Will this PC upgrade even work?

Hi guys, I know there are multiple posts on PC upgrades, but I’m struggling to find a specific enough answer to this question.

Background Info: I use blender mainly for Archiviz on the cycles engine. With the potential in the future to do ‘fly-though’ animations. I’m currently having issues with blender freezing and becoming very slow with large sub surf modifiers (I use these for displacement for roof tiles etc.).

Current Build:

  • AMD FX-8350 8-core CPU
  • 12GB DDR3 RAM (not sure on speed)
  • MSI Gaming X RX580 8GB GPU

I want to upgrade to DDR4, but due to cashflow, need to do the upgrade in 2 stages.

Stage 1 - Upgrade to Ryzen 7 3700X CPU
Upgrade to 32GB DDR4 RAM - 3600Mhz speed
Upgrade mobo and case to suit.

Stage 2 - Upgrade to RTX 2080 SUPER or 2080ti (depending on budget at the time)

My questions are - Will I see any performance increases after stage 1 at all? Specifically with the issues I’m having with large sub surf numbers (generally between 6-8 divisions).

Will the GPU bottleneck be too large while temporarily using the RX580 and make blender unusable completely?

I’ve been told I’d be better to upgrade the GPU first, but my current case won’t support a card of this size so really stuck with doing the upgrade this way round.

Sorry for the long post, tried to include as much info. as possible. Any help would be greatly appreciated!



It seems like a good plan for me.
From FX-8350 to Ryzen 7 3700X you will definitely have better single and multi-thread performance.

You will also have better performance, but you should also consider that current SubSurf implementation in 2.8 has many performance problems which are in the development plans to improve it.

Perfect, Thanks YAFU.

I wasn’t aware there was 2.8 issues with sub surfs. I’ll keep a look out for these in the patch notes.

It’s the slow, sluggish viewport and viewport rendering that has prompted the upgrade. Just wasn’t sure what hardware blender mainly uses for computing sub surfs and particle systems etc.

Thanks for the reply

You search in this forum or developers devtalk forum for topics related to OpenSubdiv 2.8 implementation or slow animation en viewport in 2.8, for example:

CPU has a lot to do, and Single thread performance is no less important than Multi thread performance. The CPU you chose for the update is good in both cases.
Also with SubSurf and particles it may be that RAM is filled and begins to intensive use virtual memory on disk, so the whole system becomes slow. But you also have plans to increase system RAM.

That’s awesome.

Feel confident in the upgrades I’ll be getting now.

Will be nice to be able to have an uninterrupted workflow eventually.

Thanks again, been a massive help.

By the way, in that thread I found this message that may interest you:

But my recommendation is that you do that upgrade anyway, the CPU change from FX-8350 is really great, and eventually SubSurf problems will be solved.

That’s an interesting read. I’m definitely going to go ahead with the upgrade.

Is it when used in animations that have the biggest problems?

I simply use a SubSurf modifier on roofs (I use ‘simple’ setting, with somewhere between 6-8 divisions). I apply a height map on top of the texture to give some depth to tiles such as pan tiles.

Once I’m happy with the result, I then turn off the subsurf visibility in the viewport to not overwork the PC

Subsurf just gives me that ability to turn on/off the high-poly that subdividing the mesh doesn’t give me.

If you disable it from the modifier or globally from the Simplify menu, then you should have no problem while working in viewport.