3900x vs i9 9900k for Unity/Blender/Zbrush

Hey,

I have Ryzen 1700 atm and I was thinking about upgrading my PC because 1700 isn't enough for single core performance.
I am a bit confused though because Zbrush is mostly multi-threaded app but when I turn off 8 threads on my 16 thread CPU I see NO difference in perfromance at all.

New AMD bioses and mobos are having problems and aren't 100% stable. Intel's platform is a lot stable and older/more mature.

Can higher single core perfromance influence my overall zbrush performance/snappiness/smoothness?

Can someone explain it to me, what's going on? Is it even worth to buy 3900x that has lower single core performance but has more threads (24) than intel? What is more worth it?

Has anyone benchmarked i9 9900k in zbrush/blender/unity?

I also stream, play video games and do video editing/rendering as well.

 

Blender is one of the standard benchmarks these days. A 3900X will absolutely destroy a 9900K in Blender.

https://www.pcworld.com/article/3405567/ryzen-3000-review-amds-12-core-ryzen-9-3900x.html?page=2

I have no idea about ZBrush but I do know Unity will benefit massively in some tasks (eg lightmapping) from very high core counts and for those tasks the 3900X will dominate the 9900K.

For the record the difference in single core performance is almost negligible. In the majority of workstation benchmarks it is almost within the range of a round error. You won't be able to see the difference outside of benchmarks.

 

I dont know which benchmarks you refer to the 3900 is on par with the 9900 in many benchmarks, like linus,

I'm waiting for the 3950 (Since I have the 8 core 2700x and would like to double those), but If I bought today I would go with the 3900x over the 9900k

 

Have you seen the compiling benchmarks that have been getting attention because some of the demanding stages fit into the massive cache of the 3900X? I happened to find out about it last night through this video.

Spoiler

 

Nice; yeah, the new cache is a big stepup from the 2xxx series. I'm just skeptical that memory bandwidth on the AM4 is enough to drive 16 cores 32 threads. Will wait until benchmarks arrive.

 

The only problem with 3900x for me is how unstable it is. (3000 series)

If you go to amd reddit all you can see is complaints how idle temps are high, how vcore is high (1.4-1.5) how boost clock isn't what they advertised (should be 4.6ghz but is 4.3.-4.4 max) and so on and on. Some people can't post (boost their PC) or have blue screen errors all the time.

This is THE ONLY thing that prevents me from buying 3900x. Not sure what to do. I want to build a new PC in 2-3 weeks tiem from now... I don't want to wait 6 months for those things to be fixed etc.

 

I have missed this, sounds like the orginal Ryzen release all over again. Why dont they learn.

 

Yep, it is sad to see because the CPU is ssooooo good!!

Why do they always have to do it at the release. It's been 1 month since the launch and they haven't fixed these problems. ;/

 

One major problem with reddit is that there usually isn't sufficient information to determine if the problem truly is the fault of the manufacturers or if it's the fault of the person building the computer. That said I don't buy a processor as soon as it's launched for the simple fact that the firmware hasn't had time to mature.

 

It wouldn't surprise me if its legit though. AMD are pisspoor at QA. It took so long until the AGESA was stable enough for one.

 

They told it might takes weeks to update bioses to fix these issues. So, these problems aren't "fake". They are legit problems many people are having. They've been confirmed by AMD.

I don't render with my CPU but with my GPUs. I don't bake lightmaps with CPU but with GPU as well.

I am not sure how important is for me to have these 4 extra cores.

One more thing, 3900x cpus will be available in my country in 1-2 months. Nobody is selling them atm - no stock.

 

Good to know.

If you need a processor now then the 9900K is an immediate and substantial performance increase over the 1700, but keep in mind you will need a new motherboard with that processor. Alternatively you could jump to a 2700X and then jump to a 3000 series processor once it's stable.

https://www.newegg.com/amd-ryzen-7-2700x/p/N82E16819113499 - 2700X - $255

https://www.newegg.com/core-i9-9th-gen-intel-core-i9-9900k/p/N82E16819117957 - 9900K - $480
https://www.newegg.com/p/N82E16813145089 - Z390 MB - $280

That motherboard is more expensive than that 2700X, and once the 3900X is stable the cost will still be nearly identical to the 9900K which means that the cost of both paths is basically identical.

 

In my country, 3900x costs less than 9900k - both at the same price but AMD doesn't require an aftermarket cooler.

I will wait for 3900x then. 1700 to 2700x isn't that big jump for me probably... I will think about it.

I might just buy 2700x and new x570 board to prepare for 3900x when it is available in my country.

 

Here is a link detailing some of the issues that people have been having with the 3000 series Ryzen CPUs:
https://www.tomshardware.com/reviews/amd-ryzen-3000-turbo-boost-frequency-analysis,6253.html

 

This doesn't make much sense. Everyone isn't having these problems, and a BIOS update implies an issue with a specific motherboard, not the processor. AMD would be unlikely to even confirm such an issue themselves, instead punt it to the actual responsible party (motherboard manufacturer).

Rather than focusing on what you're not doing, look at the multithreaded workloads you are doing. If they would benefit from the additional cores, then they may actually be important.

It is not unstable

Idle temps high is almost certainly a mounting issue. It isn't AMD's fault if you don't install your CPU heatsink properly. Remember that any moron can post on reddit.

Vcore high from all accounts appears to be caused by background applications keeping the CPU from going into an idle state, which just means you'll use a bit more power at low load than necessary, but otherwise shouldn't be any problem. Stop/disable the background apps which are waking up the CPU, or ignore it. Maybe AMD can issue a chipset driver update for this, but I wouldn't worry about it.

The complaint about boost clocks not at 4.6ghz sounds like a lack of understanding how zen 2 boost clocks work. Here's a good explanation:
https://www.gamersnexus.net/guides/3491-explaining-precision-boost-overdrive-benchmarks-auto-oc

 

Having watched quite a few benchmark videos the boost is still far superior to that of Intel, but regardless of whether it's high or low the benchmarks and how it performs in the real world are what matter. Getting a high clock rate doesn't mean much of anything these days. That obscene cache size is having a far greater impact than a couple hundred MHz would.

 

Exactly. The actual clock speed matters far less than real world performance.

The AMD 3000 series CPUs manage their clock speeds similar to how modern graphics cards manage their clock speeds. There are curves regarding input power, thermals, etc. It is jarring for some people because it feels different, but it is a solid way to deliver performance, power, and thermal goals.

 

All we need now is working gfx jobs/vulkan (Hint, hint Unity)

 

Thanks for this.

I've been very impressed with AMD's recent releases which on paper and in plenty of benchmarks look pretty awesome vs the cost of Intel. So much so that I felt like my next PC would use one ( and probably still will), but at the back of my mind I always wondered just how much of a benefit more cores would have for Unity in terms of building or just general Editor usage ( e.g. importing assets ).

However judging by your results it sadly seems to be 'not that much' of a benefit with more cores for building, probably even worse for me since most of my client work takes in the order of a few minutes maximum to build compared to your hour.

I already have a 6 core/12 thread 5930K so its kept up pretty well and while i'd certainly see some boost in core clock speeds, i'd likely need to jump to 8 or 12 cores to get a big improvement in other apps and that will add additional cost.

Shame really, I wonder if there is more work Unity could do here to make the editor and build use more cores? I tried looking for some official documentation on whether Unity utilizes more cores for importing/building but couldn't find anything at the time. Also wish all these benchmarking video's would include a Unity benchmark, something like building one of Unity's own projects.

 

I own two similar PCs, one with an AMD 3900X and the other with an Intel i9-9900K. Both systems are fast. The Intel benchmarks slightly faster in some games, but the AMD definitely feels faster when I am working. And even when a specific application is not actively using all 12 cores, the OS still finds other work to do with those other cores. The AMD system feels very responsive all of the time.

Remember that both the latest Intel and AMD systems will be a noticeable performance boost for you. Your i7-5930K is Haswell based 2013/2014 technology. Those older Intel CPUs took some real performance hits after the Spectre/Meltdown patches for the OS. My AMD rig actually replaced my old i7-4770k workstation, which was also Haswell based. The difference between the Intel i7-4770K and the AMD 3900X was absolutely massive on the same workloads.

 

Agree with everything you say ( esp. with regard to Spectre/Meltdown ), but it still stands that apparently from the original post ( which was strangely deleted a few minutes after I posted my reply??? ) that Unity doesn't appear to take advantage of more cores when building beyond a certain point. They showed a screenshot that showed about 6-8 cores from something like 24 being used whilst building.

The OP did state it improved in loading, but I guess my point was it would be nice for Unity to provide some benchmarks or guidance as to what area's can be improved with more cores vs clock speed so as to inform their customers so we can make better choices rather than guessing.

 

To chime in i have a 3900X use unity play games time to time. To me seems like a rock solid processor, changing code to going back to the editor is way faster then on my old 6700K. i totally recommend it.

 

If I bought a processor for myself, I would choose Ryzen without a doubt.

 

Let's assume for a moment that it's completely accurate. Are you only running Unity by itself in a vacuum? No. At the very least you have your operating system which can decide in the middle of the day to run Windows Update (which is now multi-threaded by the way) and you have support apps like graphics and printer control panels.

Beyond that most of us have multiple applications running. Programmers will typically be running an IDE. Artists will typically be running at least one content creation program if not multiple programs (a 2D app, a 3D app, etc).

Speaking of games I like to play one when I start Unity on a long demanding task but I can't currently do that because my Ryzen 5 1600X (6C/12T) CPU is barely strong enough to run the task in Unity. If everything works out this year my plan is to upgrade to a 4000 series (or a 3000 series chip if the price is right) CPU with at least twelve cores.

 

The mod censored my post as well.

My pipeline always jammed at building the 50 over scenes, because of small adjustment. e.g text realignment.
At this point, simply can't find any justifications for a threadripper or Mac Pro.

 

Weird your post was censored, I don't remember there being anything in it to justify that, but there was so very useful information in terms of multi-core usage.

 

In almost all multi-core tests, the Risen 9 3900x wins the 9900k. Given the fact that they cost the same, it is better to take Ryzen. Of course, there are no I9 competitors in some tasks, but I doubt that the difference is that big. Keep in mind that the 3900x has more cache memory, is built on 7-nanometer technology, and has PCIe 4.0.

https://nanoreview.net/en/cpu-compare/intel-core-i9-9900k-vs-amd-ryzen-9-3900x