Teaser Tuesday - Vectorworks Graphic Module for Sheet Layers - Vectorworks 2019

[PVA - Jim 2,381]

Today’s teaser item (or perhaps one and a half items?) is one of those underdogs that won’t likely get much commercial attention, but any regular user of Vectorworks will notice and appreciate it immediately.

 

In Vectorworks 2019, we have transitioned Sheet Layers over to the Vectorworks Graphics Module. This means that highly complex or dense viewports will now be drawn much faster than before, and that speed will scale up with your machine’s graphics capability. A lot could be said about this feature, but a direct video demonstration works best:

 

 

This enhancement is a direct response to the calls for improvement to speed and reliability of Vectorworks, and the most recent in the development we have seen in recent versions:

 

• 2015 - VGM OpenGL Graphics
• 2016 - Planar Objects and Wireframe added to VGM
• 2017 - VGM Top/Plan and Multithreaded final views
• 2018 - Multiple View Panes
• 2019 -  VGM Sheet Layers and On Demand Tessellation

 

On Demand Tessellation is a neat little upgrade that has one goal in mind: To return control to you as quickly as possible. I mention it here along with VGM Sheet Layers as it uses the same underlying technology to draw objects on-screen far more quickly than in Vectorworks 2018 and it’s other single threaded predecessors.

 

On Demand Tessellation is a complex feature, but it is relatively simple to explain: instead of waiting for all geometry in the view area to be drawn before returning control to the user, it starts drawing objects either one at a time or in groups, very rapidly, all the while allowing you to directly alter objects or the view, without waiting for the previous redraw to complete. This feature on its own will be the most obvious in dense files with either large number of objects, or complex sets of containers that slowed previous versions down. When coupled with VGM Sheet Layers, it benefits heavyweight architectural DWG imports and light plots specifically in our testing here.

 

We see a few more future performance improvements in the future of VGM, which Vectorworks engineering is already working towards in future updates. They intend to draw all selection/preselection highlighting and tool graphics with the VGM, and explore the possibility of splitting drawing and geometry calculation tasks to take full advantage of multi-threaded machines.

 

We will be hosting a live Facebook session to discuss each of the items we share in Teaser Tuesday posts. Here’s how it will work: the teaser will come out on Tuesday (Wednesday if you live on the front edge of the world!) and then you can post questions about it on Wednesday in the forum, through direct messages, or on our Facebook post. On Thursday at 2:00 p.m. ET, we will answer your questions live on the official Vectorworks Facebook page . If you can’t make it, don’t worry, we’ll also be transcribing the answers and posting to the forum too.

Feel free to ask any questions you like related to the feature that was just teased. This format is an experiment for us, so we also welcome any feedback or suggestions for the format, as well.

 

We plan to share Teaser Tuesday posts and host live Q&A sessions each week leading up to our launch of Vectorworks 2019!

 

 

[markdd 695]

This is great news. I’ve been hoping for this for ages. The performance with the light plot is dramatically different which is going to make working in sheet layers so much better from now on.

[cberg 195]

This is awesome!  Might there be other benefits associated with the switch of the sheet layers to the Vectorworks VGM? 😏  Specifically, would this enable viewports to be updated in a multithread fashion, or perhaps continuously?

Edited August 15, 2018 by cberg

[Patrick Fritsch 62]

This is what I like to see, making VW functions that already exist better in lieu of adding extra new shiny buttons to my design cockpit!

[Andy Broomell 1,020]

Awesome! Does this apply to all viewports or only ones that are in Top/Plan or Hidden Line? For example, what about a 3D view in wireframe? I've been working with some files with numerous complex meshes recently (usually figures, but also furniture), and the redraw rate before it's rendered is quite slow.

 

 

[zoomer 1,012]

Will this enhance 2D DLVPs too ?

[markdd 695]

2 minutes ago, zoomer said:

Will this enhance 2D DLVPs too ?

Great question.

[zoomer 1,012]

2 hours ago, Jim Wilson said:

They intend to draw all selection/preselection highlighting and tool graphics with the VGM, and explore the possibility of splitting drawing and geometry calculation tasks to take full advantage of multi-threaded machines.

 

This for me sounds even more exciting.

But this means VW 2020, right ?

Edited August 14, 2018 by zoomer

[cberg 195]

Just now, zoomer said:

 

This for me sounds even more exciting.

But this means VW 2020 right ?

 

That's what I was thinking... 

[PVA - Jim 2,381]

Keep the questions coming! Going to try something a little different this time: I'm gathering together all the questions we get from this post and other social media lines and see if we can get the lead engineer of this feature to answer the questions live in our stream on Thursday. If not, you still get the consolation prize of yours truly 😉

I'll post all the answers here right after the stream and then answer any subsequent questions after the fact the same as we were doing before.

[cberg 195]

Do you need to be on Facebook to see the live stream?

[PVA - Jim 2,381]

12 minutes ago, cberg said:

Do you need to be on Facebook to see the live stream?

Nope! You can view it without needing an account, you just can't ask a question in the live chat. Feel free to ask your question here and we will address it in the video. (No signing away your firstborn to Mark Z required 😉 ) I just posted the last one in the previous teaser thread, please let me know if it gives you any trouble viewing it.

[Mark Aceto 418]

First of all, thank you. Very excited to try this out on in real world tests to see if VW is significantly less CPU-constrained for those tasks.

 

The name of the game has been to find the fastest single-core clock speed available, so could you elaborate on multi-core improvements in 2019? For example, what tasks will take advantage of multiple cores that didn't before? And in what views that we couldn't before: DL, SL, 2D, 3D, DLVP, SLVP...

 

Would be really helpful to see a table that shows how different processes use/require CPU, GPU, RAM (particularly from year to year).

[Patrick Fritsch 62]

Wondering if this will help when the "hunting" of surfaces when using the Push/Pull tool?

 

[Kevin McAllister 1,146]

21 minutes ago, Patrick Fritsch said:

Wondering if this will help when the "hunting" of surfaces when using the Push/Pull tool?

 

My guess is no. I think that's more related to this -

 

4 hours ago, Jim Wilson said:

We see a few more future performance improvements in the future of VGM, which Vectorworks engineering is already working towards in future updates. They intend to draw all selection/preselection highlighting and tool graphics with the VGM, and explore the possibility of splitting drawing and geometry calculation tasks to take full advantage of multi-threaded machines.

 

I think it also means that the offset pre-selection highlighting bug that occurs with perspective/camera sheet layer viewports isn't fixed either 😕

 

I think this is a great step forward though. Does this acceleration extend to working in the annotations space of sheet layer viewports? I'm assuming it does, but annotations are sort of a weird hybrid space so that assumption may be incorrect.

 

@Jim Wilson have you considered using the notification bell system in VW2018 for posting these teasers? You might get more questions/viewers for the Q&A livestream that way....

 

Kevin

 

[PVA - Jim 2,381]

46 minutes ago, Kevin McAllister said:

@Jim Wilson have you considered using the notification bell system in VW2018 for posting these teasers? You might get more questions/viewers for the Q&A livestream that way....

I have! but we decided since the livestreaming is new we'd hold off for the first few. For instance, in this next one we plan to use a common game streaming utility to show the presenters, the demo video itself for anyone who hadnt seen it, and the Vectorworks UI live interchangeably. We'll keep ratcheting things up to see what works and what helps/hinders.

[PVA - Jim 2,381]

We debated having this as a separate teaser, but these two items really do go hand in hand. The below demonstrates On Demand Tessellation, which is a very large step in moving Vectorworks to being completely multithreaded. ODT basically splits up tasks to be handed off to the VGM into multiple CPU threads, meaning you don't just have to wait for that one single core to do everything before you can get working again. Its now going to draw objects using multiple cores.
 

This does NOT mean that the actual geometry calculations like add/subtract solids or duplication are multicore yet, those are still in the pipe, but a lot of the perceived slowness in previous versions, as you can see from the demo, were delays in simple redrawing of existing geometry.

We will be taking questions both about ODT and VGM Sheet Layers in tomorrows livestream, but again feel free to ask any additional questions here as well, both will be checked!

[Kevin McAllister 1,146]

15 minutes ago, Jim Wilson said:

We debated having this as a separate teaser, but these two items really do go hand in hand. The below demonstrates On Demand Tessellation, which is a very large step in moving Vectorworks to being completely multithreaded. ODT basically splits up tasks to be handed off to the VGM into multiple CPU threads, meaning you don't just have to wait for that one single core to do everything before you can get working again. Its now going to draw objects using multiple cores.

 

 

This looks awesome! Two questions come to mind -

 

1. Do Auto Hybrid objects take advantage of this? Exiting an Auto Hybrid is one of the slowest VW operations I know of.
2. Do operations previews (for example the preview when you're rotating an object) take advantage of this? Currently the previews often default back to a bounding box and I assumed it was because there wasn't enough redraw power.

Kevin

 

[Simon Allan 10]

Hi Jim,

 

In your first video, in the second half, you mention the lighting plot being fairly complex. Can you give some more details please on how many lighting instruments and other objects/viewports etc are in that particular drawing please?

 

Thanks.

[Chih-Pin 33]

23 hours ago, Mark Aceto said:

The name of the game has been to find the fastest single-core clock speed available, so could you elaborate on multi-core improvements in 2019? For example, what tasks will take advantage of multiple cores that didn't before? And in what views that we couldn't before: DL, SL, 2D, 3D, DLVP, SLVP

 

 

You can see more multi-core usages almost anywhere, including those listed above, especially transiting from one view to another, entering/exiting editing modes, turning on hidden layers, changing view render mode, turning on multi-view, and now switching between sheet layers. 

[PVA - Jim 2,381]

4 minutes ago, Simon Allan said:

Hi Jim,

 

In your first video, in the second half, you mention the lighting plot being fairly complex. Can you give some more details please on how many lighting instruments and other objects/viewports etc are in that particular drawing please?

 

Thanks.

 

The sheet layer itself has 6 viewports from multiple sources, but the main one is the plot itself as expected.

'

125 Lighting Devices

58 referenced hybrid symbols (many containing hanging positions or lighting devices in addition to the above)

58 2D only symbols

28 Groups (primarily with text and 2d geometry inside)

34 assorted polylines with something like 3800 vertices total (estimated)

Not the heaviest file I've ever encountered, but one where the delay was nearly all in the screen redraw, not related to any specific object or geometric slowness.

 

[markdd 695]

1 hour ago, Jim Wilson said:

125 Lighting Devices

58 referenced hybrid symbols (many containing hanging positions or lighting devices in addition to the above)

58 2D only symbols

28 Groups (primarily with text and 2d geometry inside)

34 assorted polylines with something like 3800 vertices total (estimated)

I’m interested how you know these figures. Do you have an in-house report function that users don’t have yet? 

 

It it would be great to have that kind of function so that, for example, when a drawing is becoming sluggish one can understand what might be the cause.

 

 

[AndrewG 14]

1 hour ago, Kevin McAllister said:

 

This looks awesome! Two questions come to mind -

 

1. Do Auto Hybrid objects take advantage of this? Exiting an Auto Hybrid is one of the slowest VW operations I know of.
2. Do operations previews (for example the preview when you're rotating an object) take advantage of this? Currently the previews often default back to a bounding box and I assumed it was because there wasn't enough redraw power.

Kevin

 

Hey @Kevin McAllister,

 

To answer your questions:

 

1. I do not believe Auto Hybrid objects do not take advantage of VGM ODT Threads. This is because a lot of the time is used by old drawing functions that are not VGM functions. We already have a report about this and we are considering on whether on fixing this in the future.
2. I haven't used the operations previews that you are speaking of, but if it doesn't use VGM or if generates a stale preview, it is probably not taking advantage of the ODT threads.

 

Everything else involving dynamic drawing (loading objects in wireframe, OpenGL, Hidden Line, and top plan design/sheet layers) should be using the new ODT threads.

[PVA - Jim 2,381]

18 minutes ago, markdd said:

I’m interested how you know these figures. Do you have an in-house report function that users don’t have yet? 

 

It it would be great to have that kind of function so that, for example, when a drawing is becoming sluggish one can understand what might be the cause.

Sorry to disappoint, but I just opened up the file and used Select Similar on a few key element types to get a general count.

Apologies, I know of no automated inventory, unless the engineers have been holding out on me?

[Kevin McAllister 1,146]

 

@AndrewG Thanks for the info!

 

16 minutes ago, AndrewG said:

1. .....We already have a report about this and we are considering on whether on fixing this in the future.

This made me laugh out loud. I'm assuming this means NV is undecided as to replacing the Auto Hybrid with something new or improving the current implementation. Complex Auto Hybrids aren't very useable at the moment.

 

16 minutes ago, AndrewG said:

2. I haven't used the operations previews that you are speaking of, but if it doesn't use VGM or if generates a stale preview, it is probably not taking advantage of the ODT threads.

 

I guess I wasn't clear enough in my description. By operations preview I was talking about the preview that VW generates when you use a tool like Mirror or Rotate. Here are some examples where VW gives me bounding box preview with a hybrid symbol in Top/Plan but gives me a complex preview in Top view.

 

Rotating -

 

 

Mirroring -

 

 

Kevin