diff --git a/src/libtorchaudio/forced_align/gpu/compute.cu b/src/libtorchaudio/forced_align/gpu/compute.cu
index c3be77a256..444a4f8f6d 100644
--- a/src/libtorchaudio/forced_align/gpu/compute.cu
+++ b/src/libtorchaudio/forced_align/gpu/compute.cu
@@ -8,10 +8,6 @@
 #include <cub/cub.cuh>
 #include <limits.h>
 
-#ifdef USE_ROCM
-  #include <libtorchaudio/hip_namespace_shim.h>
-#endif
-
 namespace {
 constexpr int kNumThreads =
     1024; // Number of threads to run CUDA kernel in parallel.
diff --git a/src/libtorchaudio/hip_namespace_shim.h b/src/libtorchaudio/hip_namespace_shim.h
deleted file mode 100644
index 7d03d919a7..0000000000
--- a/src/libtorchaudio/hip_namespace_shim.h
+++ /dev/null
@@ -1,31 +0,0 @@
-#pragma once
-
-#include <libtorchaudio/hip_utils.h>
-
-namespace libtorchaudio::hip {
-
-inline hipStream_t getCurrentHIPStreamMasqueradingAsCUDA(
-    torch::stable::DeviceIndex device_index = -1) {
-  return cuda::getCurrentHIPStreamMasqueradingAsCUDA(device_index);
-}
-
-inline void setCurrentHIPStreamMasqueradingAsCUDA(
-    hipStream_t stream,
-    torch::stable::DeviceIndex device_index = -1) {
-  cuda::setCurrentHIPStreamMasqueradingAsCUDA(stream, device_index);
-}
-
-inline hipStream_t getStreamFromPoolMasqueradingAsCUDA(
-    const bool isHighPriority = false,
-    torch::stable::DeviceIndex device_index = -1) {
-  return cuda::getStreamFromPoolMasqueradingAsCUDA(
-      isHighPriority, device_index);
-}
-
-inline void synchronize(
-    hipStream_t stream,
-    torch::stable::DeviceIndex device_index = -1) {
-  cuda::synchronize(stream, device_index);
-}
-
-} // namespace libtorchaudio::hip
diff --git a/src/libtorchaudio/rnnt/gpu/compute.cu b/src/libtorchaudio/rnnt/gpu/compute.cu
index 40c5945f81..7e99fec395 100644
--- a/src/libtorchaudio/rnnt/gpu/compute.cu
+++ b/src/libtorchaudio/rnnt/gpu/compute.cu
@@ -6,10 +6,6 @@
 #include <torch/csrc/stable/ops.h>
 #include <torch/headeronly/core/Dispatch_v2.h>
 
-#ifdef USE_ROCM
-  #include <libtorchaudio/hip_namespace_shim.h>
-#endif
-
 namespace torchaudio {
 namespace rnnt {
 namespace gpu {