Fix example runtimes

Author: alihassanijr

examples/65_distributed_gemm/65_distributed_gemm.cu

@@ -132,7 +132,7 @@ using namespace cute;
 using TP = _8;
 static constexpr int TP_ = TP{};
 
-#if defined(CUTLASS_ARCH_MMA_SM90A_ENABLED) && \
+#if defined(CUTLASS_ARCH_MMA_SM90_SUPPORTED) && \
   (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
 
 // Distributed GEMM tiling/sharding schedule
@@ -252,7 +252,7 @@ HostTensorB tensor_B_arr[TP_];
 HostTensorD tensor_C_arr[TP_];
 HostTensorD tensor_D_arr[TP_];
 
-#endif // (defined(CUTLASS_ARCH_MMA_SM90A_ENABLED) &&
+#endif // (defined(CUTLASS_ARCH_MMA_SM90_SUPPORTED) &&
        // (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
@@ -345,8 +345,7 @@ struct Result {
 
 };
 
-#if defined(CUTLASS_ARCH_MMA_SM90A_ENABLED) && \
-  (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
+#if ((__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
 /// GEMM setup and evaluation
@@ -804,8 +803,7 @@ int run(Options &options) {
   return 0;
 }
 
-#endif // (defined(CUTLASS_ARCH_MMA_SM90A_ENABLED) &&
-       // (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
+#endif //(__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6))
 
 ///////////////////////////////////////////////////////////////////////////////////////////////////
 
@@ -859,7 +857,7 @@ int main(int argc, char const **args) {
   // Evaluate CUTLASS kernels
   //
 
-#if (defined(CUTLASS_ARCH_MMA_SM90A_ENABLED) && (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6)))
+#if ((__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 6)))
   run(options);
 #else
     std::cerr

examples/82_blackwell_distributed_gemm/82_blackwell_distributed_gemm.cu

@@ -129,10 +129,10 @@ using namespace cute;
 /////////////////////////////////////////////////////////////////////////////////////////////////
 
 // TP size (= number of processors/GPUs)
-using TP = _8;
+using TP = _4;
 static constexpr int TP_ = TP{};
 
-#if defined(CUTLASS_ARCH_MMA_SM100A_ENABLED) && \
+#if defined(CUTLASS_ARCH_MMA_SM100_SUPPORTED) && \
   (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8))
 
 // Distributed GEMM tiling/sharding schedule
@@ -254,7 +254,7 @@ HostTensorB tensor_B_arr[TP_];
 HostTensorD tensor_C_arr[TP_];
 HostTensorD tensor_D_arr[TP_];
 
-#endif // (defined(CUTLASS_ARCH_MMA_SM100A_ENABLED) &&
+#endif // (defined(CUTLASS_ARCH_MMA_SM100_SUPPORTED) &&
        // (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8))
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
@@ -347,8 +347,7 @@ struct Result {
 
 };
 
-#if defined(CUTLASS_ARCH_MMA_SM100A_ENABLED) && \
-  (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8))
+#if ((__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8)))
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
 /// GEMM setup and evaluation
@@ -812,8 +811,7 @@ int run(Options &options) {
   return 0;
 }
 
-#endif // (defined(CUTLASS_ARCH_MMA_SM100A_ENABLED) &&
-       // (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8))
+#endif // (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8))
 
 ///////////////////////////////////////////////////////////////////////////////////////////////////
 
@@ -867,7 +865,7 @@ int main(int argc, char const **args) {
   // Evaluate CUTLASS kernels
   //
 
-#if (defined(CUTLASS_ARCH_MMA_SM100A_ENABLED) && (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8)))
+#if ((__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 8)))
   run(options);
 #else
     std::cerr