Questions regarding potential performance gains on upgrade

[S_K_I]

I'll be brief as possible, I'm looking to upgrade my rig when Ryzen 7950x comes next month out but need input on what I should expect to see performance gains wise in certain areas, but my specs on my old rig first:

I7-4930k o/c 4.1ghz
ASUS Sabertooth X79
ASUS RTX 3090
Corsair 32GB (4x8GB) DDR3 1600
1 TB Sabrent NVME drive
Corsair 860w power supply

So assuming we can speculate here for a moment, I'm more than likely going 64GB DDR 5 6000 RAM (budget wise) to pair up wih the Ryzen . Beyond that I'm already satisifed with the 3090 and intend to use that as my primary render for now and since I do very light Tyflow and Phoenix FD simulations. My understanding is it doesn't require core count so much more so how fast the clock is? Correct me if I'm wrong of course. Secondly since the RTX performance is equal if not better than a 3990x based off the benchmarks and tests I've seen on YouTube I'd rather go that route since it's also more efficient for my budget.

My main concern though are during the IPR and final renders because I'm just so damn tired of having 1 GB projects for my ArchViz and personal projects, and then having to proxy everything and wait for the compiling geometry, loading bitmaps, optix blah blah, before I can continue my workflow. I don't know what's the bottleneck in this case though and need input here. Which areas of the computer are handling these steps by the way? Any ideas how much gains I'd expect going from DDR4 to DDR5? Oh and another thing, what is rendering the previews in the material editor? CPU or GPU? Another pet peeve of mine.

Thanks in advance for feedback on this, and it's 2AM my time at the moment so I may not be able to respond till tomorrow since I'm running on E

Regards.

[Muhammed_Hamed]

but need input on what I should expect to see performance gains wise in certain areas

Hey S_K_I

Any new CPUs will be considerably faster than your machine. Your CPU dates back to 2013 and 2012 tech, it is been 10 years in which technology has progressed a lot
The biggest difference is single threaded performance, I will talk about the 5950X which is 2 years old by now. it performs at least 2x faster in single threaded tasks, most if not everything in 3Ds Max relies on single-threaded performance, all modifiers and tools, viewport..etc
you will get a much better experience

My understanding is it doesn't require core count so much more so how fast the clock is?

It is not much about frequency or clock speeds, lets say your CPU runs at 4.1 Ghz now. If I run a 5950x at 4.1 GHz it still will be a lot faster than your CPU in all respects
The 5950x and other newer CPUs have improved architecture, a lot better than your CPU

Your 3090 will perform better under a new CPU for sure

Secondly since the RTX performance is equal if not better than a 3990x based off the benchmarks and tests I've seen on YouTube I'd rather go that route since it's also more efficient for my budget

Yeah, this is correct. Although V-Ray and V-Ray GPU are 2 different engines so this comparison should be taken with a grain of salt
the general consensus is that GPU rendering is faster than CPU rendering for sure, and could be scaled easier on a personal level. Meaning you can add more GPUs to your new system, but not more CPUs

My main concern though are during the IPR and final renders because I'm just so damn tired of having 1 GB projects for my ArchViz and personal projects, and then having to proxy everything and wait for the compiling geometry, loading bitmaps, optix blah blah, before I can continue my workflow

This part is gonna be a lot faster with a new CPU, it relies on single threading. It is also something we can improve on our side specially with Forest pack and scattering plugins, it is on the GPU roadmap

Any ideas how much gains I'd expect going from DDR4 to DDR5?

Near zero, judging by the currently available CPUs from Intel that use both DDR4 and DDR5
This may or may not be different with the new AMD CPUs

what is rendering the previews in the material editor? CPU or GPU?

CPU

We have this on our GPU roadmap, to have the previews rendered with V-Ray GPU somehow. It is not an easy task sadly

Let me know if you have more questions

Best,

Muhammed

[S_K_I]

Thanks so much for this writeup and it reinforces the fact that going a faster cpu than core count is way better for a GPU build. Thanks man!