ppalanga/InferenceX
Open Source Continuous Inference Benchmarking Qwen3.5, DeepSeek, GPTOSS - GB200 NVL72 vs MI355X vs B200 vs GB300 NVL72 vs H100 & soon™ TPUv6e/v7/Trainium2/3
Repositories
ppalanga/recipes
Common recipes to run vLLM
ppalanga/vllm
A high-throughput and memory-efficient inference and serving engine for LLMs
ppalanga/sglang
SGLang is a high-performance serving framework for large language models and multimodal models.
ppalanga/AgentKernelArena
AgentKernelArena provides an end-to-end siloed-benchmarking environment where different LLM-powered agents—such as Cursor Agent, Claude Code, Codex, SWE-agent, and GEAK—can be evaluated side-by-side on the same GPU kernel tasks, using objective and reproducible metrics.
ppalanga/GEAK
Generating Efficient AI-Centric Kernels