[Triton] [Beta] Add latency to the tt.load frontend as a no-op

Summary: Adds latency as a no-op to the tt.load frontend to ensure kernel compatability. This will be replaced with the proper changes from fb experimental once this SEV is resolved.

Reviewed By: dhruvak3

Differential Revision: D85875953

fbshipit-source-id: 155f3c267aa4fcc7101bbea0834c304cb40728d0

Author: njriasan

python/triton/language/core.py

@@ -2064,7 +2064,7 @@ def dot_scaled(lhs, lhs_scale, lhs_format, rhs, rhs_scale, rhs_format, acc=None,
 
 @builtin
 def load(pointer, mask=None, other=None, boundary_check=(), padding_option="", cache_modifier="", eviction_policy="",
-         volatile=False, _semantic=None):
+         volatile=False, latency=None, _semantic=None):
     """
     Return a tensor of data whose values are loaded from memory at location defined by `pointer`: