PegaFlow is a high-performance KV cache offloading solution for vLLM v1 on single-node multi-GPU setups.
☆25Jan 7, 2026Updated 7 months ago
Alternatives and similar repositories for pegaflow
Users that are interested in pegaflow are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- High-performance KV cache storage for LLM inference — GPU offloading, SSD caching, and cross-node sharing via RDMA. Works with vLLM and S…☆198Updated this week
- High Performance KV Cache Store for LLM☆59May 20, 2026Updated 3 months ago
- This is the implementation repository of our SOSP'24 paper: Aceso: Achieving Efficient Fault Tolerance in Memory-Disaggregated Key-Value …☆24Oct 20, 2024Updated last year
- TiSpace manages VMs in K8s for developers☆14Nov 16, 2024Updated last year
- Pure Rust + CUDA LLM inference engine — no PyTorch, OpenAI-compatible, serves Qwen3 to Kimi-K2☆659Updated this week
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- ☆27Updated this week
- Skyloft: A General High-Efficient Scheduling Framework in User Space (SOSP 2024)☆38Sep 15, 2024Updated last year
- Tutorial Exercises and Code for GPU Communications Tutorial at HOT Interconnects 2026☆38Aug 21, 2026Updated last week
- The Intelligent Inference Scheduler for Large-scale Inference Services.☆69Feb 12, 2026Updated 6 months ago
- High-performance Rust benchmark client for vLLM serving endpoints.☆52Aug 3, 2026Updated 3 weeks ago
- Examples of usage for Mellanox HW offloads☆17Jan 18, 2022Updated 4 years ago
- LLM Kernel Library for Rust☆31Updated this week
- Contain some materials about CXL.☆20Feb 29, 2024Updated 2 years ago
- A scalable RAID system to aggregate performance and capacity of the next-generation storage.☆17Jan 3, 2024Updated 2 years ago
- Serverless GPU API endpoints on Runpod - Get Bonus Credits • AdSkip the infrastructure headaches. Auto-scaling, pay-as-you-go, no-ops approach lets you focus on innovating your application.
- DeepXTrace is a lightweight tool for precisely diagnosing slow ranks in DeepEP-based environments.☆102Jan 16, 2026Updated 7 months ago
- Ensō is a high-performance streaming interface for NIC-application communication.☆84Apr 11, 2026Updated 4 months ago
- [AFK] Hardware router in Chisel (THU Network Joint Lab 2020)☆14Oct 8, 2020Updated 5 years ago
- a size profiler for cuda binary☆71Jan 15, 2026Updated 7 months ago
- An Envoy inspired, ultimate LLM-first gateway for LLM serving and downstream application developers and enterprises☆27Apr 24, 2025Updated last year
- What if everything is a io_uring?☆17Nov 10, 2022Updated 3 years ago
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆117Aug 4, 2026Updated 3 weeks ago
- ☆20Mar 11, 2025Updated last year
- 鼠鼠的笔记本☆24Nov 5, 2024Updated last year
- Serverless GPU API endpoints on Runpod - Get Bonus Credits • AdSkip the infrastructure headaches. Auto-scaling, pay-as-you-go, no-ops approach lets you focus on innovating your application.
- High performance RMSNorm Implement by using SM Core Storage(Registers and Shared Memory)☆31Jan 22, 2026Updated 7 months ago
- ☆15Nov 14, 2023Updated 2 years ago
- ☆46Oct 15, 2025Updated 10 months ago
- Compare different hardware platforms via the Roofline Model for LLM inference tasks.☆126Mar 13, 2024Updated 2 years ago
- 各类内核的设计思路☆19May 19, 2021Updated 5 years ago
- KV cache store for distributed LLM inference☆433Nov 13, 2025Updated 9 months ago
- A high-throughput and memory-efficient inference and serving engine for LLMs☆17Jun 3, 2024Updated 2 years ago
- A fast and scalable distributed lock service using programmable switches.☆21Jul 30, 2024Updated 2 years ago
- FlashSampling: Fast and Memory-Efficient Exact Sampling (https://huggingface.co/papers/2603.15854)☆87Aug 5, 2026Updated 3 weeks ago
- End-to-end encrypted email - Proton Mail • AdSpecial offer: 40% Off Yearly / 80% Off First Month. All Proton services are open source and independently audited for security.
- ☆34Feb 3, 2025Updated last year
- Deft: A Scalable Tree Index for Disaggregated Memory☆22Apr 23, 2025Updated last year
- Concurrent ART (adaptive radix tree)☆181Sep 26, 2025Updated 11 months ago
- A tool for cross-checking Verilog compilers☆15Apr 16, 2025Updated last year
- 🔫 Toy ML-like Imperative Programming Language (WIP)☆10Apr 15, 2021Updated 5 years ago
- In-Memory Key-Value Store Live Migration with NetMigrate☆19Jun 22, 2024Updated 2 years ago
- A NCCL extension library, designed to efficiently offload GPU memory allocated by the NCCL communication library.☆116Dec 17, 2025Updated 8 months ago