2020 Teaser Tuesday - VGM Level of Detail - Vectorworks 2020

[JuanP]

This #TeaserTuesday, we’re teasing an update to the Vectorworks Graphics Module(VGM). The VGM affects how Vectorworks generates and presents 2D and 3D geometry. In version 2019, loading tons of geometry could slow down your device, especially on larger projects. Usersoccasionallyreported slow frame rates when panning or zooming, and often loading files could take longer than anticipated. This update addresses these issues. The software will mimic the way the human eye operates — Vectorworks won’t load unnecessary graphics as objects move farther from the virtual camera. In effect, this will substantially reduce memory requirements to use Vectorworks.

 

 

[Wesley Burrows]

Has the 4 core limit been lifted from the VGM now?

[JRA-Vectorworks-CAD]

@JuanP This is a really huge step forward in terms of speed, snappiness and design workflow.Congratulations. I would alos love to know how multi-core processors affect performance, or is it mainly GPU driven?.

 

[neal-2002]

Looks positive.....19 was just too slow for me... looks like paying more cash for the next version may be worthwhile..

[Mark Aceto]

17 hours ago, Wesley Burrows said:

Has the 4 core limit been lifted from the VGM now?

 

17 hours ago, JRA-Vectorworks-CAD said:

@JuanP This is a really huge step forward in terms of speed, snappiness and design workflow.Congratulations. I would alos love to know how multi-core processors affect performance, or is it mainly GPU driven?.

 

 

AKA "the thing of which we do not speak."

 

While this performance boost is massively appreciated, I haven't heard anything about taking a load off the CPU and moving it to the GPU. Every release since 2015 has rewritten old code to take advantage of better GPU performance (and eGPU's), so this is the first year it's been quiet. This is even more important today because while we have 8-core laptops, their single core base clock speed is well below the VW recommendation, and turbo boost speeds are thermally throttled.

 

With only one more Teaser Tuesday left, I'm hopeful for:

• 4-core limit lifted to a minimum of 8 cores
• "This is is a maintenance release - we cleaned up a bunch of old code, and made everything faster and more stable"
• Massively improved worksheets
• Improved stair tool

I'm excited about all the little things that didn't make it into this round of teasers but these have been priority #1 in user feedback for years.

 

Edited August 28, 2019 by Mark Aceto

[Clint Alderman]

Looks good.

 

As the Teaser Tuesdays are demonstrated, can you list the device info: brand, cpu, gpu, ram, etc?

[Wesley Burrows]

1 minute ago, Mark Aceto said:

 

 

AKA "the thing of which we do not speak."

 

Also, while this performance boost is massively appreciated, I haven't heard anything about taking a load off the CPU and moving it to the GPU. Every release since 2015 has rewritten old code to take advantage of better GPU performance, so this is the first year it's been quiet. This is even more important today because while we have 8-core laptops, their single core base clock speed is well below the VW recommendation, and turbo boost speeds are thermal throttled.

 

With only one more Teaser Tuesday left, I'm hopeful for:

• 4-core limit lifted to a minimum of 8 cores
• "This is is a maintenance release - we cleaned up a bunch of old code, and made everything faster and more stable"
• Improved stair tool

I'm excited about all the little things that didn't make it into this round of teasers but those three have been priority #1 in user feedback for years.

 

Right!  I really want to read "All tools/modules/processes are 100% multi-threaded now,  we'll use every core you got."   

[Mark Aceto]

3 minutes ago, Wesley Burrows said:

 

Right!  I really want to read "All tools/modules/processes are 100% multi-threaded now,  we'll use every core you got."   

 

A big theme of last years teaser videos was, "we hope to one day eliminate the performance/quality sliders altogether because the app will run at full tilt, so there will no longer be a need for them."

 

I'm still holding onto that dream... 

[Matt Overton]

So does LOD improve viewport rendering and can we use it, tweak it to get depth-based line weights?

 

Edit: Does the attitude extend to making sure there are no modal dialogue boxes?

That is the number one way I'm watching progress bars and taking up vast parts of the screen so I can't do other things. 

Edited August 27, 2019 by Matt Overton
sorry could not resist.

[RussU]

2 hours ago, Clint Alderman said:

can you list the device info: brand, cpu, gpu, ram, etc?

I think this would be a really good benchmark.

we can all do amazing things with amazing hardware, but would love to see the spec used for demonstration

looking forward to seeing the new optimisations. 

[herbieherb]

10 hours ago, Wesley Burrows said:

"All tools/modules/processes are 100% multi-threaded now,  we'll use every core you got." 

Normally making a process multi-threaded means you have to rewrite almost every single line of code. Also there are lot's of things that you can't make multi-threaded. For example, in all calculations where a part depends on the result of the last part. Such calculations are very common in CAD. All tools/modules/processes will never be multi-threaded capable.

[elepp]

I keep my excitement on the backburner for now. While I don't doubt the sincererity of the enigineers here. This teaser is really a "teaser". It's not showing how long it takes for the models to load to begin with and how fast the switching happens from 2D-Plan to 3D. That's where we experience a lot of friction. Especially now that we also have data visualisation in the design layer, I am more than suspicious about the promises of improved speed.

[line-weight]

Bit odd the way this is explained in the video, as if it's resolving a fault in users rather than in the software.

 

If setting the quality to 'very high' means that VW draws stuff in the distance in more detail than is necessary then that doesn't mean that users are making a 'wrong' choice, it means that VW is doing it wrong.

 

Anyway, I wonder if this change will have any effect on the massive slow-down seen in models with imported, complex mesh objects. That includes in my experience problems selecting things (beach ball of death) when there's a complex object behind them (even if that object is not in direct line of sight view).

[line-weight]

12 hours ago, JRA-Vectorworks-CAD said:

@JuanP This is a really huge step forward in terms of speed, snappiness and design workflow.

 

 

Have you been given a test version to try out?

[HEengineering]

Certainly one of the things I was hoping for here. Tho we have to wait and see how it truly performs in certain workflows.  Wondering if this could improve site models as well? Also wondering if the issues with contours skipping over certain vertices has been resolved? I know many were experiencing this. 

 

For us we actually chose to stick with 2017 for site models as we felt it provided more accurate site models based off provided survey data. 

[Andrew Davies]

Sounds similar to 

 

 

but for design layers??

 

Would love to say I saw a big improvement with on demand tessillation - but I’m afraid I don’t. 

[Andrew Davies]

PS - oh I do miss @Jim W @PVA - Jim

Edited August 28, 2019 by Andrew Davies

[JuanP]

Took me a while to reply but most of these improvements are  a little out of my league (under the hood ) but to resume all my conversations with the lead engineers in these topics...here I go:

 

Multi CPU cores help VGM in 2 ways:  tessellation and level-of-detail management. VGM objects are tessellated simultaneously on CPU cores. In the best case this makes tessellating nearly N times faster where N is the number of cores being utilized (which is currently 3 @Wesley Burrows), but sometimes there is a bottleneck function that only allows a single thread of execution in at a time; other threads have to wait in line (like a 4 lane highway converging to a single lane). It is situations like that which need to be avoided. VGM uses one of the cores to continuously determine if objects are at their appropriate level-of-detail. VGM also uses the GPU’s cores to accelerate OpenGL/wireframe/pan-zooming in Top/Plan.

 

From our VP of Product Development: "Transitioning our application code to support multi-core capabilities is an ongoing challenge that we place at the top of our development priority list. We have been able to do this with more and more of our codebase. We see more and more opportunities ahead."

 

I hope I was able to answer most of your questions/concerns. If there is anything else you would like to know or that I missed let me know, I'll be happy to find an answer for you.

 

One more comment, during the creation and testing of the 2020 feature videos, most of the improvements that we experienced were in complex Spotlight files with multiple 3D lighting devices, and also curvy objects.  In some other cases, the performance was the same compared with version 2019 depending on the geometry.

[jnr]

Quote

 

 

On 8/27/2019 at 5:13 PM, Mark Aceto said:

With only one more Teaser Tuesday left, I'm hopeful for:

• 4-core limit lifted to a minimum of 8 cores
• "This is is a maintenance release - we cleaned up a bunch of old code, and made everything faster and more stable"
• Massively improved worksheets
• Improved stair tool

I'm excited about all the little things that didn't make it into this round of teasers but these have been priority #1 in user feedback for years.

Mark:

 

Can I flip your list over? The stair tool blows.

[Mark Aceto]

47 minutes ago, jnr said:

 

Mark:

 

Can I flip your list over? The stair tool blows.

 

That's why they're bullets instead of numbers ;) They're all priority #1.

[Mark Aceto]

19 hours ago, JuanP said:

Took me a while to reply but most of these improvements are  a little out of my league (under the hood ) but to resume all my conversations with the lead engineers in these topics...here I go:

 

Multi CPU cores help VGM in 2 ways:  tessellation and level-of-detail management. VGM objects are tessellated simultaneously on CPU cores. In the best case this makes tessellating nearly N times faster where N is the number of cores being utilized (which is currently 3 @Wesley Burrows), but sometimes there is a bottleneck function that only allows a single thread of execution in at a time; other threads have to wait in line (like a 4 lane highway converging to a single lane). It is situations like that which need to be avoided. VGM uses one of the cores to continuously determine if objects are at their appropriate level-of-detail. VGM also uses the GPU’s cores to accelerate OpenGL/wireframe/pan-zooming in Top/Plan.

 

From our VP of Product Development: "Transitioning our application code to support multi-core capabilities is an ongoing challenge that we place at the top of our development priority list. We have been able to do this with more and more of our codebase. We see more and more opportunities ahead."

 

I hope I was able to answer most of your questions/concerns. If there is anything else you would like to know or that I missed let me know, I'll be happy to find an answer for you.

 

One more comment, during the creation and testing of the 2020 feature videos, most of the improvements that we experienced were in complex Spotlight files with multiple 3D lighting devices, and also curvy objects.  In some other cases, the performance was the same compared with version 2019 depending on the geometry.

 

@JuanP thank you for the reply.

 

For clarification, there are 2 multicore issues being discussed here:

1. Making more of the code multi-threaded, which you just answered
2. Lifting the 4-core limit (on modeling not rendering), which is still a mystery

Please let us know where we stand with #2, including a roadmap of the version/year ahead.

 

Edited August 29, 2019 by Mark Aceto

[Kevin McAllister]

1 hour ago, Mark Aceto said:

 

@JuanP thank you for the reply.

 

For clarification, there are 2 multicore issues being discussed here:

1. Making more of the code multi-threaded, which you just answered
2. Lifting the 4-core limit (on modeling not rendering), which is still a mystery

Please let us know where we stand with #2, including a roadmap of the version/year ahead.

 

My guess is that these things being multithread and/or passed off to the VGM would make a huge difference when modelling in high geometry situations -

• geometry calculations (math) which is still single core
• various types of selection highlighting which I understand are being shifted to the VGM
• snapping / snap highlighting
• the whole process of moving an object (which involves various things above) including the move preview no longer becoming a bounding box only

I would be super excited to learn that any of these tasks were completed for VW2020. I've recently been working with a high geometry file and and using the Attribute Mapping tool with snapping turned on is brutally slow...

 

Kevin

 

[Wesley Burrows]

4 minutes ago, Kevin McAllister said:

 

My guess is that these things being multithread and/or passed off to the VGM would make a huge difference when modelling in high geometry situations -

• geometry calculations (math) which is still single core
• various types of selection highlighting which I understand are being shifted to the VGM
• snapping / snap highlighting
• the whole process of moving an object (which involves various things above) including the move preview no longer becoming a bounding box only

I would be super excited to learn that any of these tasks were completed for VW2020. I've recently been working with a high geometry file and and using the Attribute Mapping tool with snapping turned on is brutally slow...

 

Kevin

 

 

My experience mirrors Kevins. 

 

[jnr]

Not that it would matter at this juncture but,I seem to recall that the sorely missed message board zen master suggested that geometry calcs were going multi-core during a conversation about why regenerating a section viewport was so friggin' slow in 2018. fingers crossed.

[JuanP]

Busy day today!!! I've been checking with the different engineers involved with the VGM to help me get the answers to all your questions. I'm hoping that tomorrow I'll be able to provide you with additional information.