File 0001-Fix-signedness-for-ARM-Neon-mask-type.patch of Package netgen

Overview Repositories Revisions Requests Users Attributes Meta

File 0001-Fix-signedness-for-ARM-Neon-mask-type.patch of Package netgen (Revision f95e58af174161f038eb0f13d74b2996)

Currently displaying revision f95e58af174161f038eb0f13d74b2996 , Show latest

From 5fce0b48b8691c4b359ec56c792087d33a9e4b81 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Stefan=20Br=C3=BCns?= <stefan.bruens@rwth-aachen.de>
Date: Thu, 27 Jan 2022 12:40:39 +0100
Subject: [PATCH] Fix signedness for ARM Neon mask type

vbslq_f64 and vandq_u64 both require uint64x2_t types as mask arguments,
and the Neon intrinsics do not allow for implicit conversion.

Fixes build errors with current GCC 11.2.1:

/home/abuild/rpmbuild/BUILD/netgen-6.2.2105/libsrc/core/simd_arm64.hpp:171:29: error: cannot convert '__Int64x2_t' to 'uint64x2_t'
 171 |     return vandq_u64 (a.Data(), b.Data());
---
 libsrc/core/simd_arm64.hpp | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/libsrc/core/simd_arm64.hpp b/libsrc/core/simd_arm64.hpp
index f2572d34..bcc0bb1d 100644
--- a/libsrc/core/simd_arm64.hpp
+++ b/libsrc/core/simd_arm64.hpp
@@ -14,9 +14,10 @@ namespace ngcore
       mask[1] = i > 1 ? -1 : 0;
     }
 
-    SIMD (bool i0, bool i1) { mask[0] = i0 ? -1:0; mask[1] = i1 ? -1 : 0; }
+    SIMD (bool i0, bool i1) { mask[0] = i0 ? -1 : 0; mask[1] = i1 ? -1 : 0; }
     SIMD (SIMD<mask64,1> i0, SIMD<mask64,1> i1) { mask[0] = i0[0]; mask[1] = i1[0]; }
-    SIMD (float64x2_t _data) : mask{_data} { }
+    // SIMD (float64x2_t _data) : mask{_data} { }
+    SIMD (int64x2_t _data) : mask{_data} { }
     auto Data() const { return mask; }
     static constexpr int Size() { return 2; }
     // static NETGEN_INLINE SIMD<mask64, 2> GetMaskFromBits (unsigned int i);
@@ -159,7 +160,8 @@ namespace ngcore
   NETGEN_INLINE SIMD<double,2> If (SIMD<mask64,2> a, SIMD<double,2> b, SIMD<double,2> c)
   {
     // return { a[0] ? b[0] : c[0], a[1] ? b[1] : c[1] };
-    return vbslq_f64(a.Data(), b.Data(), c.Data());
+    uint64x2_t mask = vreinterpretq_u64_s64(a.Data());
+    return vbslq_f64(mask, b.Data(), c.Data());
   }
   NETGEN_INLINE SIMD<int64_t,2> If (SIMD<mask64,2> a, SIMD<int64_t,2> b, SIMD<int64_t,2> c)
   {
@@ -168,7 +170,10 @@ namespace ngcore
 
   NETGEN_INLINE SIMD<mask64,2> operator&& (SIMD<mask64,2> a, SIMD<mask64,2> b)
   {
-    return vandq_u64 (a.Data(), b.Data());
+    uint64x2_t m1 = vreinterpretq_u64_s64(a.Data());
+    uint64x2_t m2 = vreinterpretq_u64_s64(b.Data());
+    uint64x2_t res = vandq_u64 (m1, m2);
+    return vreinterpretq_s64_u64(res);
   }
   
 }
-- 
2.34.1

Places

File 0001-Fix-signedness-for-ARM-Neon-mask-type.patch of Package netgen (Revision f95e58af174161f038eb0f13d74b2996)

Places