Benchmark GPU-accelerated Moon workloads against CPU baselines in CUDA environments.
npx skills add https://github.com/pilotspace/moon --skill gpu-bench
Please help me install this Agent Skill. Skill: gpu-bench Source: https://github.com/pilotspace/moon/tree/main/.claude/skills/gpu-bench Command: npx skills add https://github.com/pilotspace/moon --skill gpu-bench
GPU-accelerated benchmarking of Moon workloads to compare against CPU baselines, guiding optimization decisions.
Run gpu-bench to benchmark vector distances, batch throughput, and full GPU tests on a CUDA-enabled system.
High-intent search queries and answers about installing and using this skill.
You can benchmark GPU workloads against CPU baselines for Moon deployments by running gpu-bench to evaluate vector-distance, batch throughput, and full GPU benchmark scenarios with deterministic results.
Available GPU benchmark scenarios for vector-distance and batch throughput testing include measuring large-scale operations, memory-bound workloads, and multi-dimensional configurations across CUDA-enabled environments.
Yes, you need the CUDA toolkit to run GPU-accelerated benchmarks on Moon, as the process requires CUDA-enabled environments and GPU-enabled builds with the gpu-cuda feature.
Yes, you can measure batch throughput for memory-bound workloads using GPU benchmarking to evaluate large-scale operations and compare performance against CPU baselines.
Verify CUDA-enabled performance with deterministic bench scripts by running gpu-bench on a CUDA-enabled system to test vector distances, batch throughput, and full GPU functionality.
RigohlConfigure FAISS-GPU with IVF-PQ indexing for billion-scale vector search.
hey-kongBenchmark SGLang and Diffusers backends to optimize diffusion performance on multi-GPU setups.
pilotspaceBenchmark SIMD paths for dashtable probing, vector distance, and RESP parsing.
mkurmanEnable GPU acceleration for Python workloads across CUDA-enabled libraries.
