Block a user
core/llvm_intrinsics: remove llvm.roundeven call from call_float_roundeven (issue 396)
I'm not sure if doing another manual implementation of this function is the right thing to do in the first place. Indeed there are often edge cases with NaN, denormal numbers, etc.
core/llvm_intrinsics: remove llvm.roundeven call from call_float_roundeven (issue 396)
Floating point stuff can be deceptive. Please look at a reference roundeven implementation e.g. from glibc.
core/llvm_intrinsics: remove llvm.roundeven call from call_float_roundeven (issue 396)
Well OK but will it return the same/correct results?
sb10q
deleted branch enhance/issue-149-ndarray/nd-indices from M-Labs/nac3
2024-06-20 12:50:33 +08:00
5bade81ddb
standalone: Add test for multidim array index with one index
0452e6de78
core: Fix codegen for tuple-index into ndarray
635c944c90
core: Fix type inference for tuple-index into ndarray
Implement/Fix support for tuple-indexing into ndarrays
Type Error for equality between a subscripted ndarray and a scalar
core/llvm_intrinsics: remove llvm.roundeven call from call_float_roundeven (issue 396)
Hmm, are there any other codebases doing this?
core: reduce code duplication in codegen/builtin_fns
d3cd2a8d99
artiq: Add support for generating RPC tag for ndarray
202a63274d
artiq: Implement pyty-to-ty conversion
76dd5191f5
artiq: Implement Python-to-LLVM conversion of ndarray
8d9df0a615
artiq: Fix ndarray class ID
07adfb2a18
standalone: Add *.ll to Gitignore list