[Typo] Correct architecture selection for CUDA and CDNA

Author: NaOHCC

benchmark/matmul/benchmark_matmul.py

@@ -53,7 +53,7 @@ def get_configs(args, kwargs):
         from tilelang.carver.roller.rasterization import NoRasterization
         import torch
 
-        arch = CDNA("cuda") if torch.version.hip is None else CUDA("hip")
+        arch = CUDA("cuda") if torch.version.hip is None else CDNA("hip")
         topk = 10
 
         carve_template = MatmulTemplate(

benchmark/matmul/benchmark_matmul_intrinsic.py

@@ -187,7 +187,7 @@ def get_configs(args, kwargs):
         from tilelang.carver.roller.rasterization import NoRasterization
         import torch
 
-        arch = CDNA("cuda") if torch.version.hip is None else CUDA("hip")
+        arch = CUDA("cuda") if torch.version.hip is None else CDNA("hip")
         topk = 10
 
         carve_template = MatmulTemplate(

examples/analyze/example_conv_analyze.py

@@ -96,7 +96,7 @@ def conv(
 
 def main():
     my_func = kernel(N, C, H, W, F, K, S, D, P, 64, 128, 32, 3, 256)
-    cuda_device = CDNA("cuda") if torch.version.hip is None else CUDA("hip")
+    cuda_device = CUDA("cuda") if torch.version.hip is None else CDNA("hip")
     result = Analyzer.analysis(my_func, cuda_device)
     print(result)
     print(f"Analyzed FLOPs: {result.total_flops}")

examples/analyze/example_gemm_analyze.py

@@ -49,7 +49,7 @@ def matmul(
 def main():
     my_func = kernel(128, 128, 32, 3, 128, True)
 
-    cuda_device = CDNA("cuda") if torch.version.hip is None else CUDA("hip")
+    cuda_device = CUDA("cuda") if torch.version.hip is None else CDNA("hip")
     result = Analyzer.analysis(my_func, cuda_device)
 
     print(f"Analyzed FLOPs: {result.total_flops}")

examples/gemm/example_gemm_autotune.py

@@ -16,7 +16,7 @@ def ref_program(A, B):
 
 def get_configs(M, N, K, with_roller=False, topk=20):
     if with_roller:
-        arch = CDNA("cuda") if torch.version.hip is None else CUDA("hip")
+        arch = CUDA("cuda") if torch.version.hip is None else CDNA("hip")
         carve_template = MatmulTemplate(
             M=M,
             N=N,