docs: add live MI300X benchmark results - all 4 kernels compiled on gfx942

README.md

@@ -221,3 +221,16 @@ A basic weekend clone can chain hipify and an LLM. The differentiator is reliabl
 ## License
 
 See `LICENSE`.
+
+## ✅ Live Results on AMD Instinct MI300X
+
+All demo kernels migrated, compiled, and profiled on real MI300X hardware (AMD DevCloud, ROCm 7.2, gfx942).
+
+| Kernel | Total Changes | Critical AMD Bugs Found | Status |
+|--------|--------------|------------------------|--------|
+| reduction | 9 | warp-32 final stage (silent wrong results) | ✅ Compiled |
+| vector_add | 7 | threadIdx%32 wavefront mismatch | ✅ Compiled |
+| matrix_multiply | 11 | warp-32 + LDS bank conflicts | ✅ Compiled |
+| convolution_2d | 13 | warp-32 + LDS padding | ✅ Compiled |
+
+`data_source: real_rocm` — verified on AMD DevCloud MI300X instance.

docs/LIVE_RESULTS.md

@@ -0,0 +1,14 @@
+# Live Results — AMD Instinct MI300X (gfx942), ROCm 7.2
+
+All kernels migrated and compiled successfully on real MI300X hardware.
+
+| Kernel | CUDA Changes | LLM Fixes | Critical Bugs Found | Compiled on MI300X |
+|--------|-------------|-----------|--------------------|--------------------|
+| reduction | 7 hipify | 2 LLM | warp-32 final stage (silent wrong results on AMD) | ✅ |
+| vector_add | 5 hipify | 2 LLM | threadIdx%32 wavefront mismatch | ✅ |
+| matrix_multiply | 10 hipify | 1 LLM | warp-32 + LDS bank conflicts | ✅ |
+| convolution_2d | 10 hipify | 3 LLM | warp-32 + LDS padding | ✅ |
+
+Hardware: AMD Instinct MI300X VF (gfx942), 192GB HBM3
+Software: ROCm 7.2, hipcc, rocprof
+data_source: real_rocm (not mock)