Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1053
Article Thumbnail

SOLIDWORKS 2017: Coffee Lake CPU Comparison (i7 8700K, i5 8600K, i3 8350K)

Written on October 13, 2017 by William George
Share:

Introduction

Intel has just released an update to their mainstream Core processor series, marking the 8th generation of CPUs in this lineage. These are also the first mainstream processors to have more than four cores, even while maintaining high per-core clock speeds. As with any new launch in this segment, we want to see how well these new chips perform on a variety of applications. This article will focus on SOLIDWORKS 2017 SP4, and compare the new models to the previous-generation Intel offering as well as AMD's competitor.

Intel 8th Gen Core i3, i5, i7

Test Hardware and Methodology

To see how the different new Intel CPUs perform in SOLIDWORKS 2017, we used the following configurations:

These test configurations include three of the new Coffee Lake CPU models, the top of each tier (Core i3, i5, and i7) - as well as the previous top mainstream processor from Intel, the Core i7 7700K, which has been our go-to recommendation for SOLIDWORKS for several months. AMD's top Ryzen 7 model, the eight-core 1800X, is also included for reference.

The tests conducted on these systems were originally developed by my colleague here at Puget Systems: Matt Bach. He put together a series of AutoIt scripts that run through testing a variety of the capabilities in SOLIDWORKS, so rather than reinvent the wheel I used his. However, some of the results are being omitted due to hardware configuration issues. Specifically, the tests looking at how long it took to start SOLIDWORKS, open files, and save files are not included here. That is because I discovered after testing that the Coffee Lake system I was working with had both SATA and NVMe drives, and the Windows & software installations were on the SATA drive. I should have double-checked for that before testing, but I failed to... so those results could be impacted by the slower drive. The rest of the tests, though, will be unaffected by this.

So, what did we test?

Test Files
Assembly Rotation Audi R8 by ma73us
Rebuild/Rendering Vertical Twin Steam Engine with Reverse Gear (by Ridwan Septyawan)
Motion Study Gear Train Mechanism with Fixed and Swaying Axes (by trinityscsp)
FEA Simulation (Stress) FEA Benchmark V3

Flow Simulation (Airflow & Thermal)

Billboard - Lesson14 Case Study (SOLIDWORKS 2015 Flow Sim. training files)

FEA Simulation (Stress)

Flow Simulation
(Airflow & Thermal)

The results are broken up into sections below, followed by the conclusion. Each test was run three times on every CPU: the worst result was tossed out and the other two were averaged. This was done to help eliminate any influence from background applications running on the systems, but without cherry-picking only the very best result either.

Feel free to skip ahead if you are just interested in a specific set of results, or if you want to get right to the conclusion.

Results - Assembly Rotation (FPS)

Here are the results for rotating an assembly of an Audi R8 with 434 parts and about 1.4 million triangles:

SOLIDWORKS 2017 - Rotate Assembly

In this first set of tests, Intel's Core i7 processors - both the new 8700K and the 7700K - perform very well. They are neck and neck, with each edging into the lead in some places. Interestingly, the newer six-core Coffee Lake processor does slightly better with Realview and Ambient Occlusion enabled... but the results are so close that might just be the margin of error.

The new Core i5 8600K isn't far behind its i7 brothers, only 5-10% slower and still putting in very respectable FPS numbers. The Core i3 8350K and AMD's Ryzen 7 1800X are almost tied as well, showing that this test is not utilizing the Ryzen's substantial number of cores.

Results - Rebuilding Assembly

This graph shows how long it took to rebuild an assembly on each CPU:

SOLIDWORKS 2017 - Rebuild Assembly

Rebuilding an assembly is pretty much single-threaded, which means the high core counts of the newer i7 and Ryzen processors aren't going to help them here. The 8700K does edge out the 7700K by a couple percent, probably thanks to a slightly higher max turbo speed, and then the i5 8600K and i3 8350K are a bit slower (as they don't turbo as high - or at all, in the case of the Core i3). AMD's Ryzen scores particularly poorly here, faring even worse than the 8350K, since its high core count doesn't matter at all in this situation.

Results - Motion Study

Here is a chart showing how long it took to perform a motion study on each of the test systems:

SOLIDWORKS 2017 - Motion Study

Much like the previous Rebuilding test, the new Core i7 just barely edges out the old one here... in fact, they are close enough to be considered the same as the difference is well within any margin of error. And again, the Core i5, i3, and Ryzen processors come in behind the i7 models (in that order). This test looks to be limited by performance per clock, with little or no impact from core count.

Results - Simulation

This comparison shows the time taken to perform three different simulations on each processor:

SOLIDWORKS 2017 - Simulation

Simulations in SOLIDWORKS appear to benefit from both core count and clock speed, with AMD's Ryzen processor doing better here than in the preceding tests... but still coming in behind both Core i7 models. The aging 7700K actually puts in the fastest times on the stress simulation, which appears to depend more on pure clock speed, but the newer 8700K with its higher core count pulls ahead on the airflow and thermal simulations. The Core i3 and i5 models, lacking Hyperthreading, do much worse in those two simulation types.

Results - Rendering

Here are the times taken for both the render pre-pass and the main render itself, at 1920x895 resolution, on the various CPUs:

SOLIDWORKS 2017 - Rendering

This final test is where CPU core count really shines, and the one place the new Core i7 8700K with six cores pulls ahead of the older 7700K. However, AMD's Ryzen 7 1800X pulls a fast one on both of those with the fastest render times of all five tested CPUs. This makes sense, given its 8 cores (vs 6 on the 8700K and 4 on the 7700K), which overcomes the lower per-core clock speeds it runs at.

Conclusion

Combining the results of the previous charts, here is a summary of how these CPUs perform in SOLIDWORKS 2017. Please note that performance here is relative to the Core i7 7700K, our go-to recommendation for general SOLIDWORKS usage in recent months:

SOLIDWORKS 2017 - CPU Comparison Summary

Compared to the previous i7 7700K, the new "Coffee Lake" Core i7 8700K doesn't really bring anything to the table for general SOLIDWORKS use. It lands within a percent or two of the 7700K on most of the tests we ran, so for most folks there wouldn't be a need to upgrade. However, it is substantially faster for rendering: on average, 30% faster than the processor it is replacing. That is a pretty big jump, and entirely due to the additional CPU cores. The lower-end Core i5 and i3 models are solidly behind the i7 chips, though, so unless you have a very limited budget there is no reason to go with one of them.

AMD's Ryzen is an even more polarized option in SOLIDWORKS, though. The Ryzen 7 1800X surpasses even the new Core i7 8700K when it comes to rendering, thanks to having 8 cores, but in the rest of our testing it came in behind the Core i3. If you are doing nothing but rendering in SOLIDWORKS, then it could make sense... but the Core i7 is a much more balanced choice overall, for roughly the same price.

Are there other processors you would like to see tested in SOLIDWORKS? It can be safely assumed that the higher core count AMD Threadripper and Intel Core X series processors would do even better in rendering, but probably suffer somewhat in more general usage. Please leave your thoughts and suggestions in the comments below!

Tags: Dassault, Systemes, Solidworks, 2017, CPU, Processor, Coffee, Lake, Core, i3, i5, i7, Kaby, Ryzen
Tarantula

Hi, I would like to know if the 20%-30% deficit that the i3-8350K is due to CPU clock frequencies since some technical departments even recommend to turn HT off in order to avoid degrading some single threaded operations.
I would like to quote this article on Dasi Solutions http://blog.dasisolutions.c...

" In SOLIDWORKS the following operations are known to be parallel and can take advantage of CPU multi-threading:

Open SOLIDWORKS documents, drawings and assemblies
Photoworks rendering
Simulation calculations
Draw compare
Update drawing views
A large amount of SOLIDWORKS operations are sequential and single-threaded; therefore, Multi-Threading/Hyper Threading does not make much of a difference in such operations. In some cases, single-threaded operations can be degraded due to slower frequencies and additional pipeline stages that are necessary to accommodate thread-switching hardware. Therefore, SOLIDWORKS may not see much performance improvement when using Hyper Threading."

Posted on 2017-10-18 15:17:13

That depends on which tests / results you are looking at. With rendering, for example, the Core i3 scores very poorly because of three factors

1) Fixed clock speed (4GHz, no turbo)
2) Fewer cores (4 compared to 6 on the new Core i5 & i7 models)
3) No Hyperthreading

However, in the other tests things are more variable. For example, Motion Study appears to just depend on clock speed - and the Core i3 falls exactly where I would expect with that (about 10% behind the i5 and 20% behind the i7). In simulations, though, other factors like core count and Hyperthreading lead to a bigger difference... though not as dramatic as what you see with rendering.

As for turning HT off, there are a couple of applications I know of where that can actually help - but for the most part, it is best to just leave it on. Some applications may not benefit from it, but it is very rare for a program to perform worse with HT enabled.

Posted on 2017-10-18 16:35:02
Christopher Ragsdale

Great Article!

I am very interested in seeing some SolidWorks benchmarks for the Core X processors. Specifically, I'm curious as to the impact of quad channel memory use vs the dual channel (if any), and more importantly how the "turbo boost 3.0" increased clock rate will impact the performance during single threaded operations. I wonder if the "spool up time" for the turbo boost vs just having a higher base clock has an impact. I doubt there will be any better scaling in SolidWorks simulation (or most other FEA/ CFD applications) than what your previous testing demonstrated, but not having the communications bottleneck of a second CPU might show a closer approach to the asymptotic max performance core scaling wise.

Puget Systems has the best professional application benchmarks that I've seen on the web or otherwise and it is greatly appreciated! You truly give CAD/CAE users such as myself the ammunition to go propose hardware improvements backed by solid data. Thank you!

Posted on 2017-10-20 00:36:56

I've actually been doing some testing on the 14-core i9 7940X this week, and this is what I found compared to the i7 8700K:

- Rendering is over twice as fast, which is to be expected with a little more than twice as many cores
- Simulations are a mixed back, ranging from 5% slower to 20% faster (also somewhat expected, depending on how well threaded each type of simulation is)
- General modeling stuff is a lot slower, moreso that I would expect. I figured we'd see maybe 10-20% slower performance, based on clock speed alone, but (for example) rebuilding an assembly is taking more than 30% longer. This trend is seen across start-up, open & save times, FPS, etc.

That last part seems fishy to me, so I am going to see if there is something amiss with my OS or SOLIDWORKS installation before doing any more Core X testing (or publishing any results). I'd like to put together a follow-up article with Core X vs Threadripper, but I am also aware that 2018 just came out... so it may make more sense to update our testing for the new version before publishing more articles. At the moment, though, I would strongly recommend the Core i7 8700K for most SOLIDWORKS usage - only going above that, to the Core X models, if you do a lot of rendering.

Posted on 2017-10-20 23:32:23
Christopher Ragsdale

Thank you very much for the response!

I am as surprised as you are that the core i9 processor performed so poorly with general SolidWorks tasks (beyond the obvious clock rate difference). Maybe turbo boost isn't triggering properly since it is dependent on things like workload type, processor power consumption, and processor temp.
My biggest issue with moving to an i7 8700K has to do with availability of PCIe lanes. I am currently using an i7 6850K processor with 40 PCIe lanes (and using all of them) for things like PCIe based nvme storage, scratch disks, and networking. The 8700K only has 16 lanes available without going through the motherboard PCH bottleneck so I'm trying weigh the performance drop in I/O vs clock rate. Not so much an issue with SolidWorks, but I use NX NASTRAN fairly extensively and it is much more sensitive to I/O performance. If the 16 PCIe lane processors end up being the best option overall then I wonder what the performance difference (cost of the motherboard aside) between the 8700K and the 7740X. The core count and base clock rate differences might make for an interesting, if not so much economics minded case study.

Thanks again for your help! I look forward to the new benchmarks on SolidWorks 2018!

Posted on 2017-10-21 02:19:06
SBS Extensions

Nice article William, straight to the point and quite informative.

One thing I am wondering about is the relationship between all the SolidWorks articles/benches done by Puget Systems using desktop CPU (with the occasional X99), and Puget's 'Recommended Systems for SOLIDWORKS', where you start of with an X299 based system. Have you conducted internal tests, which go counter to the Zx70 and Ryzen systems to go with X299?

If time permits, it would be nice to see the above article, but with 8700K vs 7820K vs 1800X vs 1900X.
Thank you.

Posted on 2017-10-26 03:55:56

We have some preliminary data on a couple Core X processors, and we know they do better with rendering (which is the main reason we list a second, more powerful system for SOLIDWORKS). I was thinking of doing a Core X vs Threadripper comparison - focusing on rendering, but including other SW performance results as well... but then SW 2018 came out, so I thought it might make more sense to wait and update our testing for the new version first. But if I do that, the previous Coffee Lake and Ryzen results would not be directly comparable. I'm not sure what all I will end up being able to fit in, but I am aware that SW results with the bigger CPUs would be helpful for folks :)

Posted on 2017-10-26 16:39:25
SBS Extensions

Cool. Thank you for the reply.

Posted on 2017-10-27 02:22:31
Loonster

I would be very interested in seeing how the Xeon W-2125 compares to I7-7740X (and add in I7-7700k for a reference). All cpus are 4 core / 8 thread, 4.5 GHZ turbo. The question to answer is 'Does the improved cache structure make a non-negligible difference in CAD workstation builds?'

The difference could be anywhere from negligible to profound. My money is on negligible (<1%), but hopefully I am proved wrong.

Posted on 2017-12-10 03:09:47
Steve Zimmerman

I would be very interested in seeing the differences between the i7 and Xeon processors as well. We have been buying workstations with the Xeon lowest core fastest processor we can and now i'm wondering if the i7 is a better option.

Posted on 2018-01-05 20:52:18

We are going to be doing some updated Solidworks testing soon. May I ask which Xeons you want to see as comparisons? I cannot promise anything, but I do appreciate reader feedback and suggestions :)

Posted on 2018-01-08 21:33:52
Loonster

The Xeon W-2125 is the only one I would consider buying. This is approximately a $275 premium over the 8700k (CPU + MB + 16gb RDIMM).

The higher core skylake-w do not make economical sense to me. The W-2135 gives 2 more cores for $400 more than the W-2125 (same cache amount). The W-2145 is twice the cost of 7820x (ignoring MB + Ram costs).

Posted on 2018-01-09 04:12:57

I'm doing some updated testing for SW 2018 now, but unfortunately I don't have access to a Xeon W compatible motherboard... so I won't be able to include these CPUs, as I had hoped. At least I'll have date points for other CPUs (Coffee Lake, Skylake X, and Threadripper) to compare with if / when I am able to test some of the Xeons.

Posted on 2018-02-01 17:48:35
Özgür A. Ergin

i would like to see w-2145 in test

Posted on 2018-02-01 17:44:27