From 5a24c7538fcf5ccc04770b03fe569a98cf1b0f5d Mon Sep 17 00:00:00 2001
From: Daniel Bevenius <daniel.bevenius@gmail.com>
Date: Thu, 14 May 2026 05:28:56 +0200
Subject: [PATCH] Revert "ggml-cuda : add explicit casts to -INFINITY for float
 and half2 types"

This reverts commit a2839b4404de473bc7af127b7b308d530afda024.

Reverting this as after closer inspection these only warnings and not
errors.
---
 ggml/src/ggml-cuda/common.cuh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/ggml/src/ggml-cuda/common.cuh b/ggml/src/ggml-cuda/common.cuh
index 246a76193..10817505d 100644
--- a/ggml/src/ggml-cuda/common.cuh
+++ b/ggml/src/ggml-cuda/common.cuh
@@ -582,9 +582,9 @@ template <typename T> struct block_reduce_policy<block_reduce_method::MAX, T> {
 
     static __device__ T sentinel() {
         if constexpr (std::is_same_v<T, float>) {
-            return -(float)INFINITY;
+            return -INFINITY;
         } else if constexpr (std::is_same_v<T, half2>) {
-            return make_half2(__float2half(-(float)INFINITY), __float2half(-(float)INFINITY));
+            return make_half2(-INFINITY, -INFINITY);
         } else {
             static_assert(ggml_cuda_dependent_false_v<T>, "Unsupported type for block reduce max");
         }