From d462dee37da2e1a744c6694cb65a1c735600f262 Mon Sep 17 00:00:00 2001
From: Max Hu <hyoung2991@gmail.com>
Date: Tue, 11 Aug 2026 23:09:54 -0400
Subject: [PATCH] [Bugfix][Triton] Make fp8_min/fp8_max constexpr in
 _quantize_pad_fp8_kernel (#51872)

Signed-off-by: Max Hu <maxhu@nvidia.com>
Co-authored-by: Max Hu <maxhu@nvidia.com>
---
 vllm/kernels/triton/qkv_padded_fp8_quant.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/vllm/kernels/triton/qkv_padded_fp8_quant.py b/vllm/kernels/triton/qkv_padded_fp8_quant.py
index 74dfe10436..d8e772f648 100644
--- a/vllm/kernels/triton/qkv_padded_fp8_quant.py
+++ b/vllm/kernels/triton/qkv_padded_fp8_quant.py
@@ -34,8 +34,8 @@ def _quantize_pad_fp8_kernel(
     n_rows,
     n_cols,
     n_cols_padded,
-    fp8_min,
-    fp8_max,
+    fp8_min: tl.constexpr,
+    fp8_max: tl.constexpr,
     SKIP_SCALE: tl.constexpr,
     BLOCK_M: tl.constexpr,
     BLOCK_N: tl.constexpr,