From 482aabdea3e48508647615cab94ee6285b0f1f2b Mon Sep 17 00:00:00 2001 From: dylanyunlon Date: Sat, 1 Aug 2026 01:29:57 +0800 Subject: [PATCH] fix(muh_kernel_map): add threads >= 32 floor in Python scale_mem_bound Mirrors the C++ fix in common.cuh.