Kernel forge for the-den. Not just kernels. Kennels. Where the Hydra lives. Raw PTX tensor core path for Blackwell SM120. OMMA.SF.16864 cubins, SASS verification, fragment mapping. NVFP4-native.
-
Updated
Jun 9, 2026 - C++
#
sm120a
Here are 3 public repositories matching this topic...
Build your own AI companion. GPU-native avatar generation, Substance-Painter cognitive layers, holographic progressive memory, iDream 3D world engine, three-model cognitive hierarchy. Runs entirely on consumer GPUs. No cloud.
cuda moe quantization inference-engine blackwell llm-inference omma nvfp4 expert-offloading sm120a wh4 gpu-resident dengine
-
Updated
Jun 9, 2026
Windows NVIDIA-only Triton 3.7.0 build pipeline for RTX 5090 / Blackwell sm_120a, with FP8 tl.dot validation and peak benchmark results.
windows benchmark gpu cuda pytorch nvidia triton windows-build tensor-cores blackwell fp8 triton-lang rtx-5090 triton-windows cuda-13 sm120a fp8-matmul
-
Updated
May 26, 2026 - Batchfile
Improve this page
Add a description, image, and links to the sm120a topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the sm120a topic, visit your repo's landing page and select "manage topics."