Arminius AI Benchmark
Pure Vulkan rendering pipeline. No CUDA. No ML frameworks.
—
Runs
—
GPUs
—
Top
Linux
6MB
Windows
4MB
SteamOS
6MB
macOS
soon
|
Model
4GB
Less Energy
Less VRAM
Faster
CUDA/PyTorch comparison
Same model, same prompt, same settings on every GPU. Vulkan rendering pipeline only.
CSV Export
#
↓
Tag
GPU
Score
Burst
Sust.
Wh/100K
¢/100K
$GPU
CUDA
Speedup
