From 4abb4df215300ca14eb3f516eaa03d1dd3b3e652 Mon Sep 17 00:00:00 2001
From: Claude <claude@anthropic.com>
Date: Sat, 15 Aug 2026 05:30:15 +0000
Subject: [PATCH] =?UTF-8?q?test:=20cat=20Cu10=20CUTLASS=20files=20?=
 =?UTF-8?q?=E2=80=94=20mma=5Fcu10.h,=20iluvatar=5Fmma.hpp,=20batched=5Fgem?=
 =?UTF-8?q?m.cu,=20default=5Fmma=5Fcore=5Fcu10.h?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 qwen3_6_scripts/cat_cutlass_cu10.sh | 39 +++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100755 qwen3_6_scripts/cat_cutlass_cu10.sh

diff --git a/qwen3_6_scripts/cat_cutlass_cu10.sh b/qwen3_6_scripts/cat_cutlass_cu10.sh
new file mode 100755
index 00000000..c7580fa8
--- /dev/null
+++ b/qwen3_6_scripts/cat_cutlass_cu10.sh
@@ -0,0 +1,39 @@
+#!/bin/bash
+# cat_cutlass_cu10.sh — Cat the critical Cu10 CUTLASS files
+
+SAMPLES="/usr/local/corex-samples-3.2.3_x86_64/samples/cutlass"
+TF_INC="/usr/local/corex-3.2.3/lib64/python3/dist-packages/tensorflow/include/third_party/gpus/cuda/include"
+
+echo "=== 1. iluvatar_mma.hpp — the TCU intrinsics ==="
+cat /usr/local/corex/include/crt/iluvatar_mma.hpp
+echo ""
+
+echo "=== 2. mma_cu10.h — CUTLASS arch Cu10 MMA ==="
+cat "${SAMPLES}/include/cutlass/arch/mma_cu10.h"
+echo ""
+
+echo "=== 3. default_mma_core_cu10.h — threadblock MMA core for Cu10 ==="
+cat "${SAMPLES}/include/cutlass/gemm/threadblock/default_mma_core_cu10.h"
+echo ""
+
+echo "=== 4. batched_gemm.cu — the example we'll modify for grouped GEMM ==="
+cat "${SAMPLES}/examples/05_batched_gemm/batched_gemm.cu"
+echo ""
+
+echo "=== 5. cutlass samples tree (full, no depth limit) ==="
+find "${SAMPLES}" -type f | sort
+echo ""
+
+echo "=== 6. gemm_universal.h (device level) ==="
+cat "${SAMPLES}/include/cutlass/gemm/device/gemm_universal.h" | head -100
+echo "... ($(wc -l < "${SAMPLES}/include/cutlass/gemm/device/gemm_universal.h") total lines)"
+echo ""
+
+echo "=== 7. gemm_batched.h (device level) ==="
+cat "${SAMPLES}/include/cutlass/gemm/device/gemm_batched.h" | head -100
+echo "... ($(wc -l < "${SAMPLES}/include/cutlass/gemm/device/gemm_batched.h") total lines)"
+echo ""
+
+echo "=== 8. mma_tensor_op.h (warp level) ==="
+cat "${SAMPLES}/include/cutlass/gemm/warp/mma_tensor_op.h" | head -100
+echo "... ($(wc -l < "${SAMPLES}/include/cutlass/gemm/warp/mma_tensor_op.h") total lines)"