LLVM Weekly - #436, May 9th 2022
If you prefer, you can read a HTML version of this email at http://llvmweekly.org/issue/436.
Welcome to the four hundred and thirty-sixth issue of LLVM Weekly, a weekly newsletter (published every Monday) covering developments in LLVM, Clang, and related projects. LLVM Weekly is brought to you by Alex Bradbury. Subscribe to future issues at http://llvmweekly.org and pass it on to anyone else you think may be interested. Please send any tips or feedback to asb@asbradbury.org, or @llvmweekly or @asbradbury on Twitter.
News and articles from around the web
The videos from the 2022 LLVM Performance Workshop at CGO are now available.
The 2022 Clang-built Linux meetup has opened its call for papers. It will take place 10-11 September in Dublin, Ireland.
GCC 12.1 was released.
On the forums
-
Orlando Cazalet-Hyams proposed an RFC on assignment tracking - a better way of specifying variable locations in IR.
-
Tanya Lattner shared that a code of conduct transparency report has been published.
-
Artem Belevich sought opinions on conditional dependencies on proprietary tools (e.g. ptxas, Nvidia’s proprieatary assembler).
-
Andrzej Warzynski is seeking input on an RFC about the coding style in the Flang driver, which currently uses a mixture of LLVM’s standard style as of 2 years ago and MLIR’s style.
-
Alexandros Lamprineas provided an update on function specialisation.
-
Mehdi Amini proposes updating the MLIR developer policy on the use of recursion.
-
Csonkás Kristóf Gyula suggests adding a baremetal/none OS type.
-
River Riddle shared a public service announcement on verification of SameOperandsAndResultType in MLIR being tightened.
LLVM commits
-
Variable length instruction support was added to the decoder generator. 28e850a.
-
Initial DXContainer parsing support was added. 4070aa0.
-
Documentation on
CMAKE_BUILD_TYPE
was improved. 30e8796. -
New passes were introduced to add SPIR-V specific intrinsics. 0098f2a.
-
The SLP (superword-level parallelism) Vectorizer had its reductions analyses improved. 7ea03f0.
-
Native CPU detection, a scheduling model, and other definitions were added for the Ampere1 core. 64816e6, 7e02bc5.
Clang commits
-
__builtin_dump_struct
was reimplemented, and gained a number of new features. c4f95ef. -
The clang pseudo parser gained an implementation of the standard GLR parsing algorithm and a fuzzer. 9f38da2, 1616bd9.
-
Exploded graph dumps and CFG dumps from the static analyzer are now enabled in release builds. 5ce7050, 29dff0d.
-
-Wimplicit-int
now defaults to an error in C99 and later. 2cb2cd2.