in Premiere.
From what I understand I have 2 options:
1. I separate the imported sequence into 4-5 chunks, edit them in separate sequences and then create a master sequence where I join the 4-5 edited sequences, and export the master sequence.
2. Spend at least $2500 in a new board, CPU, GPU and RAM. (I know I could spend less but If I go the upgrade hardware route I would probably have to future proof the investment)
The kind of project I'm discussing here will be my main workflow for a while, but I think for the moment It's not worth it for me to spend all that money to overcome this performance issue. I could try upgrading the RAM to 64gb which is my board/CPU max allowed. Do you think that will help?
Thank you
Copy link to clipboard
Copied
Hi,
I upgraded the RAM 64GB. The performance issue with the lagging playhead when ripple deleting on long timelines with many cuts persists.
I've been trying to edit splitting the project into chunks but it's really inconvenient because I never know when I should split the files in order for the playhead to play smoothly and I have to be very careful not miss any sentencences between the cuts. I don't think I can do this without going crazy. The thing is all my projects are like this. long 70-130 minute videos chopped up into 1500-2500 clips (result of all the silences removed).
I'm seriously considering upgrading the hardware, I would have to do it sooner or later anyways.
I need a CPU/board that allows for multiple capture cards so PCIE lanes are very important for me so I was thinking about getting Threadripper Pro 3955WX (allows up to 120 pcie lanes) along with ASUS Pro WS WRX80E-SAGE SE WiFi board (has 7 full x16 pcie slots).
I will keep the 64 GB of RAM, the GTX 970 GPU and the storage set up as it is for now since it seems the CPU is the most important component. Let's see how it goes.
The Threadripper Pro 3955WX has 16 cores/32 threads running at 3.9-4.3 Ghz.
Do you think this CPU will fix the lagging issue when working with timelines with +2000 cuts from the same source video?
I appreciate any suggestions or guidelines.
Thank you.