SC'25 UltraAttn: Efficiently Parallelizing Attention through Hierarchical Context-Tiling
-
Updated
Aug 14, 2025 - Python
#
context-parallelism
Here are 2 public repositories matching this topic...
Hardware barrier-free distributed computing plane (FNG V3) fusing viscous Burgers' and Vorticity equations directly into XLA registers to pre-rectify high-order moment skewness & eliminate stalls inside distributed LLM attention rails.
hpc fault-tolerance distributed-computing gpu-acceleration numerical-analysis co-design mathematical-physics burgers-equation jax xla tpu-acceleration transformer-attention context-parallelism llm-infrastructure
-
Updated
Jul 21, 2026 - Python
Improve this page
Add a description, image, and links to the context-parallelism topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the context-parallelism topic, visit your repo's landing page and select "manage topics."