powderluv

@powderluv · User

GitHub profile ↗ · Compare

@nod-ai San Francisco84 followers213 repositories

Repositories

powderluv/Hyperloom

An agentic system that auto-optimizes LLM workloads on AMD GPUs.

★ 0Forks 0

powderluv/rustproof

Rustproof: a Rust + Verus isolation nucleus that safely hosts an untrusted GPU compute driver (gfx1201). Design/scaffold stage.

★ 0RustForks 0

powderluv/sglang

SGLang is a high-performance serving framework for large language models and multimodal models.

★ 0PythonForks 0

powderluv/tide

TIDE — Throughput × Interactivity Density Envelope. Single-number LLM inference performance score with regime-aware diagnostics, computed from full concurrency/TP/ISL-OSL/model sweeps. Works on InferenceX data.

★ 2PythonForks 1

powderluv/etdemo

EdgeTPU C++ Example with Cmake (without Batshoot Bazel)

★ 0C++Forks 0

powderluv/hrx-system-rs

Rust port of ROCm/hrx-system: the libhrx public C ABI (114/114 symbols, byte-identical + perf-parity vs C on MI300X) and the HIP LD_PRELOAD passthrough/interceptor layer.

★ 0RustForks 0

powderluv/agent-tracer

Analyze Claude Code + Codex session logs on this machine — unified Perfetto trace, agent + system telemetry, ranked optimization hints.

★ 0PythonForks 1

powderluv/grom

Grom ChromiumOS distribution workspace metadata and VM helpers

★ 0ShellForks 0

powderluv/codex

Lightweight coding agent that runs in your terminal

★ 0RustForks 0

powderluv/dyninst

DyninstAPI: Tools for binary instrumentation, analysis, and modification.

★ 0Forks 0

powderluv/autokernel

Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.

★ 0Forks 0

powderluv/Magpie

A lightweight, general-purpose framework for evaluating GPU kernel correctness and performance.

★ 0Forks 0

powderluv/TheRock

The HIP Environment and ROCm Kit - A lightweight open source build system for HIP and ROCm

★ 0Forks 0