Quantify Drawing Performance During Optimization Process

[Anonymous]

Greetings All;

I've been tasked with analyzing a series of very large AutoCAD files and offering suggestions for improving performance.  These files are master 2D layouts of a manufacturing environment and consist of multiple XREFs and blocks from many locations and vendors.  Understanding that there is a point where every system will bog down and become unusable, I've been working on ways increase the amount of useful data within each layout before this "bog down" point is reached. (purge, wblock, block vs. xref, etc)

Question.

Is there a practical way to simulate a set of actions that a user would typically perform like Pan, Zoom, Save, etc and then output an elapsed time for a particular drawing?  This would be similar to a conventional benchmark but instead of quantifying performance of a hardware configuration the score would be based on the contents of the drawing itself.  I would then be able to analyze some different optimization methods to find the best solutions.  I've experimented with both the Action Recorder and conventional script files which either fail after a certain level and/or give very different results on the same drawing.  I've been using the internal Time function by resetting at the beginning of the routine and then displaying at the end.

Any Thoughts?

[TheCADnoob]

You got to nail down your precision first. In other words you need to figure out whats affect the repeatability for you to be able to get consistent results.

If your times are all over the place you may need to increase your trials to see if you can get a normal distribution. 

I would suggest an isolated workstation (no connectivity) and minimum installs. In other words a computer that just has your os and autocad and is not connected to the network. 

After that might try running your scripts in coreconsole to skip any GUI impacts. 

[john.vellek]

Hi @Anonymous,

You might investigate the Cadalyst Benchmark and see if it can be adapted to this purpose.

Please select the Accept as Solution button if my post solves your issue or answers your question.

[john.vellek]

Hi @Anonymous,

I am checking back to see if my post or @TheCADnoob's helped you with your problem. Please mark a post or posts as accepted solutions if they resolved the issue or give me a bit more detail on this issue so we can continue to work towards getting this solved.

Please hit the Accept as Solution button if a post or posts solves your issue or answers your question.

[Anonymous]

John,

Thanks for the follow up.  Your comment about nailing down the precision is very valid.  Unfortunately, I'm working within a huge corporate environment and my system is locked down with the usual assortment security, AV and monitoring applications and local admin access is not an option.  I taking a look at the Catalyst Benchmark with some hope but with the precision so loosely controlled it may be no better than my scripts.

Backing up a bit to my original problem of AutoCAD bogging down.  I'm working with 2D only AutoCAD factory floor layouts from 3 different locations.  Many of these .Dwgs have been under constant revision for years and contain a seemingly random combinations of deeply nested blocks and Xrefs.  These blocks and Xrefs mostly originate from outside vendors who are contracted to install and maintain small sections of functionality within the main factory floor.  Most of these vendor supplied sections are 2D projections of the 3D assemblies in which they develop the machine components and contain far too much detail.  To complicate things further a majority of the folks at this company who have been working on these layouts have varying skill levels with AutoCAD so some of the basic rules have been missed.  For instance, I examined an Xref which was causing issues with a size of 308,512k.  This file contained over 3000 blocks, many of entire subsections from other area as well as 2200 layers.  After several purge commands the file size fell to 13,142k and the load time went from 35secs to 4secs.  I've been going through these files manually running through this optimization with some success.  Xref use is fairly common but there have been no guidelines in place for how these are named or stored.  Some are stored in the same folder as the main Dwg, some stored in sub-folders, and others scattered in various locations over a huge share drive that contains over 225,000 individual .Dwg files.  Finding Xrefs pointing to user desktops is not uncommon.  Finally, and the most disturbing, I've seen Xrefs with the exact same name stored in different locations with different file sizes and date stamps.  I don't think there is any process in place to archive obsolete data as some files date back to 2001.  Do any Autodesk admin tools exist which could help manage this situation.  I current have a Lenovo P50 laptop with 16G RAM which slows to a crawl when opening even the smaller files.

Thanks,

[TheCADnoob]

IMO This isn't a software issue, this is policy/procedure issue. You need good (or better as the case may be) standards.

If you have the authority to dictate how drawings are handled you can start the work of resolving it, but if you cant control the way these things are handled, it will only continue to get worse.

[john.vellek · Accepted solution]

Hi @Anonymous,

I suggest that you consider doing some maintenance work on the files you are currently "maintaining". It is great that they have blocks in them.  Modifying a block definition so that it is simpler and does not contain extra geometry or information, lets you update all the blocks with that definition quite easily.

I would also not worry too much about xref names being the same. While it is not a great practice, as long as each one is isolated then you should be fine.

If you want to share one of the large files, let me know and you can email to me.

Please select the Accept as Solution button if my post solves your issue or answers your question.

[Anonymous]

Thanks,

I totally get this is a policy/procedure issue. 

Hoping to leverage the software to help with damage control.

[Anonymous · Accepted solution]

Thanks John,

Cant send any files due to NDA issues.

Final simple questions;

I would like to have the ability to open a large drawing containing blocks and,

View the blocks in some detail.

Determine the block sizes and sort to find largest.

Determine if the block is nested, how deep, and with what.

The only way I've found to even view the blocks is by issuing an insert command and then canceling.

I've also used the Autodesk Seek Design Center (ctrl-2) which is helpful but doesn't offer info on sizes and nesting.

Thanks,

[john.vellek · Accepted solution]

Hi @Anonymous,

If you want to share your files with me privately you are covered by NDA and the files are private, confidential and destroyed after we work through the issue. You can email at john.vellek@autodesk.com or provide via a secure download location.

As far as the nested blocks, check out some of the block utilities in the Autodesk App Store. It appears there are some that are designed to help working with nested blocks.

Please select the Accept as Solution button if my post solves your issue or answers your question.

[TheCADnoob]

I have tackled messes like this before. Most recent I've done it with two large scale facilities and do it pretty regularly when i find an honey holes of contraband. 

I used a lot bat files, Excel, especially VBA in excel and macros and lisps in AutoCAD. I created a lisp to pull title block info before i found DATAEXTRACTION command but you might get some use out of DATAEXTRACTION if you are building a drawing index in the process of cleaning up the files. The big problem is that not all file can be treated the same. I set up macros on a tool palette to handle the things that were most frequently done to most files. 

Ive had macros that deleted all the paperspaces and inserts the company pagesetup and zoom extents and then purges reg apps, then purges everything else, then sets the view resolution then converts legacy polyline and hatches, then saves and then closes the file. The beauty of a macro is that it does this in one click. When individual files dont need to be reviewed, you can throw these steps all into coreconsole and it can roll pretty quickly through files. 

You can also check out the reference manager.