Your Best Source for CAD, CAM & CAE Articles




Home
| About | Submit Article

CAD, CAM & CAE Articles

Partners

CAD, CAM & CAE News

TenLinks.com

More News

 

 


 

  Exclusive Article    

Industry-Experts Cover
Products & Services You
Need to Know About.
Only on CADdigest.com.
Find Out More!

Hardware Review: No Limits with Lenovo's ThinkStation D30

By John Evans, January 11, 2013

Computer workstation reviews are helpful to the professional community in seeing what current trends are, and in which hardware to invest - when the time comes. Unfortunately, reviews rarely involve the experience that professionals encounter while working with real-world problems.

So this review is intended to convey a sense of what to expect from the Lenovo ThinkStation D30 desktop workstation in the daily work environment of a professional engineer. I’ll look at how design and analysis software behaves on this computer. If a new workhorse CAD/CAM/CAE workstation is on your horizon, then this review is for you.

In this article, I will tackle the following topics:

Figure 1: Lenovo ThinkStation D30 front view

Lenovo ThinkStation D30

The ThinkStation D30 is Lenovo’s answer to the question, "Exactly how much performance can I get in a box that one man can carry?" It is built like a tank, with exposed carry handles fore and aft (see figure 1).

Here is a run-down of its size and some of it specifications (see figure 2):

 

Figure 2: Interior layout

The Reference Workstations

Occasionally I’ll refer to a ‘reference workstation’ to establish some form of comparison between the D30 and other hardware that is commonly available in the industry. The following computer system supplies that reference:

The super stable Dell M6400 mobile workstation with

The smooth HP EliteBook 8740w mobile workstation with

SPECviewperf Benchmark

The industry standard benchmark that I used for this review is SPECviewperf 11. I performed the test at a screen resolution of 1920 x 1080 and with full-scene antialiasing.

Viewset  
 Composite
Catia-03    60.82
EnSight-04    43.39
Lightwave-01    68.10
Maya-03    98.91
ProE-05    11.28
Solidworks-02    60.49
TeamCenter-02    42.14
SNX-01    46.22
Modeling Performance with Inventor Professional

Using Inventor on this the D30 felt nice to me, because of the amount of RAM and number of CPUs available in this computer. While Inventor software cannot benefit directly from the 16 cores*, with this much throughput, Inventor did not have to wait or share the processor time. I doubt that the experience that the D30 provided could be bettered on another workstation; nevertheless, I would like to get a crack at it with all 256GB RAM on-board!

(* Note about Inventor and single-threaded multi-core use, from Autodesk Inventor Services & Support site: "On a dual-core computer, a CPU-intensive operation that uses 100% of the resources of a single-core processor will only use a maximum of 50% of the CPU for that same operation on a dual-core computer, and only 6% of each CPU on a 16-core computer. Due to the lack of multi-threading, Inventor is not capable of using more than 50% of the CPU on a dual-core computer, so there is no significant performance gain over a single CPU computer. The only way to take advantage of a dual-core processor when using Inventor is to run multiple Inventor sessions on your computer.")

Parts and Assemblies Benchmarking

I ran some benchmarks with Inventor, comparing the time actions took with the D30 workstation to the reference computers

Figure 3: Ray traced rendering of RC car in the Autodesk Inventor workspace

Varying arrays of assemblies were managed and revised. I tested assemblies with up to 300 components and witnessed no degradation of performance. Part modeling is impossible to relate as the results were almost instantaneous.

Graphics Benchmarking

The graphics display was exceptional. The Quadro 5000 is a wonderful workstation graphics card and so I expected nothing less. Antialiasing was good, and ray-traced images were clear and attractive. Those two-second interactive ray tracing updates with all effects on and 147 components is smoking fast!

Multiple Sessions of Inventor

I opened 16 sessions of Inventor with various parts and assemblies. As all 16 could not be worked on simultaneously, I was unable to bog down the processor this manner. However, nothing I did caused Inventor to pause or lag, such as when I moved rapidly from session to session, or adjusted views and features. Quite impressive.

Inventor Fusion

I use Fusion to handle most of my product simplification and defeaturing needs. There is not a better product for the task, but Fusion is by and large is a resource hog. This can make me irritable when running it on common workstations. The D30 seemed to offer Fusion a bit more room with which to work.

As I would expect, Fusion’s performance in most cases was improved on the D30, due to so much horsepower under its hood. Just like Inventor, the Fusion software does not support multi-threaded operations and so cannot use the multiple CPU cores. However, the experience was substantially more stable than on other computers. While still exhibiting slightly laggy responses on the D30 (as compared with Inventor), I was able to get my work completed - without walking on egg shells.

Some of the results I experienced when using Fusion on the D30 instead of other workstations:

Defeaturing

I engaged in operations that would typically crash Fusion or cause delays greater than a minute on standard workstations. While there was one time Fusion crashed on the D30 (I had asked for something insane), the overall performance was good. Operations that I could never get away with became possible on the D30, with only a small delay, which I consider to be a favorable tradeoff.

I experienced substantially improved response and stability while deleting features, something that represents 60% of my typical work in Fusion, and so this was a welcome relief.

Graphics

The graphics were good, but not exceptional. Responsiveness was good with visual effects off; with all effects on, the response was only fair.

Overall, I would score the Fusion experience much lower than with Inventor, because the experience was not exemplary. This is due to, unfortunately, the Fusion architecture.

Simulation Mechanical

Autodesk Simulation Mechanical was so much more capable on the D30 than my previous experiences with other workstations. Performing analyses on the D30 was just amazing. It’s not that elves came out of the machine and prepared the setup for me, but the machine never balked at anything.

Static and Dynamic analyses were performed not only with good speed, but superb stability (see figure 4).

Figure 4: Graphs of this complexity in Simulation Mechanical are not handled kindly without substantial resources; on the D30, they were no problem at all.

I performed simple static analyses, such as bending stresses, and the D30 handled them without effort. The real magic occurred with mechanical event simulations (MES), specifically one that contained 26 components and 91,659 mid-plane elements, and involved 200 time steps over a 1-second sine-shaped load curve (see figure 5).

Figure 5: Mesh and results visibility of a mechanical event simulation in Autodesk Simulation Mechanical

I had previously performed the analysis on a cloud server and a service provider’s in-house 16-core computer system. The analysis results were, of course, identical, but the D30 beat the service provider’s time even with less RAM:

I cannot tell you how fantastic it was to witness software running on the D30 showing successful convergences coming minutes or in many cases, seconds apart. The evaluation left the impression that the machine would solve anything, given the time to do so.

Meshing Testing

When I tested 138,607 mid-plane elements in 26 parts, the D30 meshed them in under two minutes. In another comparison, I did a nodal maximum results summary of 67,958 nodes; here, the standard workstation took 15 minutes, but the D30 took 7 minutes.

Simulation CFD

Autodesk Simulation CFD is the shining star in this review. Similar to Simulation Mechanical, CFD acted as if it had no limits.

I found that this software is usually quite stable, but in the past I have experienced some slow solutions on intense problems. In some cases, I had received reports that the analyses were unable to continue due to a lack of resources on standard workstations. I’m not stretching things when I say that CFD’s performance was quite impressive on the D30 (see figure 6).

Figure 6: Planar results and traces in Autodesk Simulation CFD

Not only was the graphics environment responsive and well behaved, but the solver was very nice on this machine. To prove the point, I ran the full cabin-comfort simulation performed for Autodesk University 2012, which was analyzed simultaneously on the cloud and on the D30. Results were kind of spooky fast, so much that I ran it twice just to be certain.

Showcase

Autodesk Showcase was no show stopper (see figure 7). It was, however, substantially better behaved and more capable on the D30 than on the reference workstations. The responsiveness was improved, which made manipulating solids better – a relief, considering the awkward controls offered by the software.

Figure 7: Ray traced rendering of the 'Full Cabin' in Autodesk Showcase

Rendering

Hardware renders were very fast. Ray tracing was handled very quickly, especially with <12 passes. Beyond that point, improvements in quality were not discernible between ever-increasingly longer renders.

Colors and patterns in ray-traced renders were improved as well, but not in relation to the overall machine capabilities. These observations are not the fault of the hardware, but in software limitations. In all cases however, the software completed the renderings without fault.

Other Interesting Results

To get a feel for how the D30 would handle being treated as a workstation, and not merely an analysis server, I decided to look at solving, rendering, and modeling. (If truth be told, I just couldn’t wait 179 bloody hours, and had to get the rest of my work done).

I ran simultaneous operations using the following software:

I ran these at the same time with no noticeable degradation by the software running any operation on the D30.

Figure 8: Every software package in this review loaded simultaneously, including 16 Inventor sessions, with no discernible delays or lags in performance

More observations:

Price

Base price is $1,319. The price of the reviewed workstation is $8,172.92.

Closing Thoughts

I used the Lenovo D30 extensively to develop and analyze the datasets for our Autodesk University 2012 Classes. I could not have prepared all the documents, analyses, and renderings in time without it.

As a solving machine, the D30 is amazing. Certain solvers ran faster than expected, and some were on par; nothing, however, seemed to slow down the machine or bog it down. As a multi-purpose high-capability modeling and solving machine, it’s over the top.

The only negative observation is that the D30 can put out a lot of heat. AnandTech.com states that the power consumption of the D30 is between 123 and 548 watts. When the unit is engaged in heavy computations, you’ll need some positive cooling for the surrounding area.

If I had to put my first experience with the D30 into a sentence, I would have to say the Lenovo ThinkStation D30 is like a bottomless well – if there is a limit, I was not able to find it.

See All Exclusive Articles | Back to Top

Additional Information

Lenovo ThinkStation D30 - corporate product page

Back to Top

About the Author

John Evans has 30 years experience in the aerospace design, engineering and fabrication, as well as 18 years with MEP and civil engineering. He is certified with AutoCAD Civil 3D and Inventor. More...

See All Exclusive Articles | Back to Top