upstream: add GEMM kernel references from 4 repos for BI-V100 porting

Sources (all CUDA 10.2 compatible, no CUTLASS/Triton dependency):
- leimao/CUDA-GEMM-Optimization: v00-v07, fp16 WMMA variant, double buffered
- siboehm/SGEMM_CUDA: kernel 1-12, warp tiling + double buffering
- wangzyon/NVIDIA_SGEMM_PRACTICE: kernel 1-7
- edtallison/sgemm-cuda: kernel 1-12 (reimplementation with notes)

Key porting issue: ALL kernels hardcode WARPSIZE=32.
BI-V100 has warp_size=64. Need to:
1. Replace all 32U / WARPSIZE constants with 64
2. Adjust warp subtile decomposition (WMITER, WNITER, WSUBM, WSUBN)
3. Adjust shared memory bank conflict avoidance (may have different bank count)
4. Test __shfl_down_sync with mask=0xFFFFFFFFFFFFFFFF (64-bit)

upstream_ref/cuda_gemm_optimization/cuda_gemm.hpp

@@ -0,0 +1,98 @@
+#ifndef CUDA_GEMM_HPP
+#define CUDA_GEMM_HPP
+
+#include <cuda_runtime.h>
+
+template <typename T>
+void launch_gemm_kernel_v00(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v01(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v02(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v02_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v03(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v03_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+template <typename T>
+void launch_gemm_kernel_v04(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v04_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v05(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+template <typename T>
+void launch_gemm_kernel_v05_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v06(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+template <typename T>
+void launch_gemm_kernel_v06_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v06_vectorized_double_buffered(
+    size_t m, size_t n, size_t k, T const* alpha, T const* A, size_t lda,
+    T const* B, size_t ldb, T const* beta, T* C, size_t ldc,
+    cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v07(size_t m, size_t n, size_t k, T const* alpha,
+                            T const* A, size_t lda, T const* B, size_t ldb,
+                            T const* beta, T* C, size_t ldc,
+                            cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v07_vectorized(size_t m, size_t n, size_t k,
+                                       T const* alpha, T const* A, size_t lda,
+                                       T const* B, size_t ldb, T const* beta,
+                                       T* C, size_t ldc, cudaStream_t stream);
+
+template <typename T>
+void launch_gemm_kernel_v07_vectorized_double_buffered(
+    size_t m, size_t n, size_t k, T const* alpha, T const* A, size_t lda,
+    T const* B, size_t ldb, T const* beta, T* C, size_t ldc,
+    cudaStream_t stream);
+#endif