Accelerating Mamba-2 with High-Performance Forward Pass Kernels in NVIDIA CUDA Tile
<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-768x432.jpg" class="webfeedsFeaturedVisual wp-post-image" alt="Decorative math image." style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-768x432.jpg 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-300x169.jpg 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-625x351.jpg 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-179x101.jpg 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-1536x864.jpg 1536w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-2048x1152.jpg 2048w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-645x363.jpg 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-660x370.jpg 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-500x281.jpg 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-160x90.jpg 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-362x204.jpg 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-196x110.jpg 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-1024x576.jpg 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-960x540.jpg 960w" sizes="(max-width: 768px) 100vw, 768px" title="Solving-Math" />NVIDIA CUDA Tile (cuTile) is a Python-based DSL and parallel programming model for NVIDIA GPUs that automatically leverages advanced hardware capabilities to...<img width="768" height="432" src="https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-768x432.jpg" class="webfeedsFeaturedVisual wp-post-image" alt="Decorative math image." style="display: block; margin-bottom: 5px; clear:both;max-width: 100%;" link_thumbnail="" decoding="async" loading="lazy" srcset="https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-768x432.jpg 768w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-300x169.jpg 300w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-625x351.jpg 625w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-179x101.jpg 179w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-1536x864.jpg 1536w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-2048x1152.jpg 2048w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-645x363.jpg 645w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-660x370.jpg 660w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-500x281.jpg 500w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-160x90.jpg 160w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-362x204.jpg 362w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-196x110.jpg 196w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-1024x576.jpg 1024w, https://developer-blogs.nvidia.com/wp-content/uploads/2025/11/Solving-Math-960x540.jpg 960w" sizes="auto, (max-width: 768px) 100vw, 768px" title="Solving-Math" /><p>NVIDIA CUDA Tile (cuTile) is a Python-based DSL and parallel programming model for NVIDIA GPUs that automatically leverages advanced hardware capabilities to deliver peak performance for tile-level GPU programming. By abstracting thread- and warp-level scheduling behind high-level, tile-centric primitives, cuTile enables you to write GPU kernels in Python while compiler optimizations automatically…</p>
<p><a href="https://developer.nvidia.com/blog/accelerating-mamba-2-with-high-performance-forward-pass-kernels-in-nvidia-cuda-tile/" rel="nofollow" data-wpel-link="internal" target="_self">Source</a></p>
Read original article ↗
Related Articles
At a creative industry event that took place today in Berlin — hosted in collaboration with Adobe, Bertelsmann, NVIDIA a
Siemens and NVIDIA Accelerate Lithography for Next-Generation AI Chips
Siemens has announced advances in its collaboration with NVIDIA to accelerate computational lithography, addressing a de
RTX Spark Ignites Global Celebration as New Windows PCs Arrive
NVIDIA RTX Spark laptops are now available, bringing NVIDIA Blackwell RTX graphics, up to 128GB of unified memory and th
Code, Test, and Deploy on One Machine with NVIDIA RTX Spark
AI agents are changing how software gets built. Developers have gone from writing every line of code by hand to directin