This morning I had intended to do some CapFrameX tests comparing HPET (High Precision Event Timer) vs. ISLC (Intelligent Standby List Cleaner.) In case you don’t know, HPET is built into Windows, and ISLC is 3rd Party software that is intended to replace HPET for better overall performance. There has been a lot of discussion on the web about whether ISLC is beneficial, so I decided to test it for myself.
All tests were done in the Bonanza G36 Improvement Project, flying over NYC. For each set of variables I flew the exact same course 7 times, discarding the best and worst result (based on 0.1% Low) and averaging the remaining 5 flights. Tests were done in 4K using DLSS 4 Quality, with mostly Ultra/High graphics settings. Traffic OFF. ‘Few Clouds’ weather preset. Photogrammetry ON.
No Frame Gen.
7950X3D, 3090 Ti 24GB, 2x32GB DDR-6000/CL30 (EXPO with tightened secondary timings.)
Process Lasso with the sim assigned to CCD0 (X3D cores) and everything else (except core Windows processes) on CCD1 (non-X3D cores.)
Hyperthreading ON
All other BIOS settings on AUTO - except that SVT (Virtualization) Disabled.
GPU mildly undervolted.
I ran the first set of tests, then rebooted and restarted the sim. The SU2 release update ran, so I didn’t have the chance to test HPET vs. ISLC in SU1. All other comparison tests were done running SU2, rebooting each time I changed HPET and ISLC settings.
It’s pretty clear to me that SU2 performance is significantly better.
I also see that HPET is slightly better than ISLC.
I will add frametime and interframe data later.
