Long time reader, first time poster. I use creo 2.0 here at work and was running another FEA on a fixture structure, like normal, I monitor cpu and ram usage. I decided to watch the GPU usage and found that it sits idle the entire time I'm doing anything in creo. maybe a blip here and there when changing model display type, but am I wrong to think that there is some way to offload some of the computation to the gpu? While running simulate, my cpu is pegged wide open and appears to be throttling itself back whenever simulate has to read/write to the ssd temp files. While my new workstation is leaps and bounds faster and more powerful than my last workstation, I find it odd that I never really seem to make the K4000 graphics card earn its keep.
I have done a lot of searching and found various threads that touched on opengl issues, but it was usually in reference to an AMD card.
I've seen mention of such config options as:
Is there anything obvious I'm missing? I appreciate and feedback or insight i can get.
Intel Xeon E5-1620v2 @ 3.7 ghz
nvidia quadro k4000
windows 7 64 bit
Creo 2.0 M130
Okay, I was kind of wondering this because i've monitored gpu usage and stats while doing a little of everything and haven't seen the gpu usage go over 20% for more than 1/10th of a second and only that when switching a very large assembly view display modes.
I know that Wildfire 4.0(which i prefer to use at all possible) performance is based almost entirely on single thread processing power, but with creo 2.0, more specifically creo simulate, it can make use of multiple processors. I know other programs and such can make use of the GPU power for computations, i.e. hardware acceleration, directcompute, scrypt-mining, etc. And with this Nvidia card's 768 cuda cores, it would be interesting to see how (IF) it could be used to help compute the thousands of small calculations in a FEA simulation.
So is there any real point or advantage of tweaking settings in the config to specify anything specific with the GPU or graphics in general? or not really because the gpu has enough horsepower to handle anything i can throw at it in any configuration so far?
ENABLE* options are used in case, when user has problems with graphic cards. They switch Creo functionality between old one (Wildfire) and new one (Creo). Example: Wildfire and Creo use different shading software algorithms.
I guess that this option is obsolete. It was implement in times, when OPENGL was born.
Interesting, because another issue we have to deal with and one of the main reasons I despise creo compared to WF4, is that I can open the exact same job in WF4 as in Creo 2.0, and without a doubt, Creo 2.0 operates multitudes slower.
I.E. switching sheets on a drawing even when the large assembly drawing model is on say sheet 1 only, and the drawing models on sheets 25 and 26 are simple blocks. It takes the amount of time to change from sheet 25 to 26 that it would to sheet 1, but in WF4, sheets draw instantly by comparison, between 24 and 25, but only changing to sheet 1 would take a moment longer to think, which is understandable.
Regardless, WF4.0 will perform any given operation 10x faster than creo 2.0.
So is there a way to force creo 2.0 to use the shading algo as in WF4? (I know it's not that simple, I'm not a software programmer, just spitballing ideas)
So if i wanted to experiment with config options, i would need to enter the ENABLE strings as:
Or do I have it wrong? If i don't have the ! it shows its invalid. And/or do i need to put YES or NO after the string?
I have tried multiple things and couldn't get any distinct effect on performance or usage graphs.
Trying to reproduce something along the lines of the 'going from a simple sheet 25 to simple sheet 26 takes a long time', I was unable to demonstrate it being slow. I suspect there is something about your data that is necessary to see the problem (though I expect the problem is in the code, not in your actions). If you could have an SPR made to show the poor performance, we can investigate, and this could likely lead to better performance not just with this drawing, but others as well. You can mention that I requested it, and I would appreciate being told the SPR # so I can know it's this issue.
1.] I think that you know that exclamation mark at the beginning of config.pro line means -> this line is comment line
2.] you cannot use ENABLE_OPENGL* options in WF4, I guess they were implemented in Creo and their purpose is to resolve graphics problems, only
3.] unfortunatelly Creo 2.0 can be slower than WF4 on the same hardware when you work with the same data ... this is known feature (for example ProE Wildfire is slower than ProE 2001, too) ... probably if you recreate data on Creo 2.0, the response can be better ...
1. Yes I know that the exclamation mark comments out a line so that it is ignored. I didn't know if it entirely ignored commented lines or if they are still read by pro/e to set certain sub-options or something like i seem to believe i have dealt with in another program in the past.
2. My WF4 works awesome, i don't need to change anything there. What disappoints me is that Creo 2.0 operates fundamentally slower than WF4. (and this is a shared opinion of 14 designers in here, not just me). I will try to make a screen recorded video that explains when i get time.
But for instance, simply placing a note in a drawing in creo 2.0 can make you wait for the program for 5+ seconds(on a big job), now, this may seem petty, but when you know you can place a note virtually instantaneously in WF4, this is like pulling teeth.
3. Honestly if a job was designed in WF4 and later up-converted to creo 2.0 (per customer requirements) in my opinion it seems to perform better than a job that was started in Creo. However, the designs are never identical so I don't really have a benchmark to test this. But most jobs that end up in creo are started in creo, and their drawings perform just as bad.
I realize that it is very possible that our config is contributing to creo's poor performance, and that it's not necessarily a clear cut issue that creo is slower, but at times it can be aggravatingly slow, for even the simplest of tasks. And as of yet, I have not found any options that impact the performance, so I didn't know if offloading some of the processing to the video card could/would be possible and/or improve the performance for us.
2 notes .
1.) you can check how old are graphic card drivers installed on your machine and install current version, if the current one is too old
2.) in Creo 2.0 when you work with big complex drawing, then set specific filter option at bottom right corner of the window (General is the default) before starting certain action ... this way you can reduce system load
Been busy here at work so haven't had much time to experiment with much.
I guess I should start by saying I was getting OpenGL and OpenCL mixed up and backwards, I realize OpenGL is for rendering graphics and displaying models and such, but I was confusing and mixing OpenGL and OpenCL.
So to sort of skip back to square one, I know you said that pro/e has always been unable to make use of a GPU for computing processing power, is that including advances like OpenCL and Nvidia's CUDA? Nvidia boasts about CUDA working with Creo for rendering and making realistic displays, but we never need to render a model as such.
So I'm just kind of curious if photo-realistic rendering is all that a high performance graphics card are/will be useful for, since even on very large assemblies I can only ever get a small spike of GPU usage when changing display modes to wireframe, etc. And that since a large percentage of the computing required for day to day modelling in WF4 or Creo is single thread dependent, that there is just not a large enough need or even possibility that there would be much to gain for the few times that Pro/e can make use of multiple CPU cores and/or could use any GPU processing.
Again, I would like to thank you for your time and input, and that I realize many things in computer programs as far as computing usage are simply not possible, and that even some that are possible, are generally a lot easier said than done. As I said before, I am not a software engineer or programmer, there are many things that I don't fully understand. I just want to make sure I am squeezing every drop of power I can get out of my workstation because I am often in time-critical situations, and it is difficult to convey that for the company to spend $$ to put a large amount of ram in my computer will pay itself off when it takes 1/10th of the time to run an FEA because it can hold much more in memory.