DONE;
})
+;; =========================================================================
+;; == Widening/narrowing Conversions
+;; =========================================================================
+
+;; -------------------------------------------------------------------------
+;; ---- [INT<-FP] Widening Conversions
+;; -------------------------------------------------------------------------
+;; Includes:
+;; - vfwcvt.rtz.xu.f.v
+;; - vfwcvt.rtz.x.f.v
+;; -------------------------------------------------------------------------
+(define_expand "<optab><vnconvert><mode>2"
+ [(set (match_operand:VWCONVERTI 0 "register_operand")
+ (any_fix:VWCONVERTI
+ (match_operand:<VNCONVERT> 1 "register_operand")))]
+ "TARGET_VECTOR"
+{
+ insn_code icode = code_for_pred_widen (<CODE>, <MODE>mode);
+ riscv_vector::emit_vlmax_insn (icode, riscv_vector::RVV_UNOP, operands);
+ DONE;
+})
+
+;; -------------------------------------------------------------------------
+;; ---- [FP<-INT] Widening Conversions
+;; -------------------------------------------------------------------------
+;; Includes:
+;; - vfwcvt.f.xu.v
+;; - vfwcvt.f.x.v
+;; -------------------------------------------------------------------------
+(define_expand "<float_cvt><vnconvert><mode>2"
+ [(set (match_operand:VF 0 "register_operand")
+ (any_float:VF
+ (match_operand:<VNCONVERT> 1 "register_operand")))]
+ "TARGET_VECTOR"
+{
+ insn_code icode = code_for_pred_widen (<CODE>, <MODE>mode);
+ riscv_vector::emit_vlmax_insn (icode, riscv_vector::RVV_UNOP, operands);
+ DONE;
+})
+
+;; -------------------------------------------------------------------------
+;; ---- [INT<-FP] Narrowing Conversions
+;; -------------------------------------------------------------------------
+;; Includes:
+;; - vfncvt.rtz.xu.f.v
+;; - vfncvt.rtz.x.f.v
+;; -------------------------------------------------------------------------
+(define_expand "<optab><mode><vnconvert>2"
+ [(set (match_operand:<VNCONVERT> 0 "register_operand")
+ (any_fix:<VNCONVERT>
+ (match_operand:VF 1 "register_operand")))]
+ "TARGET_VECTOR"
+{
+ insn_code icode = code_for_pred_narrow (<CODE>, <MODE>mode);
+ riscv_vector::emit_vlmax_insn (icode, riscv_vector::RVV_UNOP, operands);
+ DONE;
+})
+
+;; -------------------------------------------------------------------------
+;; ---- [FP<-INT] Narrowing Conversions
+;; -------------------------------------------------------------------------
+;; Includes:
+;; - vfncvt.f.xu.w
+;; - vfncvt.f.x.w
+;; -------------------------------------------------------------------------
+(define_expand "<float_cvt><mode><vnconvert>2"
+ [(set (match_operand:<VNCONVERT> 0 "register_operand")
+ (any_float:<VNCONVERT>
+ (match_operand:VWCONVERTI 1 "register_operand")))]
+ "TARGET_VECTOR"
+{
+ insn_code icode = code_for_pred_narrow (<CODE>, <MODE>mode);
+ riscv_vector::emit_vlmax_fp_insn (icode, riscv_vector::RVV_UNOP, operands);
+ DONE;
+})
+
;; =========================================================================
;; == Unary arithmetic
;; =========================================================================
(VNx1DF "VNx1SI") (VNx2DF "VNx2SI") (VNx4DF "VNx4SI") (VNx8DF "VNx8SI") (VNx16DF "VNx16SI")
])
+(define_mode_attr vnconvert [
+ (VNx1HF "vnx1qi") (VNx2HF "vnx2qi") (VNx4HF "vnx4qi") (VNx8HF "vnx8qi") (VNx16HF "vnx16qi") (VNx32HF "vnx32qi") (VNx64HF "vnx64qi")
+ (VNx1SF "vnx1hi") (VNx2SF "vnx2hi") (VNx4SF "vnx4hi") (VNx8SF "vnx8hi") (VNx16SF "vnx16hi") (VNx32SF "vnx32hi")
+ (VNx1SI "vnx1hf") (VNx2SI "vnx2hf") (VNx4SI "vnx4hf") (VNx8SI "vnx8hf") (VNx16SI "vnx16hf") (VNx32SI "vnx32hf")
+ (VNx1DI "vnx1sf") (VNx2DI "vnx2sf") (VNx4DI "vnx4sf") (VNx8DI "vnx8sf") (VNx16DI "vnx16sf")
+ (VNx1DF "vnx1si") (VNx2DF "vnx2si") (VNx4DF "vnx4si") (VNx8DF "vnx8si") (VNx16DF "vnx16si")
+])
+
(define_mode_attr VDEMOTE [
(VNx1DI "VNx2SI") (VNx2DI "VNx4SI")
(VNx4DI "VNx8SI") (VNx8DI "VNx16SI") (VNx16DI "VNx32SI")
--- /dev/null
+/* { dg-do run { target { riscv_vector } } } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-ftoi-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3.1315926 - 92.947289; \
+ } \
+ vfncvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3.1315926 + 92.947289; \
+ } \
+ vfncvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (float, int16_t, 3)
+ RUN (float, int16_t, 4)
+ RUN (float, int16_t, 7)
+ RUN (float, int16_t, 99)
+ RUN (float, int16_t, 119)
+ RUN (float, int16_t, 128)
+ RUN (float, int16_t, 256)
+ RUN (float, int16_t, 279)
+ RUN (float, int16_t, 555)
+ RUN (float, int16_t, 1024)
+ RUN (float, int16_t, 1389)
+ RUN (float, int16_t, 2048)
+ RUN (float, int16_t, 3989)
+ RUN (float, int16_t, 4096)
+ RUN (float, int16_t, 5975)
+
+ RUN2 (float, uint16_t, 3)
+ RUN2 (float, uint16_t, 4)
+ RUN2 (float, uint16_t, 7)
+ RUN2 (float, uint16_t, 99)
+ RUN2 (float, uint16_t, 119)
+ RUN2 (float, uint16_t, 128)
+ RUN2 (float, uint16_t, 256)
+ RUN2 (float, uint16_t, 279)
+ RUN2 (float, uint16_t, 555)
+ RUN2 (float, uint16_t, 1024)
+ RUN2 (float, uint16_t, 1389)
+ RUN2 (float, uint16_t, 2048)
+ RUN2 (float, uint16_t, 3989)
+ RUN2 (float, uint16_t, 4096)
+ RUN2 (float, uint16_t, 5975)
+
+ RUN (double, int32_t, 3)
+ RUN (double, int32_t, 4)
+ RUN (double, int32_t, 7)
+ RUN (double, int32_t, 99)
+ RUN (double, int32_t, 119)
+ RUN (double, int32_t, 128)
+ RUN (double, int32_t, 256)
+ RUN (double, int32_t, 279)
+ RUN (double, int32_t, 555)
+ RUN (double, int32_t, 1024)
+ RUN (double, int32_t, 1389)
+ RUN (double, int32_t, 2048)
+ RUN (double, int32_t, 3989)
+ RUN (double, int32_t, 4096)
+ RUN (double, int32_t, 5975)
+
+ RUN2 (double, uint32_t, 3)
+ RUN2 (double, uint32_t, 4)
+ RUN2 (double, uint32_t, 7)
+ RUN2 (double, uint32_t, 99)
+ RUN2 (double, uint32_t, 119)
+ RUN2 (double, uint32_t, 128)
+ RUN2 (double, uint32_t, 256)
+ RUN2 (double, uint32_t, 279)
+ RUN2 (double, uint32_t, 555)
+ RUN2 (double, uint32_t, 1024)
+ RUN2 (double, uint32_t, 1389)
+ RUN2 (double, uint32_t, 2048)
+ RUN2 (double, uint32_t, 3989)
+ RUN2 (double, uint32_t, 4096)
+ RUN2 (double, uint32_t, 5975)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv32gcv_zvfh -mabi=ilp32d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-ftoi-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfncvt\.rtz\.x\.f\.w} 3 } } */
+/* { dg-final { scan-assembler-times {\tvfncvt\.rtz\.xu\.f\.w} 3 } } */
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv64gcv_zvfh -mabi=lp64d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-ftoi-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfncvt\.rtz\.x\.f\.w} 3 } } */
+/* { dg-final { scan-assembler-times {\tvfncvt\.rtz\.xu\.f\.w} 3 } } */
--- /dev/null
+#include <stdint-gcc.h>
+
+#define TEST(TYPE1, TYPE2) \
+ __attribute__ ((noipa)) void vfncvt_##TYPE1##TYPE2 (TYPE2 *dst, TYPE1 *a, \
+ int n) \
+ { \
+ for (int i = 0; i < n; i++) \
+ dst[i] = (TYPE2) a[i]; \
+ }
+
+#define TEST_ALL() \
+ TEST (double, int32_t) \
+ TEST (double, uint32_t) \
+ TEST (float, int16_t) \
+ TEST (float, uint16_t) \
+ TEST (_Float16, int8_t) \
+ TEST (_Float16, uint8_t) \
+
+TEST_ALL ()
--- /dev/null
+/* { dg-do run { target { riscv_vector && riscv_zvfh_hw } } } */
+/* { dg-additional-options "-std=c99 -march=rv64gcv_zvfh -mabi=lp64d -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-ftoi-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3.1315926 - 8.947289; \
+ } \
+ vfcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3.1315926 + 8.947289; \
+ } \
+ vfcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (_Float16, int8_t, 3)
+ RUN (_Float16, int8_t, 4)
+ RUN (_Float16, int8_t, 13)
+ RUN (_Float16, int8_t, 40)
+
+ RUN2 (_Float16, uint8_t, 1)
+ RUN2 (_Float16, uint8_t, 8)
+ RUN2 (_Float16, uint8_t, 21)
+ RUN2 (_Float16, uint8_t, 33)
+}
--- /dev/null
+/* { dg-do run { target { riscv_vector } } } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-itof-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3 + 88932; \
+ } \
+ vfncvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (int64_t, float, 3)
+ RUN (int64_t, float, 4)
+ RUN (int64_t, float, 7)
+ RUN (int64_t, float, 99)
+ RUN (int64_t, float, 119)
+ RUN (int64_t, float, 128)
+ RUN (int64_t, float, 256)
+ RUN (int64_t, float, 279)
+ RUN (int64_t, float, 555)
+ RUN (int64_t, float, 1024)
+ RUN (int64_t, float, 1389)
+ RUN (int64_t, float, 2048)
+ RUN (int64_t, float, 3989)
+ RUN (int64_t, float, 4096)
+ RUN (int64_t, float, 5975)
+
+ RUN (uint64_t, float, 3)
+ RUN (uint64_t, float, 4)
+ RUN (uint64_t, float, 7)
+ RUN (uint64_t, float, 99)
+ RUN (uint64_t, float, 119)
+ RUN (uint64_t, float, 128)
+ RUN (uint64_t, float, 256)
+ RUN (uint64_t, float, 279)
+ RUN (uint64_t, float, 555)
+ RUN (uint64_t, float, 1024)
+ RUN (uint64_t, float, 1389)
+ RUN (uint64_t, float, 2048)
+ RUN (uint64_t, float, 3989)
+ RUN (uint64_t, float, 4096)
+ RUN (uint64_t, float, 5975)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv32gcv_zvfh -mabi=ilp32d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-itof-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfncvt\.f\.x\.w} 2 } } */
+/* { dg-final { scan-assembler-times {\tvfncvt\.f\.xu\.w} 2 } } */
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv64gcv_zvfh -mabi=lp64d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-itof-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfncvt\.f\.x\.w} 2 } } */
+/* { dg-final { scan-assembler-times {\tvfncvt\.f\.xu\.w} 2 } } */
--- /dev/null
+#include <stdint-gcc.h>
+
+#define TEST(TYPE1, TYPE2) \
+ __attribute__ ((noipa)) \
+ void vfncvt_##TYPE1##TYPE2 (TYPE2 *restrict dst, \
+ TYPE1 *restrict a, int n) \
+ { \
+ for (int i = 0; i < n; i++) \
+ dst[i] = (TYPE2) a[i]; \
+ }
+
+#define TEST_ALL() \
+ TEST (int64_t, float) \
+ TEST (uint64_t, float) \
+ TEST (int32_t, _Float16) \
+ TEST (uint32_t, _Float16) \
+
+TEST_ALL ()
--- /dev/null
+/* { dg-do run { target { riscv_vector && riscv_zvfh_hw } } } */
+/* { dg-additional-options "-std=c99 -march=rv64gcv_zvfh -mabi=lp64d -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfncvt-itof-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3 - 832; \
+ } \
+ vfncvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3 + 892; \
+ } \
+ vfncvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (int32_t, _Float16, 3)
+ RUN (int32_t, _Float16, 4)
+ RUN (int32_t, _Float16, 7)
+ RUN (int32_t, _Float16, 99)
+ RUN (int32_t, _Float16, 119)
+ RUN (int32_t, _Float16, 128)
+ RUN (int32_t, _Float16, 256)
+ RUN (int32_t, _Float16, 279)
+ RUN (int32_t, _Float16, 555)
+ RUN (int32_t, _Float16, 1024)
+ RUN (int32_t, _Float16, 1389)
+ RUN (int32_t, _Float16, 2048)
+ RUN (int32_t, _Float16, 3989)
+ RUN (int32_t, _Float16, 4096)
+ RUN (int32_t, _Float16, 5975)
+
+ RUN2 (uint32_t, _Float16, 3)
+ RUN2 (uint32_t, _Float16, 4)
+ RUN2 (uint32_t, _Float16, 7)
+ RUN2 (uint32_t, _Float16, 99)
+ RUN2 (uint32_t, _Float16, 119)
+ RUN2 (uint32_t, _Float16, 128)
+ RUN2 (uint32_t, _Float16, 256)
+ RUN2 (uint32_t, _Float16, 279)
+ RUN2 (uint32_t, _Float16, 555)
+ RUN2 (uint32_t, _Float16, 1024)
+ RUN2 (uint32_t, _Float16, 1389)
+ RUN2 (uint32_t, _Float16, 2048)
+ RUN2 (uint32_t, _Float16, 3989)
+ RUN2 (uint32_t, _Float16, 4096)
+ RUN2 (uint32_t, _Float16, 5975)
+}
--- /dev/null
+/* { dg-do run { target { riscv_vector } } } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-ftoi-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3.1315926 - 932.947289; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3.1315926 + 932.947289; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (float, int64_t, 3)
+ RUN (float, int64_t, 4)
+ RUN (float, int64_t, 7)
+ RUN (float, int64_t, 99)
+ RUN (float, int64_t, 119)
+ RUN (float, int64_t, 128)
+ RUN (float, int64_t, 256)
+ RUN (float, int64_t, 279)
+ RUN (float, int64_t, 555)
+ RUN (float, int64_t, 1024)
+ RUN (float, int64_t, 1389)
+ RUN (float, int64_t, 2048)
+ RUN (float, int64_t, 3989)
+ RUN (float, int64_t, 4096)
+ RUN (float, int64_t, 5975)
+
+ RUN2 (float, uint64_t, 3)
+ RUN2 (float, uint64_t, 4)
+ RUN2 (float, uint64_t, 7)
+ RUN2 (float, uint64_t, 99)
+ RUN2 (float, uint64_t, 119)
+ RUN2 (float, uint64_t, 128)
+ RUN2 (float, uint64_t, 256)
+ RUN2 (float, uint64_t, 279)
+ RUN2 (float, uint64_t, 555)
+ RUN2 (float, uint64_t, 1024)
+ RUN2 (float, uint64_t, 1389)
+ RUN2 (float, uint64_t, 2048)
+ RUN2 (float, uint64_t, 3989)
+ RUN2 (float, uint64_t, 4096)
+ RUN2 (float, uint64_t, 5975)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv32gcv_zvfh -mabi=ilp32d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-ftoi-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfwcvt\.rtz\.x\.f\.v} 2 } } */
+/* { dg-final { scan-assembler-times {\tvfwcvt\.rtz\.xu\.f\.v} 2 } } */
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv64gcv_zvfh -mabi=lp64d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-ftoi-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfwcvt\.rtz\.x\.f\.v} 2 } } */
+/* { dg-final { scan-assembler-times {\tvfwcvt\.rtz\.xu\.f\.v} 2 } } */
--- /dev/null
+#include <stdint-gcc.h>
+
+#define TEST(TYPE1, TYPE2) \
+ __attribute__ ((noipa)) void vfwcvt_##TYPE1##TYPE2 (TYPE2 *dst, TYPE1 *a, \
+ int n) \
+ { \
+ for (int i = 0; i < n; i++) \
+ dst[i] = (TYPE2) a[i]; \
+ }
+
+#define TEST_ALL() \
+ TEST (_Float16, int32_t) \
+ TEST (_Float16, uint32_t) \
+ TEST (float, int64_t) \
+ TEST (float, uint64_t) \
+
+TEST_ALL ()
--- /dev/null
+/* { dg-do run { target { riscv_vector && riscv_zvfh_hw } } } */
+/* { dg-additional-options "-std=c99 -march=rv64gcv_zvfh -mabi=lp64d -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfcvt_rtz-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3.1315926 - 832.947289; \
+ } \
+ vfcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3.1315926 + 832.947289; \
+ } \
+ vfcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (_Float16, int32_t, 3)
+ RUN (_Float16, int32_t, 4)
+ RUN (_Float16, int32_t, 7)
+ RUN (_Float16, int32_t, 99)
+ RUN (_Float16, int32_t, 119)
+ RUN (_Float16, int32_t, 128)
+ RUN (_Float16, int32_t, 256)
+ RUN (_Float16, int32_t, 279)
+ RUN (_Float16, int32_t, 555)
+ RUN (_Float16, int32_t, 1024)
+ RUN (_Float16, int32_t, 1389)
+ RUN (_Float16, int32_t, 2048)
+ RUN (_Float16, int32_t, 3989)
+ RUN (_Float16, int32_t, 4096)
+ RUN (_Float16, int32_t, 5975)
+
+ RUN2 (_Float16, uint32_t, 3)
+ RUN2 (_Float16, uint32_t, 4)
+ RUN2 (_Float16, uint32_t, 7)
+ RUN2 (_Float16, uint32_t, 99)
+ RUN2 (_Float16, uint32_t, 119)
+ RUN2 (_Float16, uint32_t, 128)
+ RUN2 (_Float16, uint32_t, 256)
+ RUN2 (_Float16, uint32_t, 279)
+ RUN2 (_Float16, uint32_t, 555)
+ RUN2 (_Float16, uint32_t, 1024)
+ RUN2 (_Float16, uint32_t, 1389)
+ RUN2 (_Float16, uint32_t, 2048)
+ RUN2 (_Float16, uint32_t, 3989)
+ RUN2 (_Float16, uint32_t, 4096)
+ RUN2 (_Float16, uint32_t, 5975)
+}
--- /dev/null
+/* { dg-do run { target { riscv_vector } } } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-itof-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3 - 88932; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3 + 88932; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (int16_t, float, 3)
+ RUN (int16_t, float, 4)
+ RUN (int16_t, float, 7)
+ RUN (int16_t, float, 99)
+ RUN (int16_t, float, 119)
+ RUN (int16_t, float, 128)
+ RUN (int16_t, float, 256)
+ RUN (int16_t, float, 279)
+ RUN (int16_t, float, 555)
+ RUN (int16_t, float, 1024)
+ RUN (int16_t, float, 1389)
+ RUN (int16_t, float, 2048)
+ RUN (int16_t, float, 3989)
+ RUN (int16_t, float, 4096)
+ RUN (int16_t, float, 5975)
+
+ RUN2 (uint16_t, float, 3)
+ RUN2 (uint16_t, float, 4)
+ RUN2 (uint16_t, float, 7)
+ RUN2 (uint16_t, float, 99)
+ RUN2 (uint16_t, float, 119)
+ RUN2 (uint16_t, float, 128)
+ RUN2 (uint16_t, float, 256)
+ RUN2 (uint16_t, float, 279)
+ RUN2 (uint16_t, float, 555)
+ RUN2 (uint16_t, float, 1024)
+ RUN2 (uint16_t, float, 1389)
+ RUN2 (uint16_t, float, 2048)
+ RUN2 (uint16_t, float, 3989)
+ RUN2 (uint16_t, float, 4096)
+ RUN2 (uint16_t, float, 5975)
+
+ RUN (int32_t, double, 3)
+ RUN (int32_t, double, 4)
+ RUN (int32_t, double, 7)
+ RUN (int32_t, double, 99)
+ RUN (int32_t, double, 119)
+ RUN (int32_t, double, 128)
+ RUN (int32_t, double, 256)
+ RUN (int32_t, double, 279)
+ RUN (int32_t, double, 555)
+ RUN (int32_t, double, 1024)
+ RUN (int32_t, double, 1389)
+ RUN (int32_t, double, 2048)
+ RUN (int32_t, double, 3989)
+ RUN (int32_t, double, 4096)
+ RUN (int32_t, double, 5975)
+
+ RUN2 (uint32_t, double, 3)
+ RUN2 (uint32_t, double, 4)
+ RUN2 (uint32_t, double, 7)
+ RUN2 (uint32_t, double, 99)
+ RUN2 (uint32_t, double, 119)
+ RUN2 (uint32_t, double, 128)
+ RUN2 (uint32_t, double, 256)
+ RUN2 (uint32_t, double, 279)
+ RUN2 (uint32_t, double, 555)
+ RUN2 (uint32_t, double, 1024)
+ RUN2 (uint32_t, double, 1389)
+ RUN2 (uint32_t, double, 2048)
+ RUN2 (uint32_t, double, 3989)
+ RUN2 (uint32_t, double, 4096)
+ RUN2 (uint32_t, double, 5975)
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv32gcv_zvfh -mabi=ilp32d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-itof-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfwcvt\.f\.x\.v} 3 } } */
+/* { dg-final { scan-assembler-times {\tvfwcvt\.f\.xu\.v} 3 } } */
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-std=c99 -fno-vect-cost-model -march=rv64gcv_zvfh -mabi=lp64d --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-itof-template.h"
+
+/* { dg-final { scan-assembler-times {\tvfwcvt\.f\.x\.v} 3 } } */
+/* { dg-final { scan-assembler-times {\tvfwcvt\.f\.xu\.v} 3 } } */
--- /dev/null
+#include <stdint-gcc.h>
+
+#define TEST(TYPE1, TYPE2) \
+ __attribute__ ((noipa)) \
+ void vfwcvt_##TYPE1##TYPE2 (TYPE2 *restrict dst, \
+ TYPE1 *restrict a, int n) \
+ { \
+ for (int i = 0; i < n; i++) \
+ dst[i] = (TYPE2) a[i]; \
+ }
+
+#define TEST_ALL() \
+ TEST (int16_t, float) \
+ TEST (uint16_t, float) \
+ TEST (int32_t, double) \
+ TEST (uint32_t, double) \
+ TEST (int8_t, _Float16) \
+ TEST (uint8_t, _Float16) \
+
+TEST_ALL ()
--- /dev/null
+/* { dg-do run { target { riscv_vector && riscv_zvfh_hw } } } */
+/* { dg-additional-options "-std=c99 -march=rv64gcv_zvfh -mabi=lp64d -fno-vect-cost-model --param=riscv-autovec-preference=scalable" } */
+
+#include "vfwcvt-itof-template.h"
+
+#define RUN(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * -3 - 8; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+#define RUN2(TYPE1, TYPE2, NUM) \
+ TYPE1 src##TYPE1##TYPE2##NUM[NUM]; \
+ TYPE2 dst##TYPE1##TYPE2##NUM[NUM]; \
+ for (int i = 0; i < NUM; i++) \
+ { \
+ src##TYPE1##TYPE2##NUM[i] = i * 3 + 8; \
+ } \
+ vfwcvt_##TYPE1##TYPE2 (dst##TYPE1##TYPE2##NUM, src##TYPE1##TYPE2##NUM, NUM); \
+ for (int i = 0; i < NUM; i++) \
+ if (dst##TYPE1##TYPE2##NUM[i] != (TYPE2) src##TYPE1##TYPE2##NUM[i]) \
+ __builtin_abort ();
+
+int
+main ()
+{
+ RUN (int8_t, _Float16, 3)
+ RUN (int8_t, _Float16, 4)
+ RUN (int8_t, _Float16, 7)
+ RUN (int8_t, _Float16, 12)
+ RUN (int8_t, _Float16, 20)
+ RUN (int8_t, _Float16, 27)
+
+ RUN (int8_t, _Float16, 4)
+ RUN (int8_t, _Float16, 8)
+ RUN (int8_t, _Float16, 11)
+ RUN (int8_t, _Float16, 29)
+ RUN (int8_t, _Float16, 49)
+}