I have two similar setups. Both computers are fairly decent, but Gazebo runs much slower on one of them even though it is in theory slightly better.

System 1:

• Ubuntu 16.04 with kernel 4.4.0-141-generic
• Intel(R) Core(TM) i7-6700K CPU @ 4.00GHz, cpufreq governor set to performance
• 32GiB DDR4 RAM @ 2133 MHz
• GeForce GTX 1080 (nvidia driver v430.64)
• Crucial CT525MX3

System 2:

• Ubuntu 16.04 with kernel 4.4.0-141-generic
• Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz, cpufreq governor set to performance
• 16GiB DDR4 RAM @ 3007 MHz
• GeForce GTX 1050 Ti (nvidia driver v396.45)
• SanDisk SDSSDA48

Even though system 1 should in theory be better, Gazebo simulation is much slower. Running empty.world with setting maxupdaterate to 10000.0 using /gazebo/setphysicsproperties, would result into real time factor ~4 at system 1 and ~10 at system 2.

I first thought it could be that the older generation of i7 is slower even at higher frequency, but all the benchmarks say it is faster (https://cpu.userbenchmark.com/Compare/Intel-Core-i7-7700-vs-Intel-Core-i7-6700K/3887vs3502). The same goes for gpu (https://gpu.userbenchmark.com/Compare/Nvidia-GTX-1080-vs-Nvidia-GTX-1050-Ti/3603vs3649) and disks (https://ssd.userbenchmark.com/Compare/SanDisk-SSD-Plus-240GB-vs-Crucial-MX300-275GB/m25612vs3642). Could it be RAM frequency?

What else could be a bottleneck? Which configuration I could look into?

EDIT: I benchmarked both computers using Phoronix Test Suite cpu tests and in contrary to the previous benchmark, the i7-6700K did perform worse (https://openbenchmarking.org/result/2001201-KH-D2086515975,2001208-KH-D5965159751). So this does not seem to be gazebo related problem. For future reference, what benchmarks are most relevant for gazebo?