I already posted on a 3d dedicated forum and with a help of a user (ran a scene on his workstation) I managed to reduce the probable culprit to hardware settings. Low poly, optimized scenes are going hundreds of times too slow with the CPU barely doing any job (it is not mistakenly setup for GPU) while the messy, 50mil poly scenes are going as planned with CPU usage on the limit. It's highly unlikely it is software (checked two different renders, the only improbability would be some general Max settings but I ran through them) and so it seems like the undemanding hole which doesn't seem to use CPU power at all is caused by some hardware setup. The hardware is a 12 core Xeon X5650 on a Dell motherboard/workstation with 24GB RAM. Ideas?
Thanks in advance.
It does meet the requirements, it is more than enough, although not incredibly fast. And it is not a performance limit, I'm not that dense. The performance is fine when the scene is overloaded and the CPU needs to rev up to be able to do the render cohesively. The problems start when it shouldn't, that is when I optimize the scene to run faster, use up less space etc. Then the CPU usage is in the low 30% for each core and temperature rises by 5-10C and lasts even 24hrs while for the messy, overloaded scene it is 95-100%, 20-25C and within 1h30min respectively. I really do not know what am I missing, like some 'hardware acceleration for low requirements' option-thing is left unchecked.
I really didn't think about it because I already assumed the possibility of a software cause is highly unlikely. Maybe the OS or maybe hardware. And don't overestimate the amount of support money can buy you. I think there is a threshold beyond which any customer related activity is basically sheep shearing