Hello,
I am mostly running Zemax OS on my laptop, based on Intel i7 4 x 2.9Ghz 16GB RAM (Win 10 Enterprise, 64-Bit), to speed up simulations, I recently started using workstation with AMD Ryzen Threadripper 3970X 32-core 3.7Ghz 128GB RAM (Win 10 Enterprise, 64-Bit). Pure NSC raytracing is about 8 times faster on the workstation then on my laptop. However, on the same NSC optimization tasks I am getting longer optimization times on the workstation.
For example, comparing with the results from the KA-01591 How to optimize non-sequential optical systems
Intel Quad Core CPU (2.90 GHz) and 16GB of RAM (as stated in KB Article)
Algorithm MF Value On-axis brightness (Cd) Time for optimization
DLS 6.69 238 4.4 min
OD 6.68 254 6.5 min
My laptop i7 4 x 2.9Ghz 16GB RAM
Algorithm MF Value On-axis brightness (Cd) Time for optimization
DLS 6.686 238 4.70 min
OD 6.686 248 7.43 min
AMD Threadripper 3970X 32-core 3.7Ghz 128GB RAM
Algorithm MF Value On-axis brightness (Cd) Time for optimization
DLS 6.77 217 16.43 min (not finished)
OD 6.70 253 16.35 min (not finished)
Screenshots
My laptop i7 4 x 2.9Ghz 16GB RAM
DLS OD
DLS OD
AMD Threadripper 3970X 32-core 3.7Ghz 128GB RAM
DLS OD
DLS OD
From the task manager it can be seen that the processor usage is very low during optimization on the workstation, not sure why. During pure NSC raytracing it looks better
NSC Raytracing
My laptop i7 4 x 2.9Ghz 16GB RAM AMD Threadripper 3970X 32-core 3.7Ghz 128GB RAM
But is still higher processor usage on my laptop than on workstation 100% vs. 87%
The KA-01591 article also shows universal plot with central pixel intensity vs. curvature of the mirror. Just refreshing this plot alone takes 1.7 min on the workstation and 20 sec on my laptop.
I am currently running Zemax OS 21.1.2 on the workstation and Zemax OS 20.3 on my laptop. Has someone experienced similar problems and/or knows what could be the cause for this sluggish performance on a much better HW setup ?
Best answer by Mark.Nicholson
View original