Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/908
Article Thumbnail

SOLIDWORKS 2017 AMD Ryzen 7 1700X & 1800X Performance

Written on March 2, 2017 by Matt Bach
Share:

Introduction

Whenever a new generation of CPUs is launched, the main question everyone wants answered is how fast they are. In the case of AMD's Ryzen, there are also a lot of questions surrounding how they compare to the processors available from Intel. For quite a while now Intel has held a dominant position in nearly every computing market, but there is a lot of hype around Ryzen due to the fact that you can get eight CPU cores for half the cost of an Intel processor of the same size.

There are a wide variety of tasks we could test in SOLIDWORKS, but in this article we will specifically be looking at:

  1. Opening/saving assemblies and drawings
  2. Rebuilding an assembly
  3. Performing a motion study
  4. Rotating a complex assembly model
  5. Running a simulation (FEA and flow)
  6. Rendering
Update 3/9/2017: Added benchmark results for the AMD Ryzen 7 1800X (results in original article were placeholders)

We are going to divide these tasks into three groups: general modeling, simulation, and rendering. If you would rather skip over the individual results, feel free to jump right to the conclusion section.

We also have a number of other articles looking at the performance of the AMD Ryzen 7 1700X & 1800X CPUs in other applications including:

Test Setup

To see how the new Ryzen CPUs perform in SOLIDWORKS, we used the following configurations:

These test configurations include three different platforms along with six different CPU models. For SOLIDWORKS, we would typically recommend a quad core CPU with a high frequency since most modeling tasks cannot efficiently take advantage of having more cores. However, since the Ryzen 7 1700X and 1800X CPUs both have eight cores we also wanted to include some of the "High-End" Core i7 CPUs with 6-10 cores. Having a higher number of cores should also give better performance when running simulations or renders so looking at a wide range of CPU core counts is a good idea anyway.

To make sure our results are as accurate as possible we used a combination of SOLIDWORKS macros and a custom AutoIt script to start SOLIDWORKS, load the relevant test file, then time how long it takes to perform the task we want to benchmark. The files we used were a mix of SOLIDWORKS training files and files available from GrabCad.com. These files and the associated tests are:

Test Files
File Open & Save Assembly - Vertical Twin Steam Engine with Reverse Gear (by Ridwan Septyawan)
Drawing - punch_holder (SOLIDWORKS Performance Test dataset)
Motion Study Gear Train Mechanism with Fixed and Swaying Axes (by trinityscsp)
FEA Simulation FEA Benchmark V3
Flow Simulation - Airflow Billboard - Lesson14 Case Study (SOLIDWORKS 2015 Flow Sim. training files)
Rebuild/Rendering Vertical Twin Steam Engine with Reverse Gear (by Ridwan Septyawan)
Model Rotation Audi R8 by ma73us

File Open/Save - Assembly

File Open/Save - Drawing

Motion Study

FEA Simulation

Flow Simulation - Airflow

Rebuild/Rendering

Model Rotation

 

General Modeling Tasks

SOLIDWORKS 2017 Ryzen 7 1700x 1800x General Modeling Benchmark

 

File Open

File Save

Rebuild

Motion Study

Model Rotation

 

Considering The Ryzen 7 1700X and 1800X are cheaper than the Intel Core i7 6850K (and 6900K for that matter), they fared pretty well compared to the "High End" Intel CPUs although they were certainly a good deal slower. However, most general tasks in SOLIDWORKS are single threaded which makes the Intel Core i7 7700K 4 core CPU much faster than every other CPU we tested in this article - including Ryzen. Specifically, the i7 7700K was on average about 35-40% faster than the 1700x and 1800x.

Simulation

SOLIDWORKS 2017 Ryzen 7 1700x 1800x Simulation Benchmark

Running simulations in SOLIDWORKS is able to take advantage of CPUs with a higher number of cores, although it is not as efficient as something like rendering. Interestingly, although the Ryzen CPUs were not a huge amount slower than the 6-10 core Intel CPUs for general modeling tasks, they did not do nearly as well when running simulations.

Compared to the Ryzen CPUs, the Intel Core i7 7700K (which is also less expensive) was about 22% faster on average. If you are looking for even faster simulation times, both the 6900K and 6950X are in the range of 35% faster than the 1700X - although you will certainly be paying a premium for that extra performance.

Rendering

SOLIDWORKS 2017 Ryzen 7 1700x 1800x Rendering Benchmark

Rendering in general is able to take great advantage of higher core count CPUs and the Photoview 360 rendering engine in SOLIDWORKS is no exception. Because of this the AMD Ryzen CPUs did extremely well, easily beating the i7 7700K by about 30% and the i7 6850K by about 20%. Compared to the i7 6900K, however, the Ryzen CPUs started to drop behind a bit with the 1700X coming in at about 5% slower than the 6900K while the 1800X was about 3% slower. Ryzen also couldn't keep up with the much more expensive i7 6950X (which was about 25% faster) but overall Ryzen is very attractive for pure CPU-based rendering performance as it easily beats the i7 6850K and just brushes up against the performance of the i7 6900K.

Conclusion

SOLIDWORKS 2017 AMD Ryzen 7 1700x 1800x benchmark performance

If you are looking for a short answer as to whether or not AMD Ryzen is a good fit for SOLIDWORKS, we would have to say "no". The Ryzen CPUs did extremely well in our rendering testing (easily beating the i7 6850K and matching the i7 6900K), but otherwise the performance was disappointing. In general modeling tasks, Ryzen fell far behind the Intel Core i7 7700K - performing on average about 35-40% slower. Simulations was a bit better but even then Ryzen was ~20% slower than the i7 7700K.

If you do a significant amount of rendering in SOLIDWORKS, Ryzen might be a good choice if you are on a tight budget but anyone doing that amount of rendering can likely either justify purchasing the faster and more expensive i7 6950X or will be using a GPU-based rendering engine like SW Visualize to get faster render times by leveraging the power of their GPU. For the vast majority of SW users, it is clear that you should stick with the Intel Core i7 7700K. Not only is it less expensive and uses less power (so it should run both cooler and quieter), it is also much faster for everything short of rendering.

Tags: SOLIDWORKS, CPU, Processor, Ryzen
Thomas Ewing

Really valuable data - thanks for going to the effort to put it together. I'd love to see some more CFD benchmarks comparing the AMD R7s to the extreme edition i7s, most especially with FLUENT. Has there been much interest in AMD based workstations yet?

Posted on 2017-03-09 03:38:03
Alex Taguchi

Great data! I was expecting better results in Sim/Flow comparable to the 6900k. However, based on previous articles you wrote, you saw that after the 4th core, solve times do not shorten in a linear way, and the ROI for more than 4 cores is low.

I think this shows that Sim/Flow are not for pure parallel tasks like PV360 is.

The other thing I've read elsewhere is that the x370 boards won't go higher than DDR4 2667, which hurts their RAM intensive tasks. But I see you are running 2400 on the x370 and running 2133 on the z270 and x99 test rigs, so the Intel rigs are still able to outperform in most scenarios even with slower RAM.

Posted on 2017-03-09 22:27:52
Alex Taguchi

Looks like 1800x performance is near identical to the 1700x performance. Is this maybe due to the XFR capability to allow the 1700x to boost even further and closer to 1800x clock speeds?

What cooling solution was used on the Ryzens?

Posted on 2017-03-09 23:37:01
4Head

Once again AMD fails to deliver in productivity and gaming too. Ryzen is a massive failure.

Posted on 2017-03-13 11:03:13
jopie1978

I have a question: Would I actually notice the difference in modeling-performance between a 7700K and a 1700x processor? Maybe my handlingspeed of peripherals is the true bottleneck in 3D-modeling? Then I would not notice the difference in performance at all.

If this bottleneck assumption is correct, the performance regarding modeling does not have such a big impact as the article has given it. Then the 1700X could be more interesting if people/students are on a tight budget.

Posted on 2017-03-13 16:14:30

You are right that it completely depends on your models. If you are not working with large enough models to give you <60FPS when rotating around with any of the CPUs we tested, then you probably won't notice any difference at all. However, something like the 7700K (with it's much better single-threaded performance) will be better for all the little things like how long it takes a GUI to come up, how long it takes to switch between parts/assemblies, how long it takes to open parts/assemblies, and things like that.

As for the 1700X being good for people on a budget, the problem there is that the 7700K is actually cheaper than the 1700X (around $350 versus $399) so for SOLIDWORKS the 7700K is both cheaper and faster than the 1700X. There is the Ryzen 7 1700 which is a hair cheaper than the 7700K, but in that case you might as well just use a i7 7700 (non-K). It will be slower than the 7700K by a small amount, but still much faster and cheaper than any Ryzen CPU.

Honestly, Ryzen just isn't designed for lightly-threaded applications like SOLIDWORKS so it just doesn't makes sense compared to Intel CPUs right now.

Posted on 2017-03-13 18:28:11
jopie1978

Thank you matt, your reply is very clear! I do not have experience with SW-visualize yet, so my point of view is sort of dated. With a simple Nvidia gpu, I should be able to reach faster rendertimes then with most cpu's to date and certainly mine. I feel old now, thanks!

Posted on 2017-03-14 10:15:04
Niko Nikolov

Loud love a similar ryzen+quadro test,but in a autodesk maya and max environment.Maybe as simple as a specviewperf (but in my opinion that one doesnt tell the whole story).

Solidworks works way different than max/maya,so it coud be a interesting thing to do.I hope you have the time for such a article as it can become complex.

Cheers!

Posted on 2017-07-15 14:01:18
jorge Hdez

Hi I'm trying to buy a pc for Maya, Blender, photoshop, premier, and I do not know which option to buy ...
1700x 2x gtx 1080
7700k 2x gtx 1080
2683 v3 2x gtx 1080.
Thanks........

Posted on 2017-03-15 10:24:58
Odin

Matt Bach Have you done a Cinema 4D Testing? I will be looking forward to it.

Posted on 2017-03-20 13:29:05

Cinema 4D is something we want to test in general, but we haven't developed a testing process for it quite yet. Rendering should be pretty straight-forward since we can just use Cinebench, but all the modeling and animation is harder.

For rendering, a lot of the reviews already have results such as this one from Hot Hardware: http://hothardware.com/revi... . For 3D modeling & animation, however, it should be pretty close to what we saw in this article for the "General Modeling Tasks" since both those and most everything in C4D (besides rendering) are single threaded. It may be a few percent different, but it should be pretty close.

Posted on 2017-03-20 17:37:41
ELLAS

Ryzen's rendering performance is insane. Good Stuff.

Posted on 2017-03-21 09:26:25

Super curious question, why is it when I scrub through 4k footage on windows media player it's lightning fast and when I scrub on 4k footage in Premiere it's not nearly as responsive? It's confusing the hell out of me however most people just keep saying "Adobe Premiere is your bottleneck." I'm so hoping this isn't true. Thanks in advance!

System:

Ryzen 1700x
GTX 1070
64gb DDR4 RAM
OS + Software - M.2 512GB
Projects / Assets / Saves - SSD 525GB OCZ x2 Raid 0
Cache + Page - SSD Samsung 128GB

Posted on 2017-04-04 19:15:23

Windows Media Play and Premiere Pro are completely different so it really isn't fair to compare the two. Media Player just has to decode and play the video - Premiere is doing a ton of other things in the background as well including live rendering of any effects you have applied. If you render previews in PP it might make the scrubbing between those two programs a bit closer, but it is still very much an apples to oranges comparison.

A more extreme example would be to stick a media clip in After Effects. Often you won't even be able to play it back at full resolution at all once you get above 4K resolution or so.

Posted on 2017-04-04 19:28:40

Awesome, that's what I more or less assumed though I guess I was hoping for faster scrubbing with no effects. Thanks for the added clarification, I needed a bit more resolve than just myself and non-editors who build computers for a living :)

Posted on 2017-04-05 00:54:47
trystudy

Very good article, help me to solve the problem of choice
.

Posted on 2017-06-24 12:15:56
Goh77

You should consider rerunning the Ryzens after all the bios updates. Their performance have improved in other areas after bios updates

Posted on 2017-06-29 11:20:22
Tuco Salamanka

really great article . could you please do these tests on the i7 8700k and add them to these comparisons , as I'm looking for a cpu that is best optimized for simulation and rendering, since the 8700k has relatively high core counts and much higher clockspeed than the ryzen 1700x, so I would really like to see how it will be compared to the 7700k and ryzen 1800x,1700x

Posted on 2017-10-07 23:22:15

We will be putting out a Solidworks article later this week with the new Coffee Lake CPUs, as well as the 7700K and 1800X.

Posted on 2017-10-09 16:05:53
AC

Please use Samsung B-die 3200CL14 on your Ryzen setup, or at least DDR4 2666MHz on all platforms. Also DDR4 4000+ on Z370 / DDR4 3600 on X299 if it's the EVGA mATX board.

Maybe also try a setup with a typical GPU for Solidworks too, most are on P4000 or P2000?

Posted on 2017-10-10 08:37:42

In so far as possible, we test with the top memory speed that is approved by the CPU manufacturer. In this case, that is 2666 for Ryzen and Coffee Lake, and 2400 for the older Kaby Lake. We want our testing to accurately reflect products we would actually carry, so we don't test at higher memory speeds that effectively overclock the CPU's memory controller (and thus often void the warranty). Matt wrote a good post about this, if you are curious for more details on our philosophy: https://www.pugetsystems.co...

As for the GPU, we are using the Quadro P6000 in order to avoid the video card bottlenecking anything. However, in our experience Solidworks isn't all that demanding on the GPU... especially in most of the areas we are focusing our CPU tests (rebuild time, motion study, simulation, and rendering). We are also testing FPS, which would be more directly impacted by the video card, but there your mileage is going to vary dramatically based on how complex of an assembly you are working on.

Posted on 2017-10-10 16:51:17
alan

good data, bare in mind the price of 1700x is like 300$

Posted on 2018-01-07 13:34:53