Best way to not need 4k proxy files? Faster CPU, faster GPU, or more RAM?

[GoodEnough]

What is the bottleneck for needing proxy files for 4k video?
Slow CPU, slow GPU, or not enough RAM?

ie: Best way to not need 4k proxy files? Faster CPU, faster GPU, or more RAM?

[Selur]

I'm no Vegas user, but in general:
I doubt the gpu is important, but the vpu might be. If your vpu can't decode the video, your cpu will be used, which for 4k content can be quite demanding.
If decoding through the cpu is used, you might hit ram limits if you have less than 8gb of free ram.
Additionally, creating of proxy files might be limited by:

• file access speeds, normally this only happens when the files are accessed through a network connection.
• when some other tool (like a virus scanner or similar is also accessing the file)

-> even Vegas users probably can only give some helpful advice if you share more about your setup and the formats you are dealing with.

Cu Selur

[GoodEnough]

Local HDD files
No network
16GB RAM

[Selur]

And what format is your input?

[GoodEnough]

And what format is your input?

GoPro 8
.mp4

[smrpix]

And what format is your input?

GoPro 8
.mp4

GoPro and DJI compress their files in ways that are notoriously hard on system resources.

This is practical advice, not technical advice: Just make the damn proxies. I-frame, Lo Rez. Your editing experience will be much happier and you'll be finished much faster.

[GoodEnough]

What is I-frame?
I will do proxy, but am still curious what hardware you need to throw at the problem to solve it by force, CPU, GPU, or RAM?

[buzz1891]

1) 'I-frame' is likely a reference for 'Intra-frame codec' contrary to Inter-frame ones: https://www.geeksforgeeks.org/difference-between-inter-and-intra-frame-compression/
2) The typical bottleneck in case of editing AVC/HEVC 4K sources is decoding speed. The processing can be done by CPU or GPU (only in modern sw thou). But latter approach has its drawbacks: only some codecs/color formats/features are supported. As consequence of limited compatibility and driver updates it's prone to bugs, this happens on regular basis.
3)The good news is that you don't need latest&greatest GPU for decoding, since the job is done by a separate chip NVDec, which is the same for rtx3050 and rtx4090. so you only have to pay attention to the generation of NVDec. And typically to see major improvement you have to jump over one, like from 1st to 3rd, etc.
4) Different workflow may need different performance level. Being able to play one or even two 4K30 streams at normal speed does not mean you can do comfortable forward/backward rewind at 3x speed. Or play four 4K60 streams in multicam editing.