}
[(set_attr "type" "viwalu")])
+(define_insn_and_split "*vwabda<su><mode>"
+ [(set (match_operand:VWEXTI 0 "register_operand" "+&vr")
+ (plus:VWEXTI
+ (zero_extend:VWEXTI
+ (unspec:<V_DOUBLE_TRUNC>
+ [(match_operand:<V_DOUBLE_TRUNC> 1 "register_operand" "vr")
+ (match_operand:<V_DOUBLE_TRUNC> 2 "register_operand" "vr")]
+ UNSPEC_VABD))
+ (match_operand:VWEXTI 3 "register_operand" "0")))]
+ "TARGET_ZVABD && can_create_pseudo_p ()"
+ "#"
+ "&& 1"
+ [(const_int 0)]
+{
+ rtx ops[] = {operands[0], operands[1], operands[2]};
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_widen_abd_plus<su><mode>,
+ riscv_vector::BINARY_OP, ops);
+ DONE;
+}
+[(set_attr "type" "viwalu")])
+
+;; have this since we don't canonicalize the plus in the presence of an unspec.
+(define_insn_and_split "*vwabda_right<su><mode>"
+ [(set (match_operand:VWEXTI 0 "register_operand" "+&vr")
+ (plus:VWEXTI
+ (match_operand:VWEXTI 1 "register_operand" "0")
+ (zero_extend:VWEXTI
+ (unspec:<V_DOUBLE_TRUNC>
+ [(match_operand:<V_DOUBLE_TRUNC> 2 "register_operand" "vr")
+ (match_operand:<V_DOUBLE_TRUNC> 3 "register_operand" "vr")]
+ UNSPEC_VABD))))]
+ "TARGET_ZVABD && can_create_pseudo_p ()"
+ "#"
+ "&& 1"
+ [(const_int 0)]
+{
+ rtx ops[] = {operands[0], operands[2], operands[3]};
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_widen_abd_plus<su><mode>,
+ riscv_vector::BINARY_OP, ops);
+ DONE;
+}
+[(set_attr "type" "viwalu")])
+
;; This combine pattern does not correspond to a single instruction,
;; i.e. there is no vwmul.wv instruction. This is a temporary pattern
;; produced by a combine pass and if there is no further combine into
[(set_attr "type" "vialu")])
;; -------------------------------------------------------------------------------
-;; - [INT] ABS expansion to vneg and vmax.
+;; - [INT] ABS expansion
;; -------------------------------------------------------------------------------
(define_expand "abs<mode>2"
[(set (match_operand:V_VLSI 0 "register_operand")
- (smax:V_VLSI
- (match_dup 0)
- (neg:V_VLSI
- (match_operand:V_VLSI 1 "register_operand"))))]
+ (abs:V_VLSI
+ (match_operand:V_VLSI 1 "register_operand")))]
"TARGET_VECTOR"
{
+ if (TARGET_ZVABD)
+ {
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_abs<mode>,
+ riscv_vector::UNARY_OP, operands);
+ DONE;
+ }
+
+ rtx neg = gen_reg_rtx (<MODE>mode);
+ rtx ops1[] = {neg, operands[1]};
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_neg<mode>,
+ riscv_vector::UNARY_OP, ops1);
+
+ rtx ops2[] = {operands[0], operands[1], neg};
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_smax<mode>,
+ riscv_vector::BINARY_OP, ops2);
DONE;
})
; ========
; == Absolute difference (not including sum)
; ========
-(define_expand "uabd<mode>3"
- [(match_operand:V_VLSI 0 "register_operand")
- (match_operand:V_VLSI 1 "register_operand")
- (match_operand:V_VLSI 2 "register_operand")]
- ;; Disabled until PR119224 is resolved
- "TARGET_VECTOR && 0"
- {
- rtx max = gen_reg_rtx (<MODE>mode);
- insn_code icode = code_for_pred (UMAX, <MODE>mode);
- rtx ops1[] = {max, operands[1], operands[2]};
- riscv_vector::emit_vlmax_insn (icode, riscv_vector::BINARY_OP, ops1);
-
- rtx min = gen_reg_rtx (<MODE>mode);
- icode = code_for_pred (UMIN, <MODE>mode);
- rtx ops2[] = {min, operands[1], operands[2]};
- riscv_vector::emit_vlmax_insn (icode, riscv_vector::BINARY_OP, ops2);
-
- icode = code_for_pred (MINUS, <MODE>mode);
- rtx ops3[] = {operands[0], max, min};
- riscv_vector::emit_vlmax_insn (icode, riscv_vector::BINARY_OP, ops3);
-
- DONE;
- });
+(define_insn_and_split "<su>abd<mode>3"
+ [(set (match_operand:V_VLSI 0 "register_operand" "=vr")
+ (unspec:V_VLSI
+ [(match_operand:V_VLSI 1 "register_operand" "vr")
+ (match_operand:V_VLSI 2 "register_operand" "vr")]
+ UNSPEC_VABD))]
+ "TARGET_ZVABD && can_create_pseudo_p ()"
+ "#"
+ "&& 1"
+ [(const_int 0)]
+{
+ riscv_vector::emit_vlmax_insn (CODE_FOR_pred_vabd<su><mode>,
+ riscv_vector::BINARY_OP, operands);
+ DONE;
+}
+[(set_attr "type" "vialu")])
/* BITMASK_BIT_POSITION*/ BITMASK_NOT_YET_ALLOCATED,
/* EXTRA_EXTENSION_FLAGS */ 0)
+DEFINE_RISCV_EXT (
+ /* NAME. */ zvabd,
+ /* UPPERCASE_NAME. */ ZVABD,
+ /* FULL_NAME. */ "Vector absolute difference extension.",
+ /* DESC. */ "",
+ /* URL. */ ,
+ /* DEP_EXTS. */ ({"v"}),
+ /* SUPPORTED_VERSIONS. */ ({{0, 7}}),
+ /* FLAG_GROUP. */ zvabd,
+ /* BITMASK_GROUP_ID. */ BITMASK_NOT_YET_ALLOCATED,
+ /* BITMASK_BIT_POSITION. */ BITMASK_NOT_YET_ALLOCATED,
+ /* EXTRA_EXTENSION_FLAGS. */ 0)
+
DEFINE_RISCV_EXT(
/* NAME */ sdtrig,
/* UPPERCASE_NAME */ SDTRIG,
/* BITMASK_BIT_POSITION*/ BITMASK_NOT_YET_ALLOCATED,
/* EXTRA_EXTENSION_FLAGS */ 0)
+
#include "riscv-ext-corev.def"
#include "riscv-ext-sifive.def"
#include "riscv-ext-thead.def"
TargetVariable
int riscv_zvl_subext
+TargetVariable
+int riscv_zvabd_subext
+
Mask(RVE) Var(riscv_base_subext)
Mask(RVI) Var(riscv_base_subext)
Mask(ZHINXMIN) Var(riscv_zinx_subext)
+Mask(ZVABD) Var(riscv_zvabd_subext)
+
Mask(SDTRIG) Var(riscv_sd_subext)
Mask(SHA) Var(riscv_sh_subext)
UNSPEC_SF_VFNRCLIP
UNSPEC_SF_VFNRCLIPU
UNSPEC_SF_CV
+
+ ;; abd
+ UNSPEC_VSABD
+ UNSPEC_VUABD
+ ;; abda
+ UNSPEC_VSABDA
+ UNSPEC_VUABDA
])
(define_c_enum "unspecv" [
(V32DI "V32HI") (V64DI "V64HI") (V128DI "V128HI") (V256DI "V256HI")
(V512DI "V512HI")
])
+
+(define_int_iterator UNSPEC_VABD[
+ UNSPEC_VSABD UNSPEC_VUABD
+])
+
+(define_int_iterator UNSPEC_VABDA[
+ UNSPEC_VSABDA UNSPEC_VUABDA
+])
+
+(define_int_attr su[
+ (UNSPEC_VSABD "s") (UNSPEC_VUABD "u")
+ (UNSPEC_VSABDA "s") (UNSPEC_VUABDA "u")
+])
+
+(define_int_attr u[
+ (UNSPEC_VSABD "") (UNSPEC_VUABD "u")
+ (UNSPEC_VSABDA "") (UNSPEC_VUABDA "u")
+])
riscv_vector::prepare_ternary_operands (operands);
})
+;; ------------------------------------
+;; ---- Vector absolute difference extension
+;; ----------------------------------------------------------------
+;; Includes:
+;; - vabs: Vector Single-Width Signed Integer Absolute
+;; - vabd/vabdu: Vector Single-Width Signed Integer Absolute Difference
+;; - vwabda/vwabdau: Vector Widening Signed Integer Absolute
+;; Difference and Accumulate
+;; ------------------------------------
+
+(define_insn "@pred_abs<mode>"
+ [(set (match_operand:V_VLSI 0 "register_operand" "=vd, vd, vr, vr")
+ (if_then_else:V_VLSI
+ (unspec:<VM>
+ [(match_operand:<VM> 1 "vector_mask_operand" " vm, vm, Wc1, Wc1")
+ (match_operand 4 "vector_length_operand" " rK, rK, rK, rK")
+ (match_operand 5 "const_int_operand" " i, i, i, i")
+ (match_operand 6 "const_int_operand" " i, i, i, i")
+ (match_operand 7 "const_int_operand" " i, i, i, i")
+ (reg:SI VL_REGNUM)
+ (reg:SI VTYPE_REGNUM)] UNSPEC_VPREDICATE)
+ (abs:V_VLSI
+ (match_operand:V_VLSI 3 "register_operand" "vr,vr,vr,vr"))
+ (match_operand:V_VLSI 2 "vector_merge_operand" "vu,0,vu,0")))]
+ "TARGET_ZVABD"
+ "vabs.v\t%0,%3%p1"
+ [(set_attr "type" "vialu")
+ (set_attr "mode" "<MODE>")
+ (set_attr "vl_op_idx" "4")
+ (set (attr "ta") (symbol_ref "riscv_vector::get_ta (operands[5])"))
+ (set (attr "ma") (symbol_ref "riscv_vector::get_ma (operands[6])"))
+ (set (attr "avl_type_idx") (const_int 7))])
+
+(define_insn "@pred_vabd<su><mode>"
+ [(set (match_operand:V_VLSI 0 "register_operand" "=vd, vd, vr, vr")
+ (if_then_else:V_VLSI
+ (unspec:<VM>
+ [(match_operand:<VM> 1 "vector_mask_operand" " vm, vm, Wc1, Wc1")
+ (match_operand 5 "vector_length_operand" " rK, rK, rK, rK")
+ (match_operand 6 "const_int_operand" " i, i, i, i")
+ (match_operand 7 "const_int_operand" " i, i, i, i")
+ (match_operand 8 "const_int_operand" " i, i, i, i")
+ (reg:SI VL_REGNUM)
+ (reg:SI VTYPE_REGNUM)] UNSPEC_VPREDICATE)
+ (unspec:V_VLSI
+ [(match_operand:V_VLSI 3 "register_operand" "vr,vr,vr,vr")
+ (match_operand:V_VLSI 4 "register_operand" "vr,vr,vr,vr")]
+ UNSPEC_VABD)
+ (match_operand:V_VLSI 2 "vector_merge_operand" "vu,0,vu,0")))]
+ "TARGET_ZVABD"
+ "vabd<u>.vv\t%0,%3,%4%p1"
+ [(set_attr "type" "vialu")
+ (set_attr "mode" "<MODE>")])
+
+(define_insn "@pred_widen_abd_plus<su><mode>"
+ [(set (match_operand:VWEXTI 0 "register_operand" "+&vd,&vd,&vr,&vr")
+ (if_then_else:VWEXTI
+ (unspec:<VM>
+ [(match_operand:<VM> 1 "vector_mask_operand" "vm,vm,Wc1,Wc1")
+ (match_operand 5 "vector_length_operand" "rK,rK,rK,rK")
+ (match_operand 6 "const_int_operand" "i,i,i,i")
+ (match_operand 7 "const_int_operand" "i,i,i,i")
+ (match_operand 8 "const_int_operand" "i,i,i,i")
+ (reg:SI VL_REGNUM)
+ (reg:SI VTYPE_REGNUM)] UNSPEC_VPREDICATE)
+ (unspec:VWEXTI
+ [(match_operand:<V_DOUBLE_TRUNC> 3 "register_operand" "vr,vr,vr,vr")
+ (match_operand:<V_DOUBLE_TRUNC> 4 "register_operand" "vr,vr,vr,vr")
+ (match_dup 0)]
+ UNSPEC_VABDA)
+ (match_operand:VWEXTI 2 "vector_merge_operand" "vu,0,vu,0")))]
+ "TARGET_ZVABD"
+ "vwabda<u>.vv\t%0,%3,%4%p1"
+ [(set_attr "type" "viwalu")
+ (set_attr "mode" "<V_DOUBLE_TRUNC>")])
+
(include "autovec.md")
(include "autovec-opt.md")
(include "sifive-vector.md")
@tab 1.0
@tab Minimal half-precision floating-point in integer registers extension
+@item @samp{zvabd}
+@tab 0.7
+@tab Vector absolute difference extension.
+
@item @samp{sdtrig}
@tab 1.0
@tab Debug triggers extension
--- /dev/null
+/* { dg-do compile } */
+/* { dg-additional-options "-march=rv64gcv_zvabd -mabi=lp64d -fno-vect-cost-model" } */
+
+#include <stdint-gcc.h>
+
+static int
+abs (int i)
+{
+ return i < 0 ? -i : i;
+}
+
+#define TEST_VABS(TYPE1, TYPE2) \
+ __attribute__((noipa)) void vabs_##TYPE1_##TYPE2 (TYPE1 *__restrict dst, \
+ TYPE2 *__restrict a, \
+ int n) \
+ { \
+ int i; \
+ for (i = 0; i < n; i++) \
+ dst[i] = abs (a[i]); \
+ }
+
+#define TEST_VABD(TYPE1, TYPE2) \
+ __attribute__((noipa)) void vabd_##TYPE1_##TYPE2 (TYPE1 *__restrict dst, \
+ TYPE2 *__restrict a, \
+ TYPE2 *__restrict b, \
+ int n) \
+ { \
+ int i; \
+ for (i = 0; i < n; i++) \
+ dst[i] = abs (a[i] - b[i]); \
+ }
+
+#define TEST_VWABDA(TYPE1, TYPE2) \
+ __attribute__((noipa)) void \
+ vwabda_##TYPE1_##TYPE2 (TYPE1 *__restrict dst, TYPE2 *__restrict a, \
+ TYPE2 *__restrict b, int n) \
+ { \
+ int i; \
+ for (i = 0; i < n; i++) \
+ dst[i] += abs (a[i] - b[i]); \
+ }
+
+#define TEST_ALL() \
+ TEST_VABS (int8_t, int8_t) \
+ TEST_VABS (int16_t, int16_t) \
+ TEST_VABD (int8_t, int8_t) \
+ TEST_VABD (uint8_t, uint8_t) \
+ TEST_VABD (int16_t, int16_t) \
+ TEST_VABD (uint16_t, uint16_t) \
+ TEST_VWABDA (int16_t, int8_t) \
+ TEST_VWABDA (uint16_t, uint8_t) \
+ TEST_VWABDA (int32_t, int16_t) \
+ TEST_VWABDA (uint32_t, uint16_t)
+
+TEST_ALL()
+
+/* { dg-final { scan-assembler-times {\tvabs\.v} 2 } } */
+/* { dg-final { scan-assembler-times {\tvabd\.vv} 2 } } */
+/* { dg-final { scan-assembler-times {\tvabdu\.vv} 2 } } */
+/* { dg-final { scan-assembler-times {\tvwabda\.vv} 2 } } */
+/* { dg-final { scan-assembler-times {\tvwabdau\.vv} 2 } } */
projects / thirdparty / gcc.git / commitdiff
