diff --git a/test/test_foreach.py b/test/test_foreach.py
index 7ac128d6bac8..12c2ec7ccc96 100644
--- a/test/test_foreach.py
+++ b/test/test_foreach.py
@@ -38,7 +38,6 @@ from torch.testing._internal.common_utils import (
     gradcheck,
     parametrize,
     run_tests,
-    skipIfRocmVersionLessThan,
     skipIfTorchDynamo,
     TEST_WITH_ROCM,
     TestCase,
@@ -196,7 +195,6 @@ class TestForeach(TestCase):
                         zero_size=True,
                     )
 
-    @skipIfRocmVersionLessThan((6, 0))
     @ops(
         foreach_unary_op_db
         + foreach_binary_op_db
diff --git a/test/test_linalg.py b/test/test_linalg.py
index 907db351d160..ba6cf8e03839 100644
--- a/test/test_linalg.py
+++ b/test/test_linalg.py
@@ -29,8 +29,7 @@ from torch.testing._internal.common_device_type import \
     (instantiate_device_type_tests, dtypes, has_cusolver, has_hipsolver,
      onlyCPU, skipCUDAIfNoMagma, skipCPUIfNoLapack, precisionOverride,
      skipCUDAIfNoMagmaAndNoCusolver, skipCUDAIfRocm, onlyNativeDeviceTypes, dtypesIfCUDA,
-     onlyCUDA, skipMeta, skipCUDAIfNoCusolver, skipCUDAIfNotRocm, skipCUDAIfRocmVersionLessThan,
-     dtypesIfMPS, largeTensorTest)
+     onlyCUDA, skipMeta, skipCUDAIfNoCusolver, skipCUDAIfNotRocm, dtypesIfMPS, largeTensorTest)
 from torch.testing import make_tensor
 from torch.testing._internal.common_dtype import (
     all_types, all_types_and_complex_and, floating_and_complex_types, integral_types,
@@ -7303,7 +7302,6 @@ scipy_lobpcg  | {eq_err_scipy:10.2e}  | {eq_err_general_scipy:10.2e}  | {iters2:
     @unittest.skipIf(IS_WINDOWS, "Skipped on Windows!")
     @unittest.skipIf(SM90OrLater and not TEST_WITH_ROCM, "Expected failure on sm90")
     @unittest.skipIf(IS_FBCODE and IS_REMOTE_GPU, "cublas runtime error")
-    @skipCUDAIfRocmVersionLessThan((6, 0))
     @onlyCUDA
     @parametrize("k", [16, 32])
     @parametrize("n", [16, 32])
@@ -7374,7 +7372,6 @@ scipy_lobpcg  | {eq_err_scipy:10.2e}  | {eq_err_general_scipy:10.2e}  | {iters2:
 
     @unittest.skipIf(IS_WINDOWS, "Skipped on Windows!")
     @unittest.skipIf(IS_FBCODE and IS_REMOTE_GPU, "cublas runtime error")
-    @skipCUDAIfRocmVersionLessThan((6, 0))
     @onlyCUDA
     def test__int_mm_errors(self, device):
 
diff --git a/test/test_matmul_cuda.py b/test/test_matmul_cuda.py
index b1f7f91a34de..f94a7c04da85 100644
--- a/test/test_matmul_cuda.py
+++ b/test/test_matmul_cuda.py
@@ -36,7 +36,6 @@ from torch.testing._internal.common_utils import (
     parametrize,
     run_tests,
     skipIfRocm,
-    skipIfRocmVersionLessThan,
     TEST_CUDA,
     TEST_WITH_ROCM,
     TestCase,
@@ -144,7 +143,6 @@ class TestMatmulCuda(InductorTestCase):
         torch.backends.cuda.matmul.allow_fp16_accumulation = orig_fp16_accumulate
 
     @onlyCUDA
-    @skipIfRocmVersionLessThan((5, 2))
     # imported 'tol' as 'xtol' to avoid aliasing in code above
     @toleranceOverride({torch.float16: xtol(atol=1e-1, rtol=1e-1),
                         torch.bfloat16: xtol(atol=1e-1, rtol=1e-1),
@@ -158,7 +156,6 @@ class TestMatmulCuda(InductorTestCase):
 
     @onlyCUDA
     @xfailIfSM100OrLaterAndCondition(lambda params: params.get('dtype') == torch.bfloat16 and params.get('size') == 10000)
-    @skipIfRocmVersionLessThan((5, 2))
     # imported 'tol' as 'xtol' to avoid aliasing in code above
     @toleranceOverride({torch.float16: xtol(atol=7e-1, rtol=2e-1),
                         torch.bfloat16: xtol(atol=1e1, rtol=2e-1)})
@@ -170,7 +167,6 @@ class TestMatmulCuda(InductorTestCase):
             self.cublas_addmm(size, dtype, True)
 
     @onlyCUDA
-    @skipIfRocmVersionLessThan((5, 2))
     @dtypes(torch.float16)
     # m == 4 chooses OUTPUT_TYPE reduction on H200
     # m == 8 chooses OUTPUT_TYPE reduction on A100
@@ -191,7 +187,6 @@ class TestMatmulCuda(InductorTestCase):
             torch.backends.cuda.matmul.allow_fp16_reduced_precision_reduction = orig_precision
 
     @onlyCUDA
-    @skipIfRocmVersionLessThan((5, 2))
     # imported 'tol' as 'xtol' to avoid aliasing in code above
     @toleranceOverride({torch.float16: xtol(atol=7e-1, rtol=2e-1),
                         torch.bfloat16: xtol(atol=1e1, rtol=2e-1)})
diff --git a/test/test_sparse_csr.py b/test/test_sparse_csr.py
index 90bd85e230f6..3f4729d36ee9 100644
--- a/test/test_sparse_csr.py
+++ b/test/test_sparse_csr.py
@@ -16,8 +16,7 @@ from torch.testing._internal.common_utils import \
      skipIfRocmVersionLessThan, IS_FBCODE, IS_REMOTE_GPU, suppress_warnings)
 from torch.testing._internal.common_device_type import \
     (ops, instantiate_device_type_tests, dtypes, OpDTypes, dtypesIfCUDA, onlyCPU, onlyCUDA, skipCUDAIfNoSparseGeneric,
-     precisionOverride, skipMeta, skipCUDAIf, skipCUDAIfRocm, skipCPUIfNoMklSparse, skipCUDAIfRocmVersionLessThan,
-     largeTensorTest)
+     precisionOverride, skipMeta, skipCUDAIf, skipCUDAIfRocm, skipCPUIfNoMklSparse, largeTensorTest)
 from torch.testing._internal.common_methods_invocations import \
     (op_db, sparse_csr_unary_ufuncs, ReductionOpInfo)
 from torch.testing._internal.common_cuda import TEST_CUDA
@@ -1492,8 +1491,6 @@ class TestSparseCSR(TestCase):
                 csr.matmul(bad_vec)
 
     @onlyCUDA
-    # hmm, the test passes ok on CUDA when Rocm is not available:
-    @skipCUDAIfRocmVersionLessThan((5, 2))
     @dtypes(torch.float32, torch.float64, torch.complex64, torch.complex128)
     def test_baddbmm(self, device, dtype):