Testing GTAV bench across multiple CPU configurations
Testing GTAV bench across multiple CPU configurations
Evaluated setup: CPU model i9 7920X, motherboard Asus X299 TUF mark 2, BIOS 3201 Cooler. GPU 1080Ti, driver 460.89. RAM configuration: 8x 8GB modules in quad channel, 2 per channel. Screen resolution 1080p, Vsync disabled. No major concerns at this stage. Recently discovered you can add a command line file to GTAV for benchmarking. This feature lets you start tests with multiple iterations, helping assess stability without repeating the same run. I conducted only the fourth subtest (plane drive), repeated five times per setup. I’ll display the top three results. Early data seems promising since the GPU warms up slightly between attempts. The second test likely offers sufficient stability. I’m leaning toward using the latest three runs for comparison. My approach aims to capture peak performance, assuming stable conditions. The first run aids in file caching, reducing load times on subsequent tests. Before bench sessions, I set API to DX11, reset defaults, and disabled Vsync. Averaged FPS was around 142, with 95% frame time and 99% frame time (rounded). Stock system scored 145, while GTAV achieved 99%. I disabled hyper-threading in software and BIOS, then re-enabled it. HT was turned off in BIOS for testing. Results showed minor gains—around 2% when HT was off versus software settings. Restored original configurations afterward. I ran this to verify my focus on the best possible performance. This setup helps me compare results more accurately.
I usually turn off HT for 3D benchmarks unless there aren’t enough threads for physics runs. What about the screenshots?
You brought it up again, I should note the boost clocks might change. By default, my motherboard seems to cap at 140W. HT off could let the cores push a bit higher. I haven’t really dived into GPU benchmarking, so I’m not very comfortable with how they adapt to CPU demands. The results come as text files, which aren’t particularly engaging.