3D Graphics Performance Comparison – Part 16 (Conclusion)

<< Part 15

I think its time I brought this series to a close. I say this because I have given up waiting for certain CAD vendors to send me their optimal graphics settings. For the record, ten days ago I sent out emails to all the CAD vendors I know asking them to send me their optimum graphics settings. I even sent them the STEP file of the engine so that they could play around with it. They already knew my hardware specs. Some of them started making excuses not to send me their settings. I got the message and didn’t bother them further. Some others had the decency to ask me not to feature their product in this comparison. The rest flat out refused to reply to my email. When I see things like this happen, I can’t help but wonder if,  instead of doing it in an open and transparent way, companies prefer to pay some “independent” analyst firm enough money to conduct a study and twist the results of that study to spit out a white paper that sings their praises and pisses all over their competitors. It just goes to show the kind of confidence they have in their own products.

I have heard people express the view that such head-to-head comparisons are pointless. I could not disagree more. In my initial comparison I compared CAD systems as they came out of the box using the default graphics settings. My point there was to show how certain CAD vendors had not set their default settings to be optimized for large models. I know of at least one CAD vendor that has changed its default graphics settings after I started this series. In a post titled “CPU Usage In Idle CAD Systems“, thanks to a tip from a reader, I showed how some CAD systems literally hog CPU cycles when idling. I know of at least one CAD vendor that has tweaked their graphics engine to reduce the load on the CPU. If you read part 11 featuring IRONCAD, you will realize that this series has been far from pointless. IronCAD actually went ahead and tweaked the graphics engine in IRONCAD 2011 Beta which has officially been released today. I just received an email from someone at Delcam stating that they are having someone from the PowerSHAPE development team look into the issues I raised about display quality.

Although I described some CAD systems as “Awesome” and “Freaking Awesome” in this series, the point of this series was not to have a pissing contest. Yes, CAD vendors can see exactly where they stand with respect to their competition. But this series has brought to light a much larger point for users. Which is, you don’t need to spend through your nose on high-end graphics hardware to get great graphics performance. Baring Acrobat (which is really not a CAD system), the graphics performance of all CAD systems was pretty good. And this was on a pretty common workstation laptop having a pretty common graphics card.

The key to better graphics performance is the software, not the hardware. This amazing screen shot and video of Inventor 2011 is proof enough of that. When your software is lacking, you need to resort to shelling out big money to buy expensive hardware to make up for it. Yesterday I was having a conversation with someone who was suggesting that some of the CAD systems I was comparing were not designed to work with large models. True, that may be the case. But then how do you explain MoI’s graphics?

Actually, I can explain it. I wrote a series titled “MoI’s Sexy Graphics” wherein I explained in some detail the kind of effort Michael Gibson put into the graphics system of his CAD system. Just take a look at this image. It is not a render, but a screen shot of MoI’s application window as the user works with it. And MoI is written by just one person. Completely! I mean the user interface, the graphics engine, wiring the third party modeling kernel to the user interface and graphics engine, file I/O, the whole deal. Even the product documentation. Oh, and yes, he looks after product support as well. If one man can come up with something like MoI, something that costs $295, surely large CAD vendors that have teams of developers for every part of their CAD system which sells for thousands of dollars, can come up with much better. And what’s more, MoI gives the same exquisite graphics on graphics cards that are seven or eight years old.

Bottom line, it’s the software stupid, not the hardware.

  • Donceod

    is it possible to make a ranking of all tested CAD systems?

  • I suppose I could. But that ranking would be good only for this particular model on this particular hardware setup. Some CAD systems would behave differently for models much larger than this one (eg. a detailed model of a factory). The point of this series was merely to get a sense of now various CAD systems handle reasonably large data on an average hardware set up. More importantly to see of hardware mattered in the first place.

    The videos and screen shots are all there for anyone inspect and study. Anyone can rate them if he or she so wishes. This is an open and simply study. No graphs, charts and technical mumbo jumbo.

  • Deelipreader

    it was an interesting series, thank you deelip. we'll have to wait and see the effect of this series in future releases of cad programs.
    which cad vendor tweaked graphics for lower cup usage?

  • I don’t want to get into names and make it political. The important thing is that the CAD vendors took notice and acted, which is simply marvellous.

  • Committees make camels. Add to the mix, the committee directed by, out-of-touch management and marketers and you have a complete explanation as to why CAD vendors who should be able to do better, don't.
    Our industry screams for and would benefit greatly to find and develop for in the mold of Michael Gibson.
    An interesting series of articles Deelip.

  • Dnelson

    What I think would be interesting is to compare the files sizes for each of the models created, and find out who is bloating up your hard drive. This would also be important for sharing or sending models.