Tag: #PyTorch
Kubernetes, DevOps and MLOps, written from things we ran and measured ourselves · 1 posts
Unlocking Tensor Cores Real-World Measurement 1: Matrix Multiplication 1.6x, Real-Time Ball Finding 0.9%
NVIDIA's consumer cards run FP16, BF16, and TF32 Tensor Core operations, which accumulate in FP32, at half speed. The third-party kernel module gpu-unlock lifts this limit at boot time via the GPU security processor's bo
2026-10-31 · 29 min read #GPU#텐서 코어#커널 모듈#프로파일링#PyTorch