improve DEBUG=3 [pr] (#9105)

This commit is contained in:
George Hotz
2025-02-15 18:44:56 +08:00
committed by GitHub
parent 41d143d27c
commit 81f5a7af7d
3 changed files with 8 additions and 6 deletions
+3 -3
View File
@@ -514,7 +514,7 @@ class Kernel:
for axis in to_upcast[::-1]: self.apply_opt(Opt(OptOps.UPCAST, axis, 0))
# potentially do more upcasts of non reduce axes based on a heuristic
upcasted_axis = set()
upcasted_axis: set[int] = set()
while resolve(prod(self.sts[0].shape[:self.first_reduce]) >= 1024):
xb_choices = []
for axis, upcast_amount in itertools.product(range(self.first_reduce), [3,4]): # consider all the non reduce axes, and a 3 or 4 reduce
@@ -669,8 +669,8 @@ class Kernel:
if DEBUG >= 3:
print(self.name)
if getenv("RAWAST"): print(self.ast)
print(modified_ast)
for i,(buf,st) in enumerate([(buf,st) for buf,st in zip(self.bufs, self.sts) if buf.op not in {Ops.CONST, Ops.VALID}]):
print(f"{i:2d}: {str(st.shape):25s} {str(buf.src[0].dtype).replace('dtypes.',''):20s}", st.real_strides())
print(self.applied_opts)
# verify AST matches the spec after applying opts
if __debug__: type_verify(list(modified_ast.toposort))