neon/rndm: initial implementation

arm/neon.h

@@ -129,6 +129,8 @@
 #include "neon/rev64.h"
 #include "neon/rhadd.h"
 #include "neon/rnd.h"
+#include "neon/rndm.h"
+#include "neon/rndi.h"
 #include "neon/rndn.h"
 #include "neon/rndp.h"
 #include "neon/rshl.h"

arm/neon/rndi.h

+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ *   2020-2021 Evan Nemerson <evan@nemerson.com>
+ */
+
+#if !defined(SIMDE_ARM_NEON_RNDI_H)
+#define SIMDE_ARM_NEON_RNDI_H
+
+#include "types.h"
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+SIMDE_BEGIN_DECLS_
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float32x2_t
+simde_vrndi_f32(simde_float32x2_t a) {
+  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+    return vrndi_f32(a);
+  #else
+    simde_float32x2_private
+      r_,
+      a_ = simde_float32x2_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_nearbyintf(a_.values[i]);
+    }
+
+    return simde_float32x2_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+  #undef vrndi_f32
+  #define vrndi_f32(a) simde_vrndi_f32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float64x1_t
+simde_vrndi_f64(simde_float64x1_t a) {
+  #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+    return vrndi_f64(a);
+  #else
+    simde_float64x1_private
+      r_,
+      a_ = simde_float64x1_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_nearbyint(a_.values[i]);
+    }
+
+    return simde_float64x1_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+  #undef vrndi_f64
+  #define vrndi_f64(a) simde_vrndi_f64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float32x4_t
+simde_vrndiq_f32(simde_float32x4_t a) {
+  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+    return vrndiq_f32(a);
+  #elif defined(SIMDE_X86_SSE4_1_NATIVE)
+    return _mm_round_ps(a, _MM_FROUND_CUR_DIRECTION);
+  #else
+    simde_float32x4_private
+      r_,
+      a_ = simde_float32x4_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_nearbyintf(a_.values[i]);
+    }
+
+    return simde_float32x4_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+  #undef vrndiq_f32
+  #define vrndiq_f32(a) simde_vrndiq_f32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float64x2_t
+simde_vrndiq_f64(simde_float64x2_t a) {
+  #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+    return vrndiq_f64(a);
+  #elif defined(SIMDE_X86_SSE4_1_NATIVE)
+    return _mm_round_pd(a, _MM_FROUND_CUR_DIRECTION);
+  #else
+    simde_float64x2_private
+      r_,
+      a_ = simde_float64x2_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_nearbyint(a_.values[i]);
+    }
+
+    return simde_float64x2_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+  #undef vrndiq_f64
+  #define vrndiq_f64(a) simde_vrndiq_f64(a)
+#endif
+
+SIMDE_END_DECLS_
+HEDLEY_DIAGNOSTIC_POP
+
+#endif /* !defined(SIMDE_ARM_NEON_RNDI_H) */

arm/neon/rndm.h

+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ *   2020-2021 Evan Nemerson <evan@nemerson.com>
+ */
+
+#if !defined(SIMDE_ARM_NEON_RNDM_H)
+#define SIMDE_ARM_NEON_RNDM_H
+
+#include "types.h"
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+SIMDE_BEGIN_DECLS_
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float32x2_t
+simde_vrndm_f32(simde_float32x2_t a) {
+  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+    return vrndm_f32(a);
+  #else
+    simde_float32x2_private
+      r_,
+      a_ = simde_float32x2_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_floorf(a_.values[i]);
+    }
+
+    return simde_float32x2_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+  #undef vrndm_f32
+  #define vrndm_f32(a) simde_vrndm_f32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float64x1_t
+simde_vrndm_f64(simde_float64x1_t a) {
+  #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+    return vrndm_f64(a);
+  #else
+    simde_float64x1_private
+      r_,
+      a_ = simde_float64x1_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_floor(a_.values[i]);
+    }
+
+    return simde_float64x1_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+  #undef vrndm_f64
+  #define vrndm_f64(a) simde_vrndm_f64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float32x4_t
+simde_vrndmq_f32(simde_float32x4_t a) {
+  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+    return vrndmq_f32(a);
+  #elif defined(SIMDE_X86_SSE4_1_NATIVE)
+    return _mm_round_ps(a, _MM_FROUND_TO_NEG_INF);
+  #elif defined(SIMDE_X86_SVML_NATIVE) && defined(SIMDE_X86_SSE_NATIVE)
+    return _mm_floor_ps(a);
+  #elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+    return vec_floor(a);
+  #else
+    simde_float32x4_private
+      r_,
+      a_ = simde_float32x4_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_floorf(a_.values[i]);
+    }
+
+    return simde_float32x4_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+  #undef vrndmq_f32
+  #define vrndmq_f32(a) simde_vrndmq_f32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float64x2_t
+simde_vrndmq_f64(simde_float64x2_t a) {
+  #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+    return vrndmq_f64(a);
+  #elif defined(SIMDE_X86_SSE4_1_NATIVE)
+    return _mm_round_pd(a, _MM_FROUND_TO_NEG_INF);
+  #elif defined(SIMDE_X86_SVML_NATIVE) && defined(SIMDE_X86_SSE_NATIVE)
+    return _mm_floor_pd(a);
+  #elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+    return vec_floor(a);
+  #else
+    simde_float64x2_private
+      r_,
+      a_ = simde_float64x2_to_private(a);
+
+    SIMDE_VECTORIZE
+    for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
+      r_.values[i] = simde_math_floor(a_.values[i]);
+    }
+
+    return simde_float64x2_from_private(r_);
+  #endif
+}
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+  #undef vrndmq_f64
+  #define vrndmq_f64(a) simde_vrndmq_f64(a)
+#endif
+
+SIMDE_END_DECLS_
+HEDLEY_DIAGNOSTIC_POP
+
+#endif /* !defined(SIMDE_ARM_NEON_RNDM_H) */