From e63cf1b3745204f5d33928c6bb8e77c166b90bb3 Mon Sep 17 00:00:00 2001
From: DefTruth <31974251+DefTruth@users.noreply.github.com>
Date: Wed, 15 Jan 2025 00:12:59 +0800
Subject: [PATCH] [bench] update perf plots for qkv swizzle (#40)

---
 README.md | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index f271f71..7edee51 100644
--- a/README.md
+++ b/README.md
@@ -13,17 +13,18 @@
 
 🤖[WIP] **FFPA**: Yet antother **Faster Flash Prefill Attention** with **O(1) SRAM complexity** & **O(d/4) or O(1) register complexity** for large headdim (D > 256), almost **1.5x~2x** 🎉 faster than SDPA EA with or without MMA Acc F32 on many devices: [📈L20 ~1.9x↑🎉](#L1-bench-l20), [📈 A30 ~1.8x↑🎉](#L1-bench-a30), [📈3080 ~2.9x↑🎉](#L1-bench-3080), [📈4090 ~2.1x↑🎉](#L1-bench-4090). 
 
-<!--
 <div align='center'>
   <img src='https://github.com/user-attachments/assets/447e2937-f7c8-47c8-8550-8c0c71b910e6' width="407px">
   <img src='https://github.com/user-attachments/assets/65a8d564-8fa7-4d66-86b9-e238feb86143' width="407px">
 </div> 
--->
+
+<!--
 <div align='center'>
   <img src='https://github.com/user-attachments/assets/cba2edce-ac0d-412e-823c-7eea2cc63f83' height="170px" width="270px">
   <img src='https://github.com/user-attachments/assets/447e2937-f7c8-47c8-8550-8c0c71b910e6' height="170px" width="270px">
   <img src='https://github.com/user-attachments/assets/65a8d564-8fa7-4d66-86b9-e238feb86143' height="170px" width="270px">
 </div> 
+-->
 
 
 💡NOTE: This project is still in its early dev stages and now provides some kernels and benchmarks for reference. More features will be added in the future. (Welcome to 🌟👆🏻star this repo to support me ~)