Memory Matters

[dcEcho]

Just my experience with different RAM levels...

Recently I switched out my RAM from 16GB to 64 GB. Rendering view ports in sheet layers can take up a LOT of memory(specifically with high detail in open GL), so that's where I most appreciated the performance changes.

 

With 16 GB I could render most VPs on high, a few crashes here and there, though it could take a few minutes (not including full project views).

With 64, I almost never crash, except the full project views on high detail; even then only once in a while. Most renderings now take up only a little of the RAM and complete in seconds, but larger, full project views, can max out the memory space on my computer (when doing high detail render), but gets the job done after a bit.

The first few months I was using Vectorworks 2018, I had an older computer with only 8GB RAM, which was nightmarish. If it didn’t crash, it would take forever to render simple VPs.

 

I’ve learned that even with 64 GB, saving is essential, both before updating the VP, once layers and classes are set, and after its finished rendering. It's so basic its easy to forget when I get into it. And also its better to keep VPs on medium detail in general, since it takes less memory and most views don’t need such high detail anyway.

 

[JMR]

Good to know. To the memory shoppe it is when the next pc update is due.

 

One wonders if the code is quite right though, since the crashing.

 

We've had similar issues with viewports. Setting everything to absolutely minimum and turning off all possible classes helps with the speed. Still too slow though. Current machines are 16 and 32GB.

[michael john williams]

Good to have the feedback. I went with VW recommendations and have 32 GB on my new pc to avoid crashes but VW still crashes frequently even on the simplest tasks and not just updating viewports.

[PVA - Admin]

Memory usage can also be dramatically affected by how large your file is in conjunction with how many Undo operations you have set in Vectorworks Preferences. I believe he default is 20 operations, but many people increase this to 100 which can spike memory use significantly.

We are also tracking our RAM and VRAM usage footprints and have some upgrades planned that should reduce how much memory Vectorworks and the VGM need to perform the same amount of work.

[JMR]

I've nowadays set my undos to 20, previously I kept them at 50 most of the time. Didn't make any noticable difference, though.

 

How large is a too large file - when should one start thinking of splitting the file into separate parts? Ours are usually 200-400MB, typically consisting of a ~2000sqm2 building with about 100 sheets. Here on the forums I've seen people quote working with files over 1GB, which to me seems rather large.

 

 

 

 

[PVA - Admin]

No flat size rating that determines it, speed and memory use also have to do with what KINDS of geometry you have. Meshes and some other 3d polygonal geometry types are particularly important to keep an eye on, but theres no strict regimen to stick to.

[Gunther]

@dcEchoHave you since moved to Vectorworks 2019 or is 2018 your current version of Vectorworks?  You're welcome to send me the last six digits of your serial number and I'll review the crashes in our database.  

 

@michael john williamsI'm also interested in the crashes you've been experiencing.  Anything you can duplicate at will, I'm quite interested also.  Again, send me the last six digits of your serial number and I'll take a look.   That's providing, "Error reporting" is turned on in your Vectorworks Preferences on the Session tab?

 

Thanks!