From acc2eb3633af46825b0193c4ffd01071d97e4838 Mon Sep 17 00:00:00 2001
From: Ryan Yin <xiaoyin_c@qq.com>
Date: Thu, 23 Nov 2023 12:28:22 +0800
Subject: [PATCH] feat: ai

---
 AI & Data-Science/Nvidia CUDA.md | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/AI & Data-Science/Nvidia CUDA.md b/AI & Data-Science/Nvidia CUDA.md
index c9ab42f0..04052ccb 100644
--- a/AI & Data-Science/Nvidia CUDA.md	
+++ b/AI & Data-Science/Nvidia CUDA.md	
@@ -20,6 +20,11 @@
 2. [Torch-TensorRT -  Compile Stable Diffusion and Inference](https://pytorch.org/TensorRT/tutorials/_rendered_examples/dynamo/torch_compile_stable_diffusion.html)
 3. [Compile a Torch Model(.pt) into Torch-TensorRT(xxx_trt.ts) via CLI](https://pytorch.org/TensorRT/cli/torchtrtc.html)
 
+### 2. TF32 精度
+
+1. https://pytorch.org/docs/stable/notes/cuda.html#tensorfloat-32-tf32-on-ampere-devices
+2. https://blogs.nvidia.com/blog/tensorfloat-32-precision-format/
+
 ## Nvidia GPU 架构、CUDA 版本、Nvidia 驱动版本、PyTorch 版本之间的兼容性
 
 > https://docs.nvidia.com/deeplearning/frameworks/support-matrix/index.html