I was wondering if there was some sort of computer, ram, or vid card that we could get so our idw's of our gaging machines would "regen" faster? We design and build gaging machines in which we have to show the customer's part. Usually a cylinder block or head. These parts are somewhat large 75-100+ meg. Put a couple of these in your assembly with conveyor systems and 2000+ model instances and you have a very large and slow idw.
We are using win7, 64bit, 12gig of ram (see full spec's below). These idw's take sometimes 5+ mins to fully regen.
any suggestions on new systems or additions?
We really don't have time for that. We have a certain amount of time to get a job done and in the shop so We are constantly under the gun to get them done. They (autodesk) claims that you are still able to work while the idw views regen separately. I don't know about any of you but We can't do crap while they are. If you try to zoom or pan it is always too little or too much and it is so "laggy" it is better to wait until all views are done. Very frustrating. We have been using inventor since ver.5 and this has always been a problem.
How would running simplification or shrinkwrap on the part be more time consuming then waiting five minutes everytime you need to recompute views?
because our business is to usually check location of holes and features in these part and most time have to get right up to these features, having the exact part model in them is key so interferances can be detected. it is usually the filets and radii on the customer parts that make them so hard to work with.
On these drawing with 2k+ instances is it neccessary to have every feature of the parts shown? You should have two (or more) versions of the file. One with full detail that you can use for verifying hole placement and such and one for these large layouts that just show a very simple representation of the part. A cylinder head is a very complex piece, if I tried to make an idw with 2k cylinder heads in it my computer would explode in a ball of flame.
No most times there are only two or three instances of the part at different positions in the machine and the rest of the instances are models from the machine.
It would probably still help to simplify the part. You could do a simple experiment to figure out what's taxing your system. Try suppressing the customer's parts in your machine assemby. If your idw is still dog slow then it's your assembly, if it speeds up considerably then it's the customer part, if it only speeds up some then it's both.
Are you using LOD's?
We are sorry to hear about your performance problems. I would be interested to know if
a. your very large parts are shrinkwrap or derived parts
b. are these imported parts from any translation?
b. if you have any threads in them AND are computing the views with threads enabled.
c. How many views are computing being computed in parallel.
d. If you fire up task manager and see the CPU usage, are all cores pegged?
e. If you create just a simple assembly with a couple of these very large parts, does it still take a long time to compute?
We do know of performance issues with shrinkwrap parts and threaded parts because of the massive amounts of data which requires processing - you say you have 50 - 100 MB parts. Another reason could be because of imported parts and any geometry inaccuracies introduced during the import process.
When views are computing in the background, the background processes compete for CPU time with the main Inventor process. Even though they run at a lower priority than the main process, windows still has to allocate resources and compute them. Having more RAM and more cores always helps.
If you can share your parts and a small assembly containing these parts, I can certainly take a look at them.
I can setup an ftp site for you to upload or you can try to zip it and send it to me. You can reach me at: sundarsATautodeskDOTcom
If nothing works, you can turn OFF "Enable background computes" found in the drawing tab of Inventor Application options - but doing this means, you will have to wait until all views are computed before resuming work.