GPU DETAIL SHEET
NVIDIA GeForce RTX 5070 - VolumeShader_BM Result
This page holds the gpu volumeshader_bm result breakdown for NVIDIA GeForce RTX 5070. Review every preset that users tested plus the API they selected so you can compare each gpu test result before planning your own run.
Last update: April 16, 2026
Stats refresh each weekPreset & API table
Average FPS is the straight mean of every accepted submission for this combination. Test duration is reported in seconds to help identify short burn tests versus multi-minute soaks.
| GPU Model | Preset | Avg FPS | Test Duration (s) | API Type |
|---|---|---|---|---|
| NVIDIA GeForce RTX 5070 | Ultra Low | 500.00 | 41.96 | webgl |
| NVIDIA GeForce RTX 5070 | Low | 201.96 | 49.33 | webgl |
| NVIDIA GeForce RTX 5070 | Ultra Low | 200.15 | 39.26 | webgpu |
| NVIDIA GeForce RTX 5070 | Low | 184.76 | 80.02 | webgpu |
| NVIDIA GeForce RTX 5070 | Balanced | 170.39 | 51.73 | webgpu |
| NVIDIA GeForce RTX 5070 | Reference(cznull github) | 166.58 | 40.95 | webgpu |
| NVIDIA GeForce RTX 5070 | Reference(cznull github) | 164.81 | 59.47 | webgl |
| NVIDIA GeForce RTX 5070 | High | 64.74 | 50.23 | webgpu |
| NVIDIA GeForce RTX 5070 | Apex | 47.73 | 33.92 | webgl |
| NVIDIA GeForce RTX 5070 | Very High | 39.80 | 49.48 | webgpu |
| NVIDIA GeForce RTX 5070 | Apex | 29.46 | 101.91 | webgpu |
| NVIDIA GeForce RTX 5070 | Extreme | 23.10 | 49.54 | webgpu |
Runs shown above are sourced from opt-in user submissions. Outliers caused by high system load or thermal throttling remain visible so you can spot variability.
Important context
Every figure above summarizes leaderboard submissions that community members voluntarily shared with the VolumeShader_BM project. We aggregate those opt-in runs by preset and API so you can see how other testers configured the exact workload shown on this page.
Treat the data as directional guidance, not a certification. Browsers share GPU time with creative apps, games, encoders, and background services, so FPS will shift with drivers, thermals, or multitasking load. Run controlled benchmarks on your own hardware whenever you need to validate a purchase or deployment plan.