From c4b852a90908e073fa620a5093d023758a14f3ad Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@gmail.com>
Date: Sun, 1 Feb 2015 08:52:55 +0000
Subject: [PATCH] [X86] Flesh out more of the avx512erintrin.h file.

llvm-svn: 227719
---
 clang/lib/Headers/avx512erintrin.h | 82 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 79 insertions(+), 3 deletions(-)

diff --git a/clang/lib/Headers/avx512erintrin.h b/clang/lib/Headers/avx512erintrin.h
index 0785bce..54db9975 100644
--- a/clang/lib/Headers/avx512erintrin.h
+++ b/clang/lib/Headers/avx512erintrin.h
@@ -29,23 +29,61 @@
 
 
 // rsqrt28
-#define _mm512_rsqrt28_round_pd(A, R) __extension ({ \
+#define _mm512_rsqrt28_round_pd(A, R) __extension__ ({ \
   (__m512d)__builtin_ia32_rsqrt28pd_mask((__v8df)(__m512d)(A), \
                                          (__v8df)_mm512_setzero_pd(), \
                                          (__mmask8)-1, (R)); })
 
+#define _mm512_mask_rsqrt28_round_pd(S, M, A, R) __extension__ ({ \
+  (__m512d)__builtin_ia32_rsqrt28pd_mask((__v8df)(__m512d)(A), \
+                                         (__v8df)(__m512d)(S), \
+                                         (__mmask8)(M), (R)); })
+
+#define _mm512_maskz_rsqrt28_round_pd(M, A, R) __extension__ ({ \
+  (__m512d)__builtin_ia32_rsqrt28pd_mask((__v8df)(__m512d)(A), \
+                                         (__v8df)_mm512_setzero_pd(), \
+                                         (__mmask8)(M), (R)); })
+
+#define _mm512_rsqrt28_pd(A) \
+  _mm512_rsqrt28_round_pd(A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_mask_rsqrt28_pd(S, M, A) \
+  _mm512_mask_rsqrt28_round_pd(S, M, A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_maskz_rsqrt28_pd(M, A) \
+  _mm512_maskz_rsqrt28_round_pd(M, A, _MM_FROUND_CUR_DIRECTION)
+
 #define _mm512_rsqrt28_round_ps(A, R) __extension__ ({ \
   (__m512)__builtin_ia32_rsqrt28ps_mask((__v16sf)(__m512)(A), \
                                         (__v16sf)_mm512_setzero_ps(), \
                                         (__mmask16)-1, (R)); })
 
+#define _mm512_mask_rsqrt28_round_ps(S, M, A, R) __extension__ ({ \
+  (__m512)__builtin_ia32_rsqrt28ps_mask((__v16sf)(__m512)(A), \
+                                        (__v16sf)(__m512)(S), \
+                                        (__mmask16)(M), (R)); })
+
+#define _mm512_maskz_rsqrt28_round_ps(M, A, R) __extension__ ({ \
+  (__m512)__builtin_ia32_rsqrt28ps_mask((__v16sf)(__m512)(A), \
+                                        (__v16sf)_mm512_setzero_ps(), \
+                                        (__mmask16)(M), (R)); })
+
+#define _mm512_rsqrt28_ps(A) \
+  _mm512_rsqrt28_round_ps(A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_mask_rsqrt28_ps(S, M, A) \
+  _mm512_mask_rsqrt28_round_ps(S, M, A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_maskz_rsqrt28_ps(M, A) \
+  _mm512_maskz_rsqrt28_round_ps(M, A, _MM_FROUND_CUR_DIRECTION)
+
 #define _mm_rsqrt28_round_ss(A, B, R) __extension__ ({ \
   (__m128)__builtin_ia32_rsqrt28ss_mask((__v4sf)(__m128)(A), \
-                                        (__v4sf)(__m128(B), \
+                                        (__v4sf)(__m128)(B), \
                                         (__v4sf)_mm_setzero_ps(), \
                                         (__mmask8)-1, (R)); })
 
-#define _mm_rsqrt28_round_sd (A, B, R) __extension__ ({ \
+#define _mm_rsqrt28_round_sd(A, B, R) __extension__ ({ \
   (__m128d)__builtin_ia32_rsqrt28sd_mask((__v2df)(__m128d)(A), \
                                          (__v2df)(__m128d)(B), \
                                          (__v2df)_mm_setzero_pd(), \
@@ -58,11 +96,49 @@
                                        (__v8df)_mm512_setzero_pd(), \
                                        (__mmask8)-1, (R)); })
 
+#define _mm512_mask_rcp28_round_pd(S, M, A, R) __extension__ ({ \
+  (__m512d)__builtin_ia32_rcp28pd_mask((__v8df)(__m512d)(A), \
+                                       (__v8df)(__m512d)(S), \
+                                       (__mmask8)(M), (R)); })
+
+#define _mm512_maskz_rcp28_round_pd(M, A, R) __extension__ ({ \
+  (__m512d)__builtin_ia32_rcp28pd_mask((__v8df)(__m512d)(A), \
+                                       (__v8df)_mm512_setzero_pd(), \
+                                       (__mmask8)(M), (R)); })
+
+#define _mm512_rcp28_pd(A) \
+  _mm512_rcp28_round_pd(A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_mask_rcp28_pd(S, M, A) \
+  _mm512_mask_rcp28_round_pd(S, M, A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_maskz_rcp28_pd(M, A) \
+  _mm512_maskz_rcp28_round_pd(M, A, _MM_FROUND_CUR_DIRECTION)
+
 #define _mm512_rcp28_round_ps(A, R) __extension__ ({ \
   (__m512)__builtin_ia32_rcp28ps_mask((__v16sf)(__m512)(A), \
                                       (__v16sf)_mm512_setzero_ps(), \
                                       (__mmask16)-1, (R)); })
 
+#define _mm512_mask_rcp28_round_ps(S, M, A, R) __extension__ ({ \
+  (__m512)__builtin_ia32_rcp28ps_mask((__v16sf)(__m512)(A), \
+                                      (__v16sf)(__m512)(S), \
+                                      (__mmask16)(M), (R)); })
+
+#define _mm512_maskz_rcp28_round_ps(M, A, R) __extension__ ({ \
+  (__m512)__builtin_ia32_rcp28ps_mask((__v16sf)(__m512)(A), \
+                                      (__v16sf)_mm512_setzero_ps(), \
+                                      (__mmask16)(M), (R)); })
+
+#define _mm512_rcp28_ps(A) \
+  _mm512_rcp28_round_ps(A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_mask_rcp28_ps(S, M, A) \
+  _mm512_mask_rcp28_round_ps(S, M, A, _MM_FROUND_CUR_DIRECTION)
+
+#define _mm512_maskz_rcp28_ps(M, A) \
+  _mm512_maskz_rcp28_round_ps(M, A, _MM_FROUND_CUR_DIRECTION)
+
 #define _mm_rcp28_round_ss(A, B, R) __extension__ ({ \
   (__m128) __builtin_ia32_rcp28ss_mask ((__v4sf)(__m128)(A), \
                                         (__v4sf)(__m128)(B), \
-- 
2.7.4