How are you making the solver run fast?

Make sure that the statement PREFERENCES/NTHREADS=x is written to the adm file. If should be, but you never know.

The static solver is not parallelized, so during static there will not be that high CPU utilization. 

Check your memory consumption. I know that your models are huge and if the solver needs to start swapping memory, then you are in big trouble. Throughput to disk can also be a bottle neck, but in general it does not affect the simulation times too much. But I only uses solid state hard drives for my working disks these days. I have regular spinning disks for backups and software storage (F: and C: resp), but my working disks (D: and E:) are 2TB SSD. At least i imagine that it makes a difference.

Try to tune HMAX so that the solver does not fail too often. We have discussed this before. If you are running a constant velocity, a single HMAX would be sufficient. If you are running an acceleration or braking, you might want to change HMAX during the simulation. Or even DTOUT to cover the changing frequencies, for example set the DTOUT to be every 5 deg of the gearbox output shaft. I have a CONSUB prototype I can share if you are really interested in this.

HHT should be the fastest of the solvers for most models.

Hi Jesper,

Thanks for the reply.

The adm and msg files both show "nthreads" getting set to what I ask for it to be. 

My memory model is set to "Huge".  Or are you talking about watching the memory during the run in the task manager?

I have changed my ways after our previous conversations and have been following your HMAX and ERROR settings advice.

I have tried HHT in the past in small doses.  I could try again...

I am still wondering what I.T. security things or other stuff may have been added to our machines.  Many times, these things are just pushed through without considering the impact to the analysis folks...

Hi Jesper,

Thanks for the reply.

The adm and msg files both show "nthreads" getting set to what I ask for it to be. 

My memory model is set to "Huge".  Or are you talking about watching the memory during the run in the task manager?

I have changed my ways after our previous conversations and have been following your HMAX and ERROR settings advice.

I have tried HHT in the past in small doses.  I could try again...

I am still wondering what I.T. security things or other stuff may have been added to our machines.  Many times, these things are just pushed through without considering the impact to the analysis folks...

I was talking about physical memory on the computer. It is very seldom a problem these days, but your models are not mill of the run models either. 

Regarding IT security: When I run big DOE, I usually exclude virus checks from an area of my hard drive so that the virus check doesn't go bananas every time a block of data is written to a file. Check with your IT department if youacan do somethign like that.