Tagged "kernel-optimization"
11 articles tagged kernel-optimization, 15 March 2026 to 14 October 2026. Newest first.
-
Modular's MAX Framework Delivers 5.5x Speedup Over vLLM on Kimi Models
Modular's MAX inference framework achieves 5.5x performance improvement over vLLM when running Kimi K2.5 models, demonstrating significant optimisations for local LLM inference speed and efficiency.
-
Llama.cpp Optimizes Kernel Execution with RMS_NORM and SCALE Fusion
The latest llama.cpp release fuses RMS_NORM and SCALE operations into a single kernel, eliminating 96 extra kernel launches per batch on large models like Qwen3.8-27B. This optimization reduces computational overhead without sacrificing accuracy.
-
Husky: Model-Specific Inference Engine Achieves 4.5x Speedup Over Apple MLX
A new inference engine optimised for Apple Silicon demonstrates dramatic performance improvements over existing solutions, achieving up to 4.5x faster inference than MLX for specific model architectures.
-
llama.cpp Adds Flash Attention Tuning for AMD RDNA4 and Optimizations
llama.cpp release b10905 enhances Flash Attention performance with GPU-specific tuning for AMD RDNA4 architecture and improves kernel selection logic. These optimizations reduce latency and memory bandwidth requirements for inference across AMD accelerators.
-
NVIDIA Local AI Optimization Delivers 1.9x Speedup on 24GB RTX GPUs
NVIDIA has announced performance optimizations for local AI inference on RTX GPUs with 24GB+ VRAM, achieving 1.9x speed improvements that rival cloud API latency and economics, making consumer hardware increasingly viable for production local LLM deployment.
-
llama.cpp Build 10605: Mamba2 GEMM Optimization Improves State-Space Model Performance
The latest llama.cpp release optimizes Mamba2 models by flattening input/output projections to dispatch GEMM operations instead of GEMV, delivering better GPU utilization and inference speed for state-space architectures.
-
vLLM v0.27.0 Released with Major Kernel Improvements and New Model Support
vLLM's latest release brings 561 commits from 242 contributors, including full-stack support for Kimi K3 models, new kernel optimizations, and expanded hardware compatibility. The release focuses on performance improvements critical for efficient local LLM serving.
-
vLLM v0.27.0 Brings Major Performance Improvements and New Model Support
vLLM v0.27.0 features 561 commits from 242 contributors including full-stack Kimi K3 support, new kernel optimizations, and DeepGEMM integration. This release significantly improves inference performance for local LLM serving.
-
How To Build Your Own LLM Runtime From Scratch
A comprehensive guide on constructing custom LLM inference runtimes, providing practitioners with deep knowledge to optimize and control local model deployment without relying on black-box frameworks.
-
Is Anyone Working on an AI Operating System?
An active Hacker News discussion exploring whether anyone is building operating systems designed from the ground up for AI workloads and inference, addressing questions about architecture, scheduling, and optimization for local LLM deployment infrastructure.
-
Qwen3.5-397B Achieves 282 tok/s on 4x RTX PRO 6000 Blackwell Through Custom CUTLASS Kernel
A developer achieved a 5x performance improvement on the massive Qwen3.5-397B model by building a custom CUTLASS kernel to fix SM120's broken MoE GEMM tiles, reaching 282 tokens/second on Blackwell GPUs. This breakthrough demonstrates significant optimization potential for running large models locally with multi-GPU setups.