[libclc] 1509b46 - [libclc] Improve nextafter behaviour around zero (#127469)

[via cfe-commits]

Author: Fraser Cormack
Date: 2025-02-19T10:24:12Z
New Revision: 1509b46ea5617402634270a01e840356ef935436

URL: https://github.com/llvm/llvm-project/commit/1509b46ea5617402634270a01e840356ef935436
DIFF: https://github.com/llvm/llvm-project/commit/1509b46ea5617402634270a01e840356ef935436.diff

LOG: [libclc] Improve nextafter behaviour around zero (#127469)

This commit improves the behaviour of (__clc_)nextafter around zero.
Specifically, the nextafter value of very small negative numbers in the
positive direction is now negative zero. Previously we'd return positive
zero.

This behaviour is not required as far as OpenCL is concerned: at least,
the CTS isn't testing for it. However, this change does bring our
implementation into bit-equivalence with (libstdc++'s implementation of)
std::nextafter, tested on all possible values of 32-bit float towards
both positive and negative INFINITY.

Furthermore, since the implementation of libclc's floating-point 'rtp'
and 'rtz' conversions use __clc_nextafter, the previous behaviour was
resulting in CTS validation issues. For example, when converting float
-0x1.000002p-25 to half, rounding towards zero or positive infinity,
nextafter was returning +0.0, whereas the correct conversion requires us
to return -0.0.

We could work around this issue in the conversion functions, but since
the change to nextafter is small enough and the behaviour around zero
matches libstdc++, the fix feels at home there.

This commit also converts several variables to unsigned types to avoid
undefined behaviour surrounding signed underflow on the subtractions.
It also converts some variables to be kept in floating-point types, using
fabs to get the absolute value rather than by bit-hacking.

Added: 
    

Modified: 
    libclc/clc/lib/generic/math/clc_nextafter.cl

Removed: 
    


################################################################################
diff  --git a/libclc/clc/lib/generic/math/clc_nextafter.cl b/libclc/clc/lib/generic/math/clc_nextafter.cl
index 58125485bf684..f77f3647d5985 100644
--- a/libclc/clc/lib/generic/math/clc_nextafter.cl
+++ b/libclc/clc/lib/generic/math/clc_nextafter.cl
@@ -1,5 +1,6 @@
 #include <clc/clcmacro.h>
 #include <clc/internal/clc.h>
+#include <clc/math/clc_fabs.h>
 #include <clc/relational/clc_isnan.h>
 
 // This file provides OpenCL C implementations of __clc_nextafter for
@@ -12,21 +13,30 @@
                                                     FLOAT_TYPE y) {            \
     const UINT_TYPE sign_bit = (UINT_TYPE)1                                    \
                                << (sizeof(INT_TYPE_SCALAR) * 8 - 1);           \
-    const UINT_TYPE sign_bit_mask = sign_bit - (UINT_TYPE)1;                   \
-    INT_TYPE ix = CLC_AS_TYPE(INT_TYPE)(x);                                    \
-    UINT_TYPE ax = CLC_AS_TYPE(UINT_TYPE)(ix) & sign_bit_mask;                 \
-    INT_TYPE mx = CLC_AS_TYPE(INT_TYPE)(sign_bit) - ix;                        \
-    mx = CLC_AS_TYPE(INT_TYPE)(ix) < (INT_TYPE)0 ? mx : ix;                    \
-    INT_TYPE iy = CLC_AS_TYPE(INT_TYPE)(y);                                    \
-    UINT_TYPE ay = CLC_AS_TYPE(UINT_TYPE)(iy) & sign_bit_mask;                 \
-    INT_TYPE my = CLC_AS_TYPE(INT_TYPE)(sign_bit) - iy;                        \
-    my = iy < (INT_TYPE)0 ? my : iy;                                           \
+    UINT_TYPE ix = CLC_AS_TYPE(UINT_TYPE)(x);                                  \
+    FLOAT_TYPE absx = __clc_fabs(x);                                           \
+    UINT_TYPE mxu = sign_bit - ix;                                             \
+    INT_TYPE mx = CLC_AS_TYPE(INT_TYPE)(mxu);                                  \
+    mx = CLC_AS_TYPE(INT_TYPE)(ix) < (INT_TYPE)0 ? mx                          \
+                                                 : CLC_AS_TYPE(INT_TYPE)(ix);  \
+    UINT_TYPE iy = CLC_AS_TYPE(UINT_TYPE)(y);                                  \
+    FLOAT_TYPE absy = __clc_fabs(y);                                           \
+    UINT_TYPE myu = sign_bit - iy;                                             \
+    INT_TYPE my = CLC_AS_TYPE(INT_TYPE)(myu);                                  \
+    my = CLC_AS_TYPE(INT_TYPE)(iy) < (INT_TYPE)0 ? my                          \
+                                                 : CLC_AS_TYPE(INT_TYPE)(iy);  \
     INT_TYPE t = mx + (mx < my ? (INT_TYPE)1 : (INT_TYPE)-1);                  \
-    INT_TYPE r = CLC_AS_TYPE(INT_TYPE)(sign_bit) - t;                          \
-    r = t < (INT_TYPE)0 ? r : t;                                               \
+    UINT_TYPE r = sign_bit - CLC_AS_TYPE(UINT_TYPE)(t);                        \
+    r = (t < (INT_TYPE)0 || (t == (INT_TYPE)0 && mx < my))                     \
+            ? r                                                                \
+            : CLC_AS_TYPE(UINT_TYPE)(t);                                       \
     r = __clc_isnan(x) ? ix : r;                                               \
-    r = __clc_isnan(y) ? CLC_AS_TYPE(INT_TYPE)(iy) : r;                        \
-    r = ((ax | ay) == (UINT_TYPE)0 || ix == iy) ? iy : r;                      \
+    r = __clc_isnan(y) ? iy : r;                                               \
+    r = ((CLC_AS_TYPE(UINT_TYPE)(absx) | CLC_AS_TYPE(UINT_TYPE)(absy)) ==      \
+             (UINT_TYPE)0 ||                                                   \
+         ix == iy)                                                             \
+            ? iy                                                               \
+            : r;                                                               \
     return CLC_AS_TYPE(FLOAT_TYPE)(r);                                         \
   }