CPU and RAM question.

[Drmattbob]

I currently have a 9900x and 32gb of ddr 3000 ram and a 2080.  I am buying a second system and someone else will be using that one.    Not thinking about money I am trying to figure out would there be a noticable difference getting a 3990x 64 cores CPU with like 64+ GB of quad channel ddr 4000 ram and a 2080ti or perhaps waiting on the video card as it seems the 3080 is due in a month? I experience a lot of waiting specially in 3d and when I am finalizing a large house plan that has been loaded out with everything.

 

 

Would it be money better spent getting like a 3970x instead and getting more ram or spending extra somewhere else? Not sure if going from 32 to 64 in chief would be a performance boost or not.  I do plan on setting up a high end cooling setup in this for over clocking.

[TheKitchenAbode]

In respect to CA, it's questionable as to whether you would notice any performance benefits from the upgrades you are proposing unless Raytracing is a significant part of your workflow, it's the one feature in CA that can take full advantage of all available CPU cores. That AMD 3990x is actually a bit slower in base and boost frequency than the Intel 9900X so single core and lightly threaded processes will be a bit slower, many processes in CA fall into this category. 32GB of ram is more than sufficient for CA. The speed of your ram should be as designated by the motherboard specs, using higher rated ram could cause instability issues. Moving up from a 2080 to a 2080ti might provide a bit of improvement but that would likely only be noticeable if you do complex PBR scenes, standard view types like plan and elevation are not very demanding. Chances are that any 3D view stuttering you are experiencing is CPU related, not your GPU. This is also applicable to plan view stuttering when panning or zooming. The best you can do here is to get the fastest base and boost frequency CPU you can afford, a high core count CPU will provide no benefit here.

 

Please keep in mind that the above relates to CA only. You may be using other software on a regular basis that may derive a benefit from your proposed upgrades.

[MarkMc]

On 8/15/2020 at 10:04 AM, Drmattbob said:

perhaps waiting on the video card as it seems the 3080 is due in a month? 

From what I've read it would be worth the wait...at least to find out after launch.

[TheKitchenAbode]

39 minutes ago, MarkMc said:

From what I've read it would be worth the wait...at least to find out after launch.

 

It will be interesting to see but I'm a bit dubious as to whether this will translate into improved CA performance unless CA can significantly change it's coding. Every time I have explored this GPU issue it always seems to come back to the CPU as being the primary bottleneck. This is not uncommon, many software packages are similar in that they are highly CPU dependent and as such the benefit between say a mid level GPU and a Top of the Line one is minimal at best. Unfortunately the way these cards are tested the results are not highly relatable to software such as CA. The best place I have been able to find testing that provides some insight into this are the test articles published by Puget Systems. If you take the time to read through them I believe it will become apparent that the benefit of high end GPU's is highly dependent on the type of software program one is using. They also have many other articles on all hardware aspects, CPU's, RAM and more.

[MarkMc]

2 hours ago, TheKitchenAbode said:

. If you take the time to read through them I

I had read them; did not see anything about CA. If building a desktop as originally planned I would do exactly what I suggest.

 

Things changed so getting a new laptop. The only suggestion I managed to get was definitely get an RTX card. Mobile 3000 aren't likely until Q1 or 2 of 2021 so not waiting on the lappie.

[TheKitchenAbode]

2 minutes ago, MarkMc said:

I had read them; did not see anything about CA.

 

There are no specific articles about CA, but there are articles about similar types of programs plus other graphics oriented software. The point I was attempting to demonstrate is that for many of these programs performance improvements based on generalities is not always reliable. In other words, my zooming and panning in plan view is laggy, therefore if I buy a graphics card that is 3 times more powerful than my current one then this lag will go away. I may end up disappointed as most of this type of lag is related to the CPU. Same as lag in 3D models when one makes a change to say the roof, the majority of lag is in the 3D model rebuild process that is also highly CPU dependent. Will a high core count CPU help in these situations, not much as the involved CPU processes are mainly single or only lightly threaded. For this you need as fast a base and boost frequency CPU you can get. I'm not stating that there is no potential benefit, just that it may not be as significant when basing this on generalities.