Performance Evaluation Confusion

So, I've got this model my company uses anytime we need a B&R computer, and I've got a google drive link cause it's a 32MB file: https://drive.google.com/file/d/1FcmTCSakJ3KR3-deU6UJPN5YBc-Rzsm7/view?usp=drive_web

Opening it takes several seconds, my computer noticeably lags when it's displayed, and loading/display times for any assembly it gets put into just *tank.* I got sick of this and built the attached model. Identical critical dimensions, mostly the same for some pretty details, opens instantly, and barely registers in an assembly. Out of curiosity, I ran Performance Evaluation on them both, results below:

Fast model rebuilds in .38 seconds, and fully half of that is the two boundary features that serve no purpose beyond looking nice. The full model reports rebuilding in .19. Half the god damn time. For a model 100x the file size, with infinitely more features and details, and more everything. There's no way the full model "performs" better than my fast model, and yet that's what performance evaluation is reporting. All I can guess is that performance eval is being thrown off by the imported features instead of actually building the thing, but if that's the case then performance eval is useless. Are they any settings I can change to actually evaluate the real performance of the two models? Geometry or feature evaluations or something?

SolidworksParts And Features