add 1210

Author: yzh119

csrc/xqa/mha.cu

@@ -89,7 +89,7 @@ constexpr uint32_t cvtExpansion = exactDiv(inputElemSize, cacheElemSize);
 constexpr uint32_t preferedKHeadPartBytes = 64;
 __constant__ constexpr uint32_t cacheVTileSeqLen = 32;
 #else
-#if __CUDA_ARCH__ == 860 || __CUDA_ARCH__ == 890 || __CUDA_ARCH__ == 1200
+#if __CUDA_ARCH__ == 860 || __CUDA_ARCH__ == 890 || __CUDA_ARCH__ == 1200 || __CUDA_ARCH__ == 1210
 constexpr uint32_t preferedKHeadPartBytes = 64;
 __constant__ constexpr uint32_t cacheVTileSeqLen = 32;
 #elif __CUDA_ARCH__ == 800 || __CUDA_ARCH__ == 870 || __CUDA_ARCH__ == 900 || \

csrc/xqa/utils.cuh

@@ -42,7 +42,7 @@ inline constexpr int32_t kBAD_PAGE_INDEX = -1;
 __constant__ constexpr float kE4M3_MAX = 448.F;
 
 #ifdef __CUDA_ARCH__
-#if __CUDA_ARCH__ == 860 || __CUDA_ARCH__ == 890 || __CUDA_ARCH__ == 1200
+#if __CUDA_ARCH__ == 860 || __CUDA_ARCH__ == 890 || __CUDA_ARCH__ == 1200 || __CUDA_ARCH__ == 1210
 constexpr uint32_t kMAX_SMEM_SIZE = (99u << 10);
 #elif __CUDA_ARCH__ == 800 || __CUDA_ARCH__ == 870
 constexpr uint32_t kMAX_SMEM_SIZE = (163u << 10);