[PATCH] D72798: [llvm][docs] LangRef for IR attribute `vector-function-abi-variants`.
Andrzej Warzynski via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Thu Jan 16 06:10:29 PST 2020
andwar added a comment.
@fpetrogalli Thank you for working on this - just a few nits from me!
@jdoerfert I also assume that we have already agreed on this. Happy to fall back to llvm-dev if you feel otherwise.
================
Comment at: llvm/docs/LangRef.rst:1852
+ This attribute can be attached to a CallInst to list the vector
+ function that can be used to vectorize calls to the scalar
+ function. The Attribute consists of a comma separated list of
----------------
`function` -> `functions`?
================
Comment at: llvm/docs/LangRef.rst:1853
+ function that can be used to vectorize calls to the scalar
+ function. The Attribute consists of a comma separated list of
+ mangled names. The syntax for the mangled names is as follows:
----------------
`scalar function` -> `scalar function that would otherwise we invoked by the call instruction`? Maybe that's too verbose.
================
Comment at: llvm/docs/LangRef.rst:1859
+ When present, the attribute informs the compiler that the function
+ ``<scalar_name>`` has a correspondent vector variant that can be
+ used to perform the concurrent invocation of ``<scalar_name>`` on
----------------
`correspondent` -> `corresponding` ?
================
Comment at: llvm/docs/LangRef.rst:1865
+ ``_ZGV<isa><mask><vlen><parameters>_<scalar_name>``. When present,
+ the optional ``(<vector_redirection>)`` informs the compiler that
+ a custom name is provided instead of the standard one
----------------
`the optional` -> `the optional name`?
================
Comment at: llvm/docs/LangRef.rst:1866
+ the optional ``(<vector_redirection>)`` informs the compiler that
+ a custom name is provided instead of the standard one
+ ``_ZGV<isa><mask><vlen><parameters>_<scalar_name>``. The
----------------
Why would one provide a custom name? Perhaps it's worth expanding here.
================
Comment at: llvm/docs/LangRef.rst:1868
+ ``_ZGV<isa><mask><vlen><parameters>_<scalar_name>``. The
+ definition or declaration of the variant must be present in the IR
+ Module. The signature of the vector variant is determined by the
----------------
`Declaration` is a _necessary_ condition, `definition` is a _sufficient_ condition. Wouldn't `declaration` cover both?
================
Comment at: llvm/docs/LangRef.rst:1879
+ X86 and Arm are used to represent the Internal LLVM Vector Isa
+ when ``<isa> = __LLVM__``.
+
----------------
This is a bit confusing - with ``<isa> = __LLVM__``, which token specification one should use? The one for X86 or Arm? Could you clarify what ``<isa> == __LLVM__`` means in practical terms?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D72798/new/
https://reviews.llvm.org/D72798
More information about the llvm-commits
mailing list