from:"Stefan Teleman via Phabricator via cfe\-commits"

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2018-10-31 Thread Stefan Teleman via Phabricator via cfe-commits

steleman created this revision.
steleman added a reviewer: joel_k_jones.
Herald added subscribers: kristina, kristof.beyls, javed.absar.

This changeset adds the required Builtins for enabling AArch64 vectorization of 
libm trigonometry functions via SLEEF: http://sleef.org/.

- A new argument is added to -fveclib=: SLEEF.
- A number of Builtins that were previously unimplemented are now enabled.

This changeset depends on https://reviews.llvm.org/D53927.

List of SLEEF vectorized trigonometry functions:

acos
asin
atan
atanh
cos
cosh
exp
exp2
exp10
lgamma
log10
log2
log
pow
sin
sinh
sqrt
tan
tanh
tgamma


Repository:
  rC Clang

https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.h
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: include/clang/Frontend/CodeGenOptions.h
===
--- include/clang/Frontend/CodeGenOptions.h
+++ include/clang/Frontend/CodeGenOptions.h
@@ -54,7 +54,8 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+SLEEF   // SLEEF - SIMD Library for Evaluating Elementary Functions.
   };
 
 
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1388,7 +1388,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,SLEEF,none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -196,6 +196,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1133,6 +1136,10 @@
 LIBBUILTIN(exp2f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(exp2l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(expm1, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1370,10 +1377,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -557,6 +557,8 @@
   Opts.setVecLib(CodeGenOptions::Accelerate);
 else if (Name == "SVML")
   Opts.setVecLib(CodeGenOptions::SVML);
+else if (Name == "SLEEF")
+  Opts.setVecLib(CodeGenOptions::SLEEF);
 else if (Name == "none")
   Opts.setVecLib(CodeGenOptions::NoLibrary);
 else
Index: lib/CodeGen/BackendUtil.cpp
===
--- lib/CodeGen/BackendUtil.cpp
+++ lib/CodeGen/BackendUtil.cpp
@@ -344,6 +344,9 @@
   case CodeGenOptions::SVML:
 TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SVML);
 break;
+  case CodeGenOptions::SLEEF:
+TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SLEEF);
+break;
   default:
 break;
   }
Index: lib/CodeGen/CGBuiltin.cpp
===
--- lib/CodeGen/CGBuiltin.cpp
+++ lib/CodeGen/CGBuiltin.cpp
@@ -1306,14 +1306,70 @@

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2018-11-05 Thread Stefan Teleman via Phabricator via cfe-commits

steleman updated this revision to Diff 172638.
steleman added a comment.

- changed the -fveclib= argument value to 'sleefgnuabi'.
- added atan2 and pow.
- spreadsheet with comparison between libm and sleef is here:

https://docs.google.com/spreadsheets/d/1lcpESCnuzEoTl_XHBqE9FLL0tXJB_tZGR8yciCx1yjg/edit?usp=sharing

- comprehensive test case in C with timings is here:

https://drive.google.com/open?id=1PGKRUdL29_ANoYebOo3Q59syhKp_mNSj


https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Driver/Options.td
  include/clang/Frontend/CodeGenOptions.h
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: include/clang/Frontend/CodeGenOptions.h
===
--- include/clang/Frontend/CodeGenOptions.h
+++ include/clang/Frontend/CodeGenOptions.h
@@ -54,7 +54,8 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+SLEEFGNUABI // SLEEF - SIMD Library for Evaluating Elementary Functions.
   };
 
 
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1388,7 +1388,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,sleefgnuabi,none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -196,6 +196,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1133,6 +1136,10 @@
 LIBBUILTIN(exp2f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(exp2l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(expm1, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1370,10 +1377,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -558,6 +558,8 @@
   Opts.setVecLib(CodeGenOptions::Accelerate);
 else if (Name == "SVML")
   Opts.setVecLib(CodeGenOptions::SVML);
+else if (Name == "sleefgnuabi")
+  Opts.setVecLib(CodeGenOptions::SLEEFGNUABI);
 else if (Name == "none")
   Opts.setVecLib(CodeGenOptions::NoLibrary);
 else
Index: lib/CodeGen/BackendUtil.cpp
===
--- lib/CodeGen/BackendUtil.cpp
+++ lib/CodeGen/BackendUtil.cpp
@@ -344,6 +344,9 @@
   case CodeGenOptions::SVML:
 TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SVML);
 break;
+  case CodeGenOptions::SLEEFGNUABI:
+TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SLEEFGNUABI);
+break;
   default:
 break;
   }
Index: lib/CodeGen/CGBuiltin.cpp
===
--- lib/CodeGen/CGBuiltin.cpp
+++ lib/CodeGen/CGBuiltin.cpp
@@ -1333,14 +1333,78 @@
 case Builtin::BI__builtin_copysignf128:
   return RValue::get(emitBinaryBuiltin(*this, E, Intrinsic::copysign));
 
+case Builtin::BIacos:
+case Builti

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2019-02-06 Thread Stefan Teleman via Phabricator via cfe-commits

steleman updated this revision to Diff 185578.
steleman added a comment.
Herald added a project: clang.

This is a small - but signifcant - update to the original changeset.

What's new in this update:

${top_srcdir}/lib/Frontend/CompilerInvocation.cpp:

When -fveclib=SLEEF was passed on compile-line, neither clang nor LLVM emit any 
warnings if SLEEF is not supported on the Target architecture. SLEEF is 
currently supported only on AArch64. So, for other ISA's, -fveclib=SLEEF would 
be quietly ignored, without any hints.

This new version of the changeset adds a diagnostic if -fveclib=SLEEF is called 
on an ISA that doesn't (yet) support SLEEF.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Basic/CodeGenOptions.h
  include/clang/Driver/Options.td
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1407,7 +1407,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,SLEEF,none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/CodeGenOptions.h
===
--- include/clang/Basic/CodeGenOptions.h
+++ include/clang/Basic/CodeGenOptions.h
@@ -53,7 +53,9 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+SLEEF   // SLEEF - SIMD Library for Evaluating Elementary Functions.
+// (Experimental).
   };
 
 
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -197,6 +197,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1146,6 +1149,10 @@
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(fdim, "ddd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(fdimf, "fff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(fdiml, "LdLdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1379,10 +1386,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -670,7 +670,14 @@
   Opts.setVecLib(CodeGenOptions::Accelerate);
 else if (Name == "SVML")
   Opts.setVecLib(CodeGenOptions::SVML);
-else if (Name == "none")
+else if (Name == "SLEEF") {
+  if (Triple.getArch() == llvm::Triple::aarch64 ||
+  Triple.getArch() == llvm::Triple::aarch64_be)
+Opts.setVecLib(CodeGenOptions::SLEEF);
+  else
+Diags.Report(diag::err_drv_invalid_value) << A->getAsString(Args)
+  << Name;
+} else if (Name == "none")
   Opts.setVecLib(CodeGenOptions::NoLibrary);
 else
   Diags.Report(diag::err_drv_invalid_value) << A->getAsString(Args) << Name;
Index: lib/CodeGen/BackendUtil.cpp
===
--- lib/CodeGen/BackendUtil.cpp
+++ lib/CodeGen/BackendUtil.cpp

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2019-02-13 Thread Stefan Teleman via Phabricator via cfe-commits

steleman added a comment.

Hi Renato,

Thank you very much for the comments.

I will create a new changeset incorporating your comments, and then re-submit.

I will also add a Diagnostic for SVML and X86/X86_64.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2019-02-14 Thread Stefan Teleman via Phabricator via cfe-commits

steleman updated this revision to Diff 186963.
steleman added a comment.

Addressed comments from Renato:

- Diagnostic for unsupported argument to -fveclib= is now:

  error: unsupported option '' for target ''



- Added check and diagnostic for SVML on non-Intel ISA's.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Basic/CodeGenOptions.h
  include/clang/Driver/Options.td
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1407,7 +1407,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,SLEEF,none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/CodeGenOptions.h
===
--- include/clang/Basic/CodeGenOptions.h
+++ include/clang/Basic/CodeGenOptions.h
@@ -53,7 +53,9 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+SLEEF   // SLEEF - SIMD Library for Evaluating Elementary Functions.
+// (Experimental).
   };
 
 
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -197,6 +197,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1147,6 +1150,10 @@
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(fdim, "ddd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(fdimf, "fff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(fdiml, "LdLdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1380,10 +1387,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/Frontend/CompilerInvocation.cpp
===
--- lib/Frontend/CompilerInvocation.cpp
+++ lib/Frontend/CompilerInvocation.cpp
@@ -668,9 +668,21 @@
 StringRef Name = A->getValue();
 if (Name == "Accelerate")
   Opts.setVecLib(CodeGenOptions::Accelerate);
-else if (Name == "SVML")
-  Opts.setVecLib(CodeGenOptions::SVML);
-else if (Name == "none")
+else if (Name == "SVML") {
+  if (Triple.getArch() == llvm::Triple::x86 ||
+  Triple.getArch() == llvm::Triple::x86_64)
+Opts.setVecLib(CodeGenOptions::SVML);
+  else
+Diags.Report(diag::err_drv_unsupported_opt_for_target)
+  << Name << Triple.getArchName();
+} else if (Name == "SLEEF") {
+  if (Triple.getArch() == llvm::Triple::aarch64 ||
+  Triple.getArch() == llvm::Triple::aarch64_be)
+Opts.setVecLib(CodeGenOptions::SLEEF);
+  else
+Diags.Report(diag::err_drv_unsupported_opt_for_target)
+  << Name << Triple.getArchName();
+} else if (Name == "none")
   Opts.setVecLib(CodeGenOptions::NoLibrary);
 else
   Diags.Report(diag::err_drv_invalid_value) << A->getAsString(Args) << Name;
Index: lib/CodeGen/BackendUtil.cpp
===
--- lib/CodeGen/BackendUtil.cpp
+++ lib/CodeGen/BackendUt

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2018-12-17 Thread Stefan Teleman via Phabricator via cfe-commits

steleman updated this revision to Diff 178477.
steleman added a comment.
Herald added a subscriber: arphaman.

Updated version of this changeset/patch, as per Renato's latest comments from 
D53927  (https://reviews.llvm.org/D53927).

New benchmark comparing SLEEF performance when built with GCC 8.2 vs. clang ToT 
from 12/15/2018:

https://docs.google.com/spreadsheets/d/1QgdJjgCTFA9I2_5MvJU0uR4JT4knoW5xu-ujxszRbd0/edit?usp=sharing


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Basic/CodeGenOptions.h
  include/clang/Driver/Options.td
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp
  lib/Index/CommentToXML.cpp

Index: include/clang/Basic/CodeGenOptions.h
===
--- include/clang/Basic/CodeGenOptions.h
+++ include/clang/Basic/CodeGenOptions.h
@@ -54,7 +54,9 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+// SLEEF is experimental for now.
+SLEEF   // SLEEF - SIMD Library for Evaluating Elementary Functions.
   };
 
 
Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1404,7 +1404,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,SLEEF (experimental),none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -196,6 +196,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1137,6 +1140,10 @@
 LIBBUILTIN(exp2f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(exp2l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(expm1, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1374,10 +1381,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/Index/CommentToXML.cpp
===
--- lib/Index/CommentToXML.cpp
+++ lib/Index/CommentToXML.cpp
@@ -227,27 +227,27 @@
   { }
 
   // Inline content.
-  void visitTextComment(const TextComment *C);
-  void visitInlineCommandComment(const InlineCommandComment *C);
-  void visitHTMLStartTagComment(const HTMLStartTagComment *C);
-  void visitHTMLEndTagComment(const HTMLEndTagComment *C);
+  void VisitTextComment(const TextComment *C);
+  void VisitInlineCommandComment(const InlineCommandComment *C);
+  void VisitHTMLStartTagComment(const HTMLStartTagComment *C);
+  void VisitHTMLEndTagComment(const HTMLEndTagComment *C);
 
   // Block content.
-  void visitParagraphComment(const ParagraphComment *C);
-  void visitBlockCommandComment(const BlockCommandComment *C);
-  void visitParamCommandComment(const ParamCommandComment *C);
-  void visitTParamCommandComment(const TParamCommandComment *C);
-  void visitVerbatimBlockComment(const VerbatimBlockComment *C);
-  void visitVerbatimBlockLineComment(const VerbatimBlockLineComment *C);
-  void visitVerbatimLineComment

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2018-12-19 Thread Stefan Teleman via Phabricator via cfe-commits

steleman updated this revision to Diff 178884.
steleman added a comment.

Removed spurious patch for an unrelated change.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928

Files:
  include/clang/Basic/Builtins.def
  include/clang/Basic/CodeGenOptions.h
  include/clang/Driver/Options.td
  lib/CodeGen/BackendUtil.cpp
  lib/CodeGen/CGBuiltin.cpp
  lib/Frontend/CompilerInvocation.cpp

Index: include/clang/Driver/Options.td
===
--- include/clang/Driver/Options.td
+++ include/clang/Driver/Options.td
@@ -1404,7 +1404,7 @@
   Group, Flags<[CC1Option]>,
   HelpText<"Disables an experimental new pass manager in LLVM.">;
 def fveclib : Joined<["-"], "fveclib=">, Group, Flags<[CC1Option]>,
-HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,none">;
+HelpText<"Use the given vector functions library">, Values<"Accelerate,SVML,SLEEF (experimental),none">;
 def fno_lax_vector_conversions : Flag<["-"], "fno-lax-vector-conversions">, Group,
   HelpText<"Disallow implicit conversions between vectors with a different number of elements or different element types">, Flags<[CC1Option]>;
 def fno_merge_all_constants : Flag<["-"], "fno-merge-all-constants">, Group,
Index: include/clang/Basic/CodeGenOptions.h
===
--- include/clang/Basic/CodeGenOptions.h
+++ include/clang/Basic/CodeGenOptions.h
@@ -54,7 +54,9 @@
   enum VectorLibrary {
 NoLibrary,  // Don't use any vector library.
 Accelerate, // Use the Accelerate framework.
-SVML// Intel short vector math library.
+SVML,   // Intel short vector math library.
+// SLEEF is experimental for now.
+SLEEF   // SLEEF - SIMD Library for Evaluating Elementary Functions.
   };
 
 
Index: include/clang/Basic/Builtins.def
===
--- include/clang/Basic/Builtins.def
+++ include/clang/Basic/Builtins.def
@@ -196,6 +196,9 @@
 BUILTIN(__builtin_exp2 , "dd"  , "Fne")
 BUILTIN(__builtin_exp2f, "ff"  , "Fne")
 BUILTIN(__builtin_exp2l, "LdLd", "Fne")
+BUILTIN(__builtin_exp10 , "dd"  , "Fne")
+BUILTIN(__builtin_exp10f, "ff"  , "Fne")
+BUILTIN(__builtin_exp10l, "LdLd", "Fne")
 BUILTIN(__builtin_expm1 , "dd", "Fne")
 BUILTIN(__builtin_expm1f, "ff", "Fne")
 BUILTIN(__builtin_expm1l, "LdLd", "Fne")
@@ -1137,6 +1140,10 @@
 LIBBUILTIN(exp2f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(exp2l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
 
+LIBBUILTIN(exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
+LIBBUILTIN(exp10l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
+
 LIBBUILTIN(expm1, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1f, "ff", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(expm1l, "LdLd", "fne", "math.h", ALL_LANGUAGES)
@@ -1374,10 +1381,6 @@
 LIBBUILTIN(__tanpi, "dd", "fne", "math.h", ALL_LANGUAGES)
 LIBBUILTIN(__tanpif, "ff", "fne", "math.h", ALL_LANGUAGES)
 
-// Similarly, __exp10 is OS X only
-LIBBUILTIN(__exp10, "dd", "fne", "math.h", ALL_LANGUAGES)
-LIBBUILTIN(__exp10f, "ff", "fne", "math.h", ALL_LANGUAGES)
-
 // Blocks runtime Builtin math library functions
 LIBBUILTIN(_Block_object_assign, "vv*vC*iC", "f", "Blocks.h", ALL_LANGUAGES)
 LIBBUILTIN(_Block_object_dispose, "vvC*iC", "f", "Blocks.h", ALL_LANGUAGES)
Index: lib/CodeGen/BackendUtil.cpp
===
--- lib/CodeGen/BackendUtil.cpp
+++ lib/CodeGen/BackendUtil.cpp
@@ -347,6 +347,9 @@
   case CodeGenOptions::SVML:
 TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SVML);
 break;
+  case CodeGenOptions::SLEEF: // Experimental for now.
+TLII->addVectorizableFunctionsFromVecLib(TargetLibraryInfoImpl::SLEEF);
+break;
   default:
 break;
   }
Index: lib/CodeGen/CGBuiltin.cpp
===
--- lib/CodeGen/CGBuiltin.cpp
+++ lib/CodeGen/CGBuiltin.cpp
@@ -1501,6 +1501,62 @@
 case Builtin::BI__builtin_copysignf128:
   return RValue::get(emitBinaryBuiltin(*this, E, Intrinsic::copysign));
 
+case Builtin::BIacos:
+case Builtin::BIacosf:
+case Builtin::BIacosl:
+case Builtin::BI__builtin_acos:
+case Builtin::BI__builtin_acosf:
+case Builtin::BI__builtin_acosl:
+  return RValue::get(emitUnaryBuiltin(*this, E, Intrinsic::acos));
+
+case Builtin::BIacosh:
+case Builtin::BIacoshf:
+case Builtin::BIacoshl:
+case Builtin::BI__builtin_acosh:
+case Builtin::BI__builtin_acoshf:
+case Builtin::BI__builtin_acoshl:
+  return RValue::get(emitUnaryBuiltin(*this, E, Intrinsic::acosh));
+
+case Builtin::BIasin:
+case Builtin::BIasinf:
+case Builtin::BIasinl:
+case Builtin::BI__builtin_asin:
+case Builtin::BI__builtin_asinf:
+case Builtin::BI_

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

2019-01-10 Thread Stefan Teleman via Phabricator via cfe-commits

steleman added a comment.

Ping!

Yes, I know, everyone was away for the holidays. :-)

Could someone please take a look at this. We'd like to make sure everything is 
OK on this side before D53927  can proceed.

Yes, it's a bit of a circular dependency in that sense.

This changeset doesn't cover any of the problems with OpenMP and conflicting 
pragmas and semantics described above by Renato. This is just the clang side of 
straightforward mapping of SLEEF vectorized function calls.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D53928/new/

https://reviews.llvm.org/D53928



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

[PATCH] D53928: Enable builtins necessary for SLEEF [AArch64] vectorized trigonometry libm functions

8 matches

Site Navigation

Mail list logo

Footer information