[submodule update] Bump cudnn-frontend to v0.6.1 (#1353)

* bump version

* add guard

* fix the cond

cudnn-frontend @ fa611998

-Subproject commit 7b83dba83fa31381aeca508d89aab94f4639ac6d
+Subproject commit fa611998a360cbabaa2dcc7c9859748144114fc0

setup.py

@@ -58,6 +58,13 @@ def append_nvcc_threads(nvcc_extra_args):
     return nvcc_extra_args
 
 
+def check_cudnn_version_and_warn(global_option: str, required_cudnn_version: int) -> bool:
+    green = torch.backends.cudnn.is_available() and torch.backends.cudnn.version() >= required_cudnn_version
+    if not green:
+        warnings.warn(f"Skip `{global_option}` as it requires cuDNN {required_cudnn_version} or later")
+    return green
+
+
 if not torch.cuda.is_available():
     # https://github.com/NVIDIA/apex/issues/486
     # Extension builds after https://github.com/pytorch/pytorch/pull/23408 attempt to query torch.cuda.get_device_capability(),
@@ -649,6 +656,7 @@ if "--fast_bottleneck" in sys.argv:
 if "--fused_conv_bias_relu" in sys.argv:
     sys.argv.remove("--fused_conv_bias_relu")
     raise_if_cuda_home_none("--fused_conv_bias_relu")
+    if check_cudnn_version_and_warn("--fused_conv_bias_relu", 8400):
         subprocess.run(["git", "submodule", "update", "--init", "apex/contrib/csrc/cudnn-frontend/"])
         ext_modules.append(
             CUDAExtension(