Well, as I mentioned before, Beaker's been through a lot in his efforts to help us build our SOLIDWORKS World presentation "Modeling Methods and Environment Configuration for Optimal Performance". (It's a long name, I know) The efforts of Beaker have given us some great data and some hair loss.
First, some of the great data:
We were able to somewhat quantify the
effects of using external references, at least in our environment. What we did was put 30 different locations in reference documents in reference file locations of system options and turn on search file locations for external references. Now 30 locations may seem like a lot for some users, but I once saw a company that had 75 different locations in there. The difference in load time of the assembly was significant. It took ~3X longer to open the assembly. (Your results may vary)
Second, Hair loss:
I have also been pulling my hair out in an effort to get quality data comparing RAID arrays for modeling purposes. This isn't going well and as of this afternoon I actually lost a 1TB Hard Disk to the battle. After attempting to run a test comparing disk access speeds of a single disk versus a 2 disk RAID 0 (SOLIDWORKS not included). The single disk that had my OS on it would no longer boot. (Part of the reason I'm writing this article is that no one in my office wants to hear me complaining about it anymore.) I still have not gotten what I would consider valid data from RAID configuration tests and it's not from a lack of trying. Software RAID from the Windows OS, A third party RAID controller, and the built in RAID controller that comes with my T3400 (Beaker), have all generated suspect data.
If Beaker gives me anymore grief like this I might just pop a cap in his monitor!
The next step is exciting. As soon as I reload the #$^% OS, I'm going to run the benchmark with a single Solidstate hard drive. I can't wait to see how this helps. After that I'll make another run at RAID 0 but with the Solidstate drives.
I'll keep you all posted.