[FRONTEND] Remove cache key from metadata (#2082)

python/triton/compiler/compiler.py

@@ -519,7 +519,6 @@ def compile(fn, **kwargs):
 
     # Add device type to meta information
     metadata["device_type"] = device_type
-    metadata["cache_key"] = fn_cache_manager.key
 
     first_stage = list(stages.keys()).index(ext)
     asm = dict()
@@ -629,7 +628,6 @@ class CompiledKernel:
         if "tensormaps_info" in metadata:
             self.tensormaps_info = metadata["tensormaps_info"]
         self.constants = metadata["constants"]
-        self.cache_key = metadata["cache_key"]
         self.device_type = metadata["device_type"]
         self.device_backend = get_backend(self.device_type) if self.device_type not in ["cuda", "hip"] else None
         # initialize asm dict
@@ -682,7 +680,7 @@ class CompiledKernel:
             # tuple for hashable
             args_ptr = tuple([arg.data_ptr() if hasattr(arg, 'data_ptr') else arg for arg in args])
             for i, e in enumerate(self.tensormaps_info):
-                args_with_tma.append(CompiledKernel.tensormap_manager[(self.cache_key, e, args_ptr)])
+                args_with_tma.append(CompiledKernel.tensormap_manager[(e, args_ptr)])
         return args_with_tma
 
     def __getitem__(self, grid):

python/triton/compiler/utils.py

@@ -283,7 +283,7 @@ class TensorMapManager:
         if key in self.tensormaps_device:
             return int(self.tensormaps_device[key])
         else:
-            (cache_key, e, args) = key
+            (e, args) = key
             t_tensormap = e.tensormap(args)
             TENSORMAP_SIZE_IN_BYTES = 128
             t_tensormap_device = driver.utils.cuMemAlloc(TENSORMAP_SIZE_IN_BYTES)