misc: add debug flag for ncu profile (#288)

* misc: add new zhihu tech blog link

* misc: add debug flag for ncu profile

Author: DefTruth

kernels/openai-triton/merge-attn-states/cuda_merge_attn_states.py

@@ -15,6 +15,7 @@
         "-U__CUDA_NO_BFLOAT16_CONVERSIONS__",
         "--expt-relaxed-constexpr",
         "--expt-extended-lambda",
+        "--generate-line-info -g", # for NCU debugging
         # "--use_fast_math"
     ], 
     extra_cflags=['-std=c++17'],