That post by i_build_nanosuits made me think - and thank you for it too.
acad.exe uses 20 threads when idle - 850MB of RAM when it has opened a 44MB 3D model. While it was opening it (~15 seconds) it maxed out one of the two cores in my 2020M CPU at 100% (50% overall).
With the model open there are now 12-15 threads to the process - panning the whole 3D model in wireframe maxed out the CPU at 50%.
Interestingly - Windows 8.1 shows two logical processors - both with identical usage??? I was expecting 100% for one and a nominal percentage for the rest of the O/S.
I regened the model to conceptual (a full shaded image) and orbited it - still both logical cores are the same as one another. Still maxed at 60% overall. I'm assuming that acad accounted for 50% and the balance was the shitty GPU/OS overhead.
With a hidden view - same story. Around 15 threads and 50% CPU...
This information is quite valuable. The major obstacle to smooth workflow with these size models is a) the time taken to regenerate the views (from wireframe to shaded for example). B) framerate when panning /orbiting around the models. In either of these situations, the 2020M only seems to max out 1 core.
Now - my questions based on this are:
if the bottleneck is the CPU and the single core issue am I best off going for the CPU with the fastest single thread score (the 4690K)? Possibly with overclocking? Would the iGPU help?
if the bottleneck is the lack of hyperthreading, am I better off going with the 1230V3 (as it supports it)? But - no iGPU?
It would certainly seem that the passmark scores for multi-threads are irrelevant if the program does not use them (hence the poor scores for the FX AMDs in other reviews for CAD. Do you think this is correct?
Much of the bottleneck must be the shitty GPU on this el-cheapo laptop. As I stated earlier I'm shooting for a GTX 760. With all this in mind - 4690K or 1230V3? The i7 is out of budget range...
Cheers - Dave