1 files changed, 83 insertions, 0 deletions
diff --git a/math/aarch64/sve/sv_log1pf_inline.h b/math/aarch64/sve/sv_log1pf_inline.h
new file mode 100644
index 000000000000..238079c61a5b
--- /dev/null
+++ b/math/aarch64/sve/sv_log1pf_inline.h
@@ -0,0 +1,83 @@
+/*
+ * Helper for SVE routines which calculate log(1 + x) and do not
+ * need special-case handling
+ *
+ * Copyright (c) 2023-2024, Arm Limited.
+ * SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
+ */
+
+#ifndef MATH_SV_LOG1PF_INLINE_H
+#define MATH_SV_LOG1PF_INLINE_H
+
+#define SignExponentMask 0xff800000
+
+static const struct sv_log1pf_data
+{
+  float c0, c2, c4, c6;
+  float c1, c3, c5, c7;
+  float ln2, exp_bias, quarter;
+  uint32_t four, three_quarters;
+} sv_log1pf_data = {
+  /* Do not store first term of polynomial, which is -0.5, as
+     this can be fmov-ed directly instead of including it in
+     the main load-and-mla polynomial schedule.  */
+  .c0 = 0x1.5555aap-2f,		.c1 = -0x1.000038p-2f, .c2 = 0x1.99675cp-3f,
+  .c3 = -0x1.54ef78p-3f,	.c4 = 0x1.28a1f4p-3f,  .c5 = -0x1.0da91p-3f,
+  .c6 = 0x1.abcb6p-4f,		.c7 = -0x1.6f0d5ep-5f, .ln2 = 0x1.62e43p-1f,
+  .exp_bias = 0x1p-23f,		.quarter = 0x1p-2f,    .four = 0x40800000,
+  .three_quarters = 0x3f400000,
+};
+
+static inline svfloat32_t
+sv_log1pf_inline (svfloat32_t x, svbool_t pg)
+{
+  const struct sv_log1pf_data *d = ptr_barrier (&sv_log1pf_data);
+
+  /* With x + 1 = t * 2^k (where t = m + 1 and k is chosen such that m
+			 is in [-0.25, 0.5]):
+   log1p(x) = log(t) + log(2^k) = log1p(m) + k*log(2).
+
+   We approximate log1p(m) with a polynomial, then scale by
+   k*log(2). Instead of doing this directly, we use an intermediate
+   scale factor s = 4*k*log(2) to ensure the scale is representable
+   as a normalised fp32 number.  */
+  svfloat32_t m = svadd_x (pg, x, 1);
+
+  /* Choose k to scale x to the range [-1/4, 1/2].  */
+  svint32_t k
+      = svand_x (pg, svsub_x (pg, svreinterpret_s32 (m), d->three_quarters),
+		 sv_s32 (SignExponentMask));
+
+  /* Scale x by exponent manipulation.  */
+  svfloat32_t m_scale = svreinterpret_f32 (
+      svsub_x (pg, svreinterpret_u32 (x), svreinterpret_u32 (k)));
+
+  /* Scale up to ensure that the scale factor is representable as normalised
+     fp32 number, and scale m down accordingly.  */
+  svfloat32_t s = svreinterpret_f32 (svsubr_x (pg, k, d->four));
+  svfloat32_t fconst = svld1rq_f32 (svptrue_b32 (), &d->ln2);
+  m_scale = svadd_x (pg, m_scale, svmla_lane_f32 (sv_f32 (-1), s, fconst, 2));
+
+  /* Evaluate polynomial on reduced interval.  */
+  svfloat32_t ms2 = svmul_x (svptrue_b32 (), m_scale, m_scale);
+
+  svfloat32_t c1357 = svld1rq_f32 (svptrue_b32 (), &d->c1);
+  svfloat32_t p01 = svmla_lane_f32 (sv_f32 (d->c0), m_scale, c1357, 0);
+  svfloat32_t p23 = svmla_lane_f32 (sv_f32 (d->c2), m_scale, c1357, 1);
+  svfloat32_t p45 = svmla_lane_f32 (sv_f32 (d->c4), m_scale, c1357, 2);
+  svfloat32_t p67 = svmla_lane_f32 (sv_f32 (d->c6), m_scale, c1357, 3);
+
+  svfloat32_t p = svmla_x (pg, p45, p67, ms2);
+  p = svmla_x (pg, p23, p, ms2);
+  p = svmla_x (pg, p01, p, ms2);
+
+  p = svmad_x (pg, m_scale, p, -0.5);
+  p = svmla_x (pg, m_scale, m_scale, svmul_x (pg, m_scale, p));
+
+  /* The scale factor to be applied back at the end - by multiplying float(k)
+   by 2^-23 we get the unbiased exponent of k.  */
+  svfloat32_t scale_back = svmul_lane_f32 (svcvt_f32_x (pg, k), fconst, 1);
+  return svmla_lane_f32 (p, scale_back, fconst, 0);
+}
+
+#endif //  SV_LOG1PF_INLINE_H