New Ryzen 7700x(7900x) worth over 5950x?

Planning of building new system, what is the better pick? Ryzen 7000 series has better single core performance, but will it do better than 5950x? Thanks!

Check opendata.blender.org :slight_smile:
Single core performance is more important for blender, but truly, your best info will come from OpenData

The 7900X has a score of 463, the 5950X has a 419. It’s not a major difference, honestly, but that’s just for rendering. I’d imagine the 7900X is better for non rendering tasks, but it all depends on Blender’s optimization and multi core support

Thanks!

Yeah, for rendering there is 3090 on the way😁.Do you think that 7900x will do noticeably better in simulations, animations, etc?

1 Like

It’s quite hard to say. There’s no benchmark for that kind of thing, you’ll need anecdotal evidence and I have none to offer

1 Like

So you think even if pared with ddr5 ram new Ryzens not necessarily will do a much better?

Sorry, I really can’t speculate further- they are similar enough in rendering scores that there’s no obvious superiority, which means it’s very literally a tossup which one is better for simulations and animation

2 Likes

Unless you plan on pushing all 16 cores at full load (which only something like rendering does), then I would expect the 7700X to be a better option, simply because of its faster single core performance and a lot of things are still either single threaded or only a few threads are pushed to max.

Based on that alone, I would totally expect any 7000 series CPU to be faster then the 5950X in most Blender usage (along with most other apps).

I can even give a direct example of running a geometry node animation in a viewport on my 5900X. I recently replaced my 1070 Ti with a 3080 Ti and ran some comparison tests, and on the viewport playback, the fps was exactly the same, so it was CPU bottlenecked. On looking at the CPU in task manager, while all 12 cores are being used, it was running at a total of 50-60% usage.

Most of the cores where ticking along at 50%, but two cores where peaking at 100%. If the CPU was just generally faster, then I would have got a higher frame rate. Likely that even a 7700X would have been faster, even with 4 less cores.

As to how much faster, well that would require some testing and I don’t have the parts to do that. Even so, I would expect a 7700X to make simulations, animation playback, etc to be faster then a 5950X, while a 7900X would easily be noticeably better.

While not Blender specific, I base this somewhat on reviews like this: https://www.youtube.com/watch?v=s04TOQkzv3c

The two benchmarks to look at there are the Premiere and Photoshop, which gives some idea as to lightly multi-threaded and single threaded performance.

2 Likes

My recent benchmarks with my 5950x on Linux was 466, on windows it was 404.

1 Like

I wouldn’t replace the 5950x for at least another generation unless you have an absurd amount of money burning a hole in your pocket.

Real world performance (blender responsiveness) isn’t really noticeable unless you are doing leaps of 2x 3x faster single core performance. Just a part of the human condition.

For rendering performance or a very rough estimate of multi core performance open data is your best bet for figuring out the difference.

The reason I say it’s a rough estimate is because Cycles scales very poorly on cpus after Cycles X.

Here are some tests I did with a python script I made.
Left = render time in milliseconds.
Bottom = threads used to render.

Sheit just noticing this is an old post agian. Wish BA didn’t suggest old posts to me when I’m on a helping spree lol.