045ea5df79
From corex-samples batched_gemm.cu, changed: float → half_t, OpClassSimt → OpClassTensorOp, Sm61 → Cu10 Uses __ivcorex_matrix_mad_f32x4_f16x4 via mma_cu10.h Default config: TB<128,128,32> Warp<32,32,32> Inst<16,16,16> Standalone test: correctness + perf for MoE decode (8 × 1x4096@4096x11008)
26 lines
753 B
Bash
Executable File
26 lines
753 B
Bash
Executable File
#!/bin/bash
|
|
# build_test_cutlass_batched.sh — Compile and test Cu10 TensorOp batched GEMM
|
|
set -eo pipefail
|
|
|
|
SAMPLES="/usr/local/corex-samples-3.2.3_x86_64/samples/cutlass"
|
|
SRC="ex_engine/xllm_kernels/cuda/moe_cutlass_batched.cu"
|
|
|
|
echo "=== Compile Cu10 TensorOp batched HGEMM ==="
|
|
/usr/local/corex/bin/clang++ \
|
|
--cuda-gpu-arch=ivcore10 --cuda-path=/usr/local/corex \
|
|
-I"${SAMPLES}/include" \
|
|
-I/usr/local/corex/include \
|
|
-L/usr/local/corex/lib64 -lcudart -lcutlass \
|
|
-DBUILD_STANDALONE_TEST \
|
|
-O2 -std=c++17 \
|
|
"$SRC" -o /tmp/test_cutlass_batched 2>&1
|
|
|
|
if [ -f /tmp/test_cutlass_batched ]; then
|
|
echo "Compile: SUCCESS"
|
|
echo ""
|
|
echo "=== Run ==="
|
|
/tmp/test_cutlass_batched
|
|
else
|
|
echo "Compile: FAILED"
|
|
fi
|