Large integers are generated in Circt commonly which exceed 4kbits. This aligns the maximum bitwidth in MLIR and LLVM.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
mlir/include/mlir/IR/BuiltinTypes.h | ||
---|---|---|
142 | How did you come up with this number? |
LLVM uses "MAX_INT_BITS = (1<<24)-1". Unless there is a reason to be different, it would probably be good to align with that.
Nit: on review description, this is updating the maximum _bitwidth_ of integer types, not maximum integer size (i4096 can represent integers up to 1.04e+1233)
mlir/include/mlir/IR/BuiltinTypes.h | ||
---|---|---|
142 | Can you just also document that this is aligned with LLVM? |
Can you please add a proper commit summary with the rationale (basically whatever is in the discussion now)? Also, as @jpienaar points out, the commit title isn't really accurate.
Update the maximum integer bitwidth in MLIR.
Large integers are generated in Circt commonly which exceed 4k. This aligns MLIR's width to llvm.
mlir/include/mlir/IR/BuiltinTypes.h | ||
---|---|---|
142 |
How did you come up with this number?