feat(rocm-support): support mamba2 on rocm (#18565)

Signed-off-by: Islam Almersawi <islam.almersawi@openinnovation.ai>
Co-authored-by: Islam Almersawi <islam.almersawi@openinnovation.ai>

csrc/mamba/causal_conv1d/causal_conv1d.cu

@@ -13,6 +13,10 @@
 #include <cub/block/block_load.cuh>
 #include <cub/block/block_store.cuh>
 
+#ifdef USE_ROCM
+    namespace cub = hipcub;
+#endif
+
 #include "static_switch.h"
 
 
@@ -501,15 +505,9 @@ void causal_conv1d_fwd_launch(ConvParamsBase &params, cudaStream_t stream) {
         auto kernel = &causal_conv1d_fwd_kernel<Ktraits>;
 
         if (kSmemSize >= 48 * 1024) {
-            #ifndef USE_ROCM
-            C10_CUDA_CHECK(cudaFuncSetAttribute(
-                kernel, cudaFuncAttributeMaxDynamicSharedMemorySize, kSmemSize));
-            #else
-            // There is a slight signature discrepancy in HIP and CUDA "FuncSetAttribute" function.
             C10_CUDA_CHECK(cudaFuncSetAttribute(
                 (void *) kernel, cudaFuncAttributeMaxDynamicSharedMemorySize, kSmemSize));
             std::cerr << "Warning (causal_conv1d fwd launch): attempting to set maxDynamicSharedMemorySize on an AMD GPU which is currently a non-op (in ROCm versions <= 6.1). This might lead to undefined behavior. \n" << std::endl;
-            #endif
         }
         kernel<<<grid, Ktraits::kNThreads, kSmemSize, stream>>>(params);