5 repository-uri
Extends tensor parallelism across machines by joining local CUDA ranks into a global NCCL communicator coordinated by a head node.
Distinct from Tensor Parallelism: Distinct from general Tensor Parallelism: specifically addresses multi-node coordination with NCCL communicators.
Explore 5 awesome GitHub repositories matching artificial intelligence & ml · Multi-Node. Refine with filters or upvote what's useful.
mistral.rs is an inference engine for large language models that runs locally and exposes models behind OpenAI and Anthropic-compatible APIs. It serves as a multi-model serving platform, capable of loading several models in a single server process with per-request routing and on-demand loading and unloading. The engine supports multimodal inference, processing text alongside images, video, audio, and speech inputs, and includes a quantized model deployment runtime that reduces memory use and speeds up inference on consumer hardware. The project distinguishes itself through an agentic tool exe
Extends tensor parallelism across machines by joining local CUDA ranks into a global NCCL communicator.
Dynamo is a distributed inference orchestration platform designed for large language models. It functions as a system to coordinate prefill and decode phases across GPU nodes, utilizing a multi-backend runtime adapter to connect engines like vLLM and TensorRT-LLM through a unified block-oriented memory interface. An OpenAI-compatible API server provides the frontend for integration with existing tools and clients. The project is distinguished by its disaggregated serving architecture, which separates prompt processing and token generation onto independent GPU pools to optimize throughput and
Spreads tensor-parallel inference across multiple hardware nodes using global NCCL communicators.
FlashInfer is a library of high-performance GPU kernels purpose-built for accelerating large language model inference. It provides optimized implementations for attention operations (including flash attention, page attention, multi-head latent attention, and cascade attention) using paged key-value caches, fused kernel composition, and just-in-time compilation. The library also includes specialized kernels for mixture-of-experts layers, block-scaled low-precision quantization (FP8, FP4), and distributed collective communication. What distinguishes FlashInfer is its fused all-reduce communicat
Provides GPU buffer allocation with multicast access for multi-node communication in distributed inference.
NCCL este o bibliotecă de comunicare de înaltă performanță și un framework de calcul distribuit pe GPU, conceput pentru executarea schimburilor de date colective și punct-la-punct pe mai multe GPU-uri în sisteme cu un singur nod sau multi-nod. Servește ca strat de transport RDMA pentru GPU și orchestrator de memorie, facilitând sincronizarea cu lățime de bandă mare a datelor și a gradienților de model pentru antrenarea și inferența distribuită pe GPU. Biblioteca se distinge prin capacitatea sa de a executa primitive de comunicare direct din kernel-urile GPU, eliminând CPU-ul gazdă din calea critică. Utilizează selecția de căi conștientă de topologie pentru a optimiza mișcarea datelor și folosește transportul de rețea bazat pe RDMA, inclusiv InfiniBand și NVLink, pentru a permite accesul la memorie zero-copy între dispozitive pe diferite noduri fizice. Proiectul acoperă o gamă largă de tipare de comunicare colectivă, inclusiv reduceri, broadcast-uri, gather-uri și schimburi all-to-all, alături de accesul la memorie la distanță punct-la-punct. Oferă gestionare cuprinzătoare a comunicatorului pentru inițializarea, partiționarea și redimensionarea grupurilor GPU, precum și gestionarea specializată a memoriei pentru înregistrarea bufferelor și coordonarea memoriei partajate a dispozitivului. Sistemul include o suită de instrumente de monitorizare și observabilitate pentru urmărirea stării, logarea diagnostică și monitorizarea evenimentelor în timp real, precum și interfețe de integrare pentru framework-uri de machine learning, CUDA graphs, MPI și Python.
Coordinates GPU communication across multiple physical nodes by establishing NVLink domains and identifiers.
Distributed-llama is a distributed inference engine and command line tool for running large language models across multiple networked machines. It functions as a compute cluster manager that coordinates worker nodes to share the computational load of a single model. The system utilizes tensor parallelism to shard model weights across different hosts, allowing the execution of models that exceed the memory capacity of a single piece of hardware. It includes a dedicated format converter to transform standard model files into a compatible binary layout optimized for distributed loading. The eng
Groups independent machines into a virtual compute resource to enable multi-node tensor parallelism.