This revision was automatically updated to reflect the committed changes.
Closed by commit rL270499: [X86][SSE] Replace lossless i32/f32 to f64
conversion intrinsics with generic IR (authored by RKSimon).
Changed prior to commit:
http://reviews.llvm.org/D20528?vs=58146=58160#toc
Repository:
mkuper accepted this revision.
mkuper added a comment.
This revision is now accepted and ready to land.
> The existing llvm\test\CodeGen\X86\vec_fpext.ll and
> llvm\test\CodeGen\X86\vec_int_to_fp.ll already demonstrate the correct
> optimized DAG codegen using the same IR as output in the
>
RKSimon added a comment.
In http://reviews.llvm.org/D20528#437165, @mkuper wrote:
> Presumably, the fast-isel lowering of the IR pattern is already correct, and
> in any case, it isn't affected by this patch.
> I just want to make sure we don't regress the optimized DAG codegen - that
> is,
mkuper added a comment.
Presumably, the fast-isel lowering of the IR pattern is already correct, and in
any case, it isn't affected by this patch.
I just want to make sure we don't regress the optimized DAG codegen - that is,
it still produces the instruction we'd expect from the intrinsic (or
RKSimon added a comment.
In http://reviews.llvm.org/D20528#437117, @mkuper wrote:
> Sorry, I didn't intend to imply the rest of the llvm work is necessary for
> this to go in. Just that I'd be happier with this patch knowing that we have
> a regression test for doing the (shuffle + fpext, say)
RKSimon updated this revision to Diff 58146.
Repository:
rL LLVM
http://reviews.llvm.org/D20528
Files:
include/clang/Basic/BuiltinsX86.def
lib/Headers/avxintrin.h
lib/Headers/emmintrin.h
test/CodeGen/avx-builtins.c
test/CodeGen/builtins-x86.c
test/CodeGen/sse2-builtins.c
mkuper added a comment.
In http://reviews.llvm.org/D20528#437090, @RKSimon wrote:
> In http://reviews.llvm.org/D20528#436893, @mkuper wrote:
>
> > This looks right, but we may lose some end-to-end tests, since right now we
> > have a clang-level test that checks the builtin is lowered to the
>
RKSimon added a comment.
In http://reviews.llvm.org/D20528#436893, @mkuper wrote:
> This looks right, but we may lose some end-to-end tests, since right now we
> have a clang-level test that checks the builtin is lowered to the intrinsic,
> and (hopefully) a CG-level test that the intrinsic is
RKSimon created this revision.
RKSimon added reviewers: qcolombet, craig.topper, mkuper, andreadb, spatel.
RKSimon added a subscriber: cfe-commits.
RKSimon set the repository for this revision to rL LLVM.
Both the (V)CVTDQ2PD(Y) (i32 to f64) and (V)CVTPS2PD(Y) (f32 to f64) conversion