paperg / NCCL_GPLinks
Separate from hardware and used to learn some NCCL mechanisms
☆19Updated last year
Alternatives and similar repositories for NCCL_GP
Users that are interested in NCCL_GP are comparing it to the libraries listed below
Sorting:
- example code for using DC QP for providing RDMA READ and WRITE operations to remote GPU memory☆132Updated 10 months ago
- NCCL Profiling Kit☆138Updated 11 months ago
- RDMA and SHARP plugins for nccl library☆197Updated this week
- ☆186Updated 2 years ago
- ☆73Updated 2 months ago
- Repository for MLCommons Chakra schema and tools☆109Updated last week
- ☆203Updated 3 weeks ago
- Hooked CUDA-related dynamic libraries by using automated code generation tools.☆158Updated last year
- A prefill & decode disaggregated LLM serving framework with shared GPU memory and fine-grained compute isolation.☆87Updated last month
- DeepSeek-V3/R1 inference performance simulator☆149Updated 2 months ago
- ☆36Updated 5 months ago
- ☆16Updated 4 months ago
- Paella: Low-latency Model Serving with Virtualized GPU Scheduling☆59Updated last year
- Synthesizer for optimal collective communication algorithms☆108Updated last year
- NCCL Fast Socket is a transport layer plugin to improve NCCL collective communication performance on Google Cloud.☆116Updated last year
- Microsoft Collective Communication Library☆350Updated last year
- Artifact of OSDI '24 paper, ”Llumnix: Dynamic Scheduling for Large Language Model Serving“☆61Updated last year
- Injecting Adrenaline into LLM Serving: Boosting Resource Utilization and Throughput via Attention Disaggregation☆24Updated 3 months ago
- ASTRA-sim2.0: Modeling Hierarchical Networks and Disaggregated Systems for Large-model Training at Scale☆378Updated last week
- TACCL: Guiding Collective Algorithm Synthesis using Communication Sketches☆73Updated last year
- Repository for MLCommons Chakra schema and tools☆39Updated last year
- High performance Transformer implementation in C++.☆125Updated 5 months ago
- ☆37Updated 6 months ago
- A lightweight design for computation-communication overlap.☆143Updated this week
- REEF is a GPU-accelerated DNN inference serving system that enables instant kernel preemption and biased concurrent execution in GPU sche…☆94Updated 2 years ago
- ☆138Updated last year
- ☆23Updated 11 months ago
- Ultra and Unified CCL☆165Updated this week
- An interference-aware scheduler for fine-grained GPU sharing☆140Updated 4 months ago
- ☆35Updated 3 years ago