Hi, Martin here from GarageFarm.NET.
I have tried your new benchmark utility and noticed an issue that the benchmark won't run on a multi socket setup.
We have nodes with E5-4669v4 CPUs which can be run in 3 configurations: 1CPU, 2CPU, 4CPU. I can run the benchmark without issues on 1CPU setup, but not on 2CPU or 4CPU.
I launched Blender Benchmark from command line and here's the error log:
location: <unknown location>:-1
Traceback (most recent call last):
File "P:\blender\blender-benchmark-1.0beta-windows64\2.80\scripts\benchmark\benchmark\space\__init__.py", line 602, in draw
self.draw_welcome(context)
File "P:\blender\blender-benchmark-1.0beta-windows64\2.80\scripts\benchmark\benchmark\space\__init__.py", line 536, in draw_welcome
compute_device_list_get(None, context)
File "P:\blender\blender-benchmark-1.0beta-windows64\2.80\scripts\benchmark\benchmark\space\__init__.py", line 937, in compute_device_list_get
global_cached_system_info = system_info_get(ctx)
File "P:\blender\blender-benchmark-1.0beta-windows64\2.80\scripts\benchmark\benchmark\space\__init__.py", line 413, in system_info_get
info = system_info.gatherSystemInfo(ctx)
File "P:\blender\blender-benchmark-1.0beta-windows64\2.80\scripts\benchmark\benchmark\foundation\system_info.py", line 111, in gatherSystemInfo
system_info["num_cpu_cores"] = int(cpu_topology['num_physical_cores'])
ValueError: invalid literal for int() with base 10: '22\r\r\n\r\r\n\r\r\nNumberOfCores'It is a problem with reading the number of cores. I was able to launch the benchmark after editing file system_info.py and setting fixed number of cores in that line (22 in our case).