Gilbert Osmond

That's incredibly slow! Create a new admin account on your machine: '>http://kbase.vectorworks.net/questions/805/Creating+a+New+User+Account+%28Mac+OS+X%29 Then once loaded in, time it twice, the first time you launch it then shut it down and time the second time as well in the same manner you did for your previous test, please. Hmm, a bit embarrassed here -- did as you suggested, in the clean user account I did have to re-enter my serial number & re-activate. Once that was done, on the 5th consecutive launch I am getting: Time to splash screen: 16 seconds Time to Doc open/ready to work: 19 seconds. What do I need to examine in my main user folder(s)?

I just did some tests on this. This is on an clean install that has now been launched ~15 times total, working out any "first-launch" setup issues, and getting the OS caching to an optimal state. From dock click to splash screen: 18 seconds. Splash screen to "Untitled 1" document open & ready to work: 35 seconds. Grand total startup time, 52-56 seconds (range.) This is with the default Architect workspace selected, and with an empty Resource Browser. The rest of the system operates very fast indeed -- new Samsung EVO 850 SSD boot device (500GB, only 20% used,) which is on the internal SATA-III bus of the Mac Mini. VW 2016 and all associated files & content are also located on this SSD. No other storage devices connected or mounted. Safari launches in <1 second, Mail (with 40,000 messsages and 300+ subfolders) in 3 to 4 seconds, LibreOffice v5 (a pretty cantankerous mass of code) in 3 to 4 seconds. Again, all of these are "cached/optimized" after several launch/quit cycles. Let me know if I can do other testing or submit logs on this. The hardware & OS config in my profile signature is up to date.

I also notice that VW 2016 is slow to launch in 10.11.2. Particularly so because it is *slower* launching a clean install of VW 2016 SP2 on a clean install of 10.11.2 on a brand new, mostly empty Samsung EVO 850 SSD located on the internal SATA-III bus --- than launching a year-old install of VW 2016 on my mucked-up, well-used Yosemite 10.10 setup on a much smaller, older, almost 95% full SSD.

I doubt I'd ever have become a VW user (and lately, aficionado) if there wasn't an a la carte option. I don't like companies that offer software ONLY on a subscription basis. That said, as a now-committed VW user [to the program, not an insane asylum.. ] I am a VSS subscriber. I do consider the cost carefully each year, and I will say that last year's early-ish and in-depth previews of VW2016 features were quite important in convincing me to remain with VSS. I hope April 2016 brings similar previews of VW 2017 features, and that that pattern holds through future years.

Just chiming in here -- if your budget is limited and/or you just prefer extracting maximum value from your older hardware, if you have any Mac with a Thunderbolt port (Mini, iMac, any laptop, even a Mac Pro for that matter,) there are fairly well-tested eGPU (external GPU) solutions now available. The best of them are custom-built and do require some hardware hacking. I can say that I am completely satisfied with the 3D & OpenGL performance of the Nvidia GTX970 card I have running with/on my 2012 Mac Mini. See the link in my signature below if you want more details.

In general, moving to a back-end that supports the full capabilities of modern browsers. Especially, for example, not having to go to separate "confirmation" web page for every action, before being taken back to the previous page. I agree ++ with the remark that long-term accessibility should be prioritized. Of course good content has to be curated to remain good. The StackExchange model (see http://www.stackexchange.com if not already familiar) of vetting content and questions seems to work very well. And SE also vets users as well. Other large org's like Apple/Filemaker seem to have official discussion boards which are semi-moderated by well-respected and experienced users who are not employees. Maybe use a system that allows multiple levels of moderators so that experienced users can help with the work of funnelling & sorting posts by quality and importance. I would like to be able to set a default posts-displayed time horizon of more than 3 months. I realize this default may have been chosen in an earlier era when server & network bandwidth were more precious. It is a drag to have to continually choose "Show posts 1 up to 1 year ago..." or longer. I look forward to whatever the new forums solution turns out to be. I am confident it is going to be an improvement over what now exists.

Watched all the videos, looks beautiful and I will buy a copy soon.

Marissa, on the general subject of monetizing Marionette objects and nodes -- when will encryption / runtime-only ability be added to The Marionette environment? I.e. so that one can create and sell M. content the way Vectorscript plugins have long been sold?

I really appreciate your frankness & confirmation on this JimW. Having this info helps me avoid fruitlessly throwing a lot of money at hardware in an attempt to make VW faster -- i.e. in "core" computation it's not going to make a revolutionary difference using a Mac Pro w/ multiple Xeons vs. a core i7 2.6Ghz. Only an incremental difference not worth ~$3-$4K. Of course VW remains fantastic in so many other ways, I'm not complaining, just hoping that in the not-too-distant VW gets full multi-threading/multi-CPU awareness in the "Core" functionality. As far as marketing & the annual product cycle go, of course this is not as sexy a feature as, for example, ENERGOS, point-cloud import, curtain walls, and much else -- but for the existing user base of VW it would be a *major* boon to take full advantage of existing hardware.

+++1 also, very much agreed.

Leica now sells the Disto S910, which is basically a miniature total-station with alt/azimuth + distance measuring. It's what I had hoped I could turn my D810 into; I wish I had waited 6 months or a year for the release of the S910 but you can never tell with product releases, so. However -- on careful consideration, actually a tape measure is still a lot faster than a Disto for accessible areas & dimensions under 15 feet or so. For example I would not want to use a Disto S910 to shoot points to capture a) a house wall b) 2 different windows and 1 door in that wall, c) the eave height & foundation height for that wall d) electrical outlets & hose bibs in that wall, and e) perhaps a small set of wood or concrete stairs on that wall. That would be a lot of complex pointing & shooting along with notes to link the points to actual features. I think much faster to just drag a tape measure.

Let's say I have a layer or a class of objects with very high polygon counts, for example high-quality 3D plant & tree models: I would like to be able to set the on-screen rendering quality for objects by class and/or by layer so that ONLY those objects are rendered at a much lower quality level (or even just outline-box place holders), while continuing to display the rest of the drawing at whatever default OpenGL and/or Renderworks quality levels I've specified. Basically, copying the "Sketch: Document default" function from the OIP into both Classes and Layers settings.

Why is it that Hardscape objects with 3D attributes (such as Pad or Texture Bed) can apply only to the PROPOSED site model? I want to be able to modify the *existing* site with Hardscape objects. I know I can just create separate site modifiers & manage the existing site that way but it seems like extra work. Why not have the same "Apply to: Existing / Proposed" drop-down for Hardscape objects 3D-display as there is for Site modifiers? Is there an easy way to work around this? Thanks.

I still use a tape measure for about half of my site measurements, a Leica Disto D810 for the other half. I looked at Magic Plan for iPad but it wasn't a good fit for me. The Disto D810 does at a lot of value for long distances, inaccessible objects, plan-distance (i.e. horizontal distance with slope already subtracted out,) object heights, etc. In the long run I am *certain* that 3D point clouding, either with laser scanners or with Project-Tango-type camera-based solutions, will replace most or all hand measuring, even down into residential-level architecture & landscape design. It's just a question of time until the equipment cost drops. With laser point scanners most of the cost is in the extremely high precision rotating-part machinery, both the laser mirror(s) and even more so the survey-station 3-axis mount. Perhaps the cost will come down if they can figure out how to move the scanning process into the solid-state realm, i.e. with MEMS mirrors, something along the lines of how they've made DLP projectors better with fewer moving parts. More likely however is Project-Tango style 3D capturing based on multiple cameras & parallax, maybe with a simple laser distance meter built in to augment or cross-check measurements at key points.

I've read through that (previously) in detail. I think my main wish is for wire-frame to become multi-core aware. As follows: When I am working OpenGL mode, pan/tilt/rotate axes all flow very smoothly with my GTX970 graphics card assisting them, even at highest quality settings. Where it still lags is in *zooming* in/out, specifically and only. Here's what I think the problem may be: the GPU is handling any pan/tilt/rotate movements alone because it has the full model loaded into it already. But when I zoom in or out of the model, the wireframe has to be re-calculated for the shift in perspective. Specifically on *zooming alone,* I watch Activity Monitor and see the Mac CPU bounce up to 100% (out of 400% possible, i.e. 4 available cores). If wireframe became multi-core-capable I bet the hiccups & pauses on zoom in/ out in OpenGL mode would disappear. I should note also that this zoom lag is independent of OpenGL quality settings. High or low, anti alias or not, etc. don't make a difference. That confirms for me that the lag is in the CPU as it deals with recalculating the wireframe geometry when zooming.

If you need portability AND you want high-end GPU performance you can do so with an external GPU setup. As you can see over at the TechInferno forums, perhaps the majority of eGPU users are laptop owners. Many of the setups are physically "klugey" with wires going everywhere but there are some with high marks for fit, finish, and portability. (You have to carry around an extra box the size of a large toaster, but with a good carry-case it could travel without too much trouble.

More GPU support would be fantastic. I understand that there are programming & structural issues that may or may not be possible to overcome, i.e. GPUs are not general-purpose compute units. But aside from the GPU question -- I've spent the last day or so carefully observing Vectorworks to learn exactly where it is CPU-bound, and I am pretty disappointed to regulary (*usually* it seems) see VW using only a single CPU core (out of 4 available) when doing basic operations & geometry updates / re-renders. I gather this by carefully watching Activity Monitor: I almost never see VW take more than 100% of CPU, [which means 1/4th of the available 400% figure.] Am I missing something or have I misconfigured something? I've checked all my preferences carefully. Having VW regularly use all (or a configurable number of) CPU cores would yield instant 200-800% speed increases in many operations I would think. If there is a better thread to discuss this in please point me to it, thank you.

Follow-up, here are Cinebench R15 results with the Nvidia GeForce GTX970 external-GPU solution I set up for my Late-2012 Mac Min. (Note, the CPU scores are a bit low because I was running many apps in the background, did not attempt to optimize.) So far the card runs flawlessly in all parts of OS X, including Vectorworks. I hope it's possible to one day move more of the compute-intensive tasks, e.g. hidden-line, polygon, and CineRender (Renderworks) modes onto GPUs.

Updating this thread w/ useful info: I am now running an NVIDIA GeForce GTX 970, in an external PCIe <--> Thunderbolt2 box, connected to the Mac Mini's single Thunderbolt port. Setup is running natively in OS X 10.10.5, no Bootcamp / Windows. (But reports indicate it can also work in those modes, w/ some caveats.) This solution works with most *any* Mac with a Thunderbolt port. 300% OpenGL performance increase, makes a big difference for using Vectorworks in Wireframe & OpenGL render modes. (But not other modes -- those are the only 2 that benefit heavily from GPU performance as of this writing. Polygon, hidden-line, and of course Renderworks are still heavily CPU-bound.) In VW2016 the amount of detail, esp. including textures, anti-aliasing, edges, and shadows that the OpenGL mode can display is impressive, and since OpenGL is sped-up by the GPU there is a big difference when handling complex models. You can read about the details of my install at the TechInferno forums, where I gathered info before buying my own equipment. This cost about $670 and around 10-15 hours of research & setup time. You could get the same setup, pre-built, for about $1000 from http://www.bizon-tech.com. Caveat emptor, BizonTech is a small / unknown company, basically all they are doing is re-packaging the same parts I listed in my post above and making a few cosmetic improvments.

I had been running a late-2012 Mac Mini with Intel HD4000 graphics on-board -- slower than the Iris graphics in the MacBook you spec'ced. Finally got fed up, did some research and figured out how to get a real external video card to work with the Mac Mini. I am now running an NVIDIA GeForce GTX 970, in an external PCIe <--> Thunderbolt2 box, connected to the Mac Mini's single Thunderbolt port. 300% OpenGL performance increase, makes a big difference for using Vectorworks in Wireframe & OpenGL render modes. (But not other modes -- those are the only 2 that benefit heavily from GPU performance as of this writing. Polygon, hidden-line, and of course Renderworks are still heavily CPU-bound.) You can read about the details of my install at the TechInferno forums, where I gathered info before buying my own equipment. This cost about $670 and around 10-15 hours of research & setup time. You could get the same setup, pre-built, for about $1000 from http://www.bizon-tech.com. Caveat emptor, BizonTech is a small / unknown company, basically all they are doing is re-packaging the same parts I listed in my post above and making a few cosmetic improvments.

I'm adding a follow-up here: I was having a similar problem where, despite setting all classes and all layers to Visible, and even copy-pasting just the Site Model to a brand-new blank document, I STILL could not see the Site Model Source Data (points, e.g. 3d loci). Solution: Go to Vectorworks preferences, then to the Display pane, and be sure "Display 3D loci" is set to "Always." (I had mine set to "none" previously, so I could never see any of my 3D loci.)

Xfrog.com Although, at the moment it appears they do not offer a native VW format. You can import .3DS or .OBJ files on a case-by-case basis. Importing is fine if you have 10 or 20 plants to import, not so good if you need 300 or 3000 species.

I am considering buying a 3000+ plant library (from Xfrog.com). They do not offer a native Vectorworks format. Their .3DS/.OBJ files do convert nicely to VW 2016 but only with laborious single-file manual importing. Not going to work for 3000 plants. Is there a way to script the import / conversion of many individual .3DS or .OBJ files from within VW? Or a third-party utility that would do this? Thanks.

Thank you, that fixes it.

Interesting, I am having the OP's (original poster) exact same issue in VW2015 SP4 (with renderworks.) I have created a viewport from 2 or 3 design layers, all at the same scale, nothing super fancy. When I go to set the Render options for the viewport, Orthogonal & Cavalier views work fine, but when I select "Perspective" I get the same tiny (~10 or 12 pixels square) box in the middle of an otherwise-blank Sheet Layer. When I change the perspective to "Orthogonal" and do Update the viewport instantly appears as it should. Any further thoughts on this? Should I attach a sample file?