LoongArch: Fix FP vector comparsons [PR113034]

We had the following mappings between <x>vfcmp submenmonics and RTX
codes:

    (define_code_attr fcc
      [(unordered "cun")
       (ordered   "cor")
       (eq       "ceq")
       (ne       "cne")
       (uneq      "cueq")
       (unle      "cule")
       (unlt      "cult")
       (le       "cle")
       (lt       "clt")])

This is inconsistent with scalar code:

    (define_code_attr fcond [(unordered "cun")
                             (uneq "cueq")
                             (unlt "cult")
                             (unle "cule")
                             (eq "ceq")
                             (lt "slt")
                             (le "sle")
                             (ordered "cor")
                             (ltgt "sne")
                             (ne "cune")
                             (ge "sge")
                             (gt "sgt")
                             (unge "cuge")
                             (ungt "cugt")])

For every RTX code for which the LSX/LASX code is different from the
scalar code, the scalar code is correct and the LSX/LASX code is wrong.
Most seriously, the RTX code NE should be mapped to "cneq", not "cne".
Rewrite <x>vfcmp define_insns in simd.md using the same mapping as
scalar fcmp.

Note that GAS does not support [x]vfcmp.{c/s}[u]{ge/gt} (pseudo)
instruction (although fcmp.{c/s}[u]{ge/gt} is supported), so we need to
switch the order of inputs and use [x]vfcmp.{c/s}[u]{le/lt} instead.

The <x>vfcmp.{sult/sule/clt/cle}.{s/d} instructions do not have a single
RTX code, but they can be modeled as an inversed RTX code following a
"not" operation.  Doing so allows the compiler to optimized vectorized
__builtin_isless etc. to a single instruction.  This optimization should
be added for scalar code too and I'll do it later.

Tests are added for mapping between C code, IEC 60559 operations, and
vfcmp instructions.

[1]:https://gcc.gnu.org/pipermail/gcc-patches/2023-December/640713.html

gcc/ChangeLog:

	PR target/113034
	* config/loongarch/lasx.md (UNSPEC_LASX_XVFCMP_*): Remove.
	(lasx_xvfcmp_caf_<flasxfmt>): Remove.
	(lasx_xvfcmp_cune_<FLASX:flasxfmt>): Remove.
	(FSC256_UNS): Remove.
	(fsc256): Remove.
	(lasx_xvfcmp_<vfcond:fcc>_<FLASX:flasxfmt>): Remove.
	(lasx_xvfcmp_<fsc256>_<FLASX:flasxfmt>): Remove.
	* config/loongarch/lsx.md (UNSPEC_LSX_XVFCMP_*): Remove.
	(lsx_vfcmp_caf_<flsxfmt>): Remove.
	(lsx_vfcmp_cune_<FLSX:flsxfmt>): Remove.
	(vfcond): Remove.
	(fcc): Remove.
	(FSC_UNS): Remove.
	(fsc): Remove.
	(lsx_vfcmp_<vfcond:fcc>_<FLSX:flsxfmt>): Remove.
	(lsx_vfcmp_<fsc>_<FLSX:flsxfmt>): Remove.
	* config/loongarch/simd.md
	(fcond_simd): New define_code_iterator.
	(<simd_isa>_<x>vfcmp_<fcond:fcond_simd>_<simdfmt>):
	New define_insn.
	(fcond_simd_rev): New define_code_iterator.
	(fcond_rev_asm): New define_code_attr.
	(<simd_isa>_<x>vfcmp_<fcond:fcond_simd_rev>_<simdfmt>):
	New define_insn.
	(fcond_inv): New define_code_iterator.
	(fcond_inv_rev): New define_code_iterator.
	(fcond_inv_rev_asm): New define_code_attr.
	(<simd_isa>_<x>vfcmp_<fcond_inv>_<simdfmt>): New define_insn.
	(<simd_isa>_<x>vfcmp_<fcond_inv:fcond_inv_rev>_<simdfmt>):
	New define_insn.
	(UNSPEC_SIMD_FCMP_CAF, UNSPEC_SIMD_FCMP_SAF,
	UNSPEC_SIMD_FCMP_SEQ, UNSPEC_SIMD_FCMP_SUN,
	UNSPEC_SIMD_FCMP_SUEQ, UNSPEC_SIMD_FCMP_CNE,
	UNSPEC_SIMD_FCMP_SOR, UNSPEC_SIMD_FCMP_SUNE): New unspecs.
	(SIMD_FCMP): New define_int_iterator.
	(fcond_unspec): New define_int_attr.
	(<simd_isa>_<x>vfcmp_<fcond_unspec>_<simdfmt>): New define_insn.
	* config/loongarch/loongarch.cc (loongarch_expand_lsx_cmp):
	Remove unneeded special cases.

gcc/testsuite/ChangeLog:

	PR target/113034
	* gcc.target/loongarch/vfcmp-f.c: New test.
	* gcc.target/loongarch/vfcmp-d.c: New test.
	* gcc.target/loongarch/xvfcmp-f.c: New test.
	* gcc.target/loongarch/xvfcmp-d.c: New test.
	* gcc.target/loongarch/vector/lasx/lasx-vcond-2.c: Scan for cune
	instead of cne.
	* gcc.target/loongarch/vector/lsx/lsx-vcond-2.c: Likewise.
This commit is contained in:
Xi Ruoyao 2023-12-17 01:09:20 +08:00
parent 3dd6f73a23
commit c5651e9bc0
No known key found for this signature in database
GPG Key ID: ACAAD20E19E710E3
10 changed files with 385 additions and 222 deletions

View File

@ -32,9 +32,7 @@
UNSPEC_LASX_XVBITREVI
UNSPEC_LASX_XVBITSET
UNSPEC_LASX_XVBITSETI
UNSPEC_LASX_XVFCMP_CAF
UNSPEC_LASX_XVFCLASS
UNSPEC_LASX_XVFCMP_CUNE
UNSPEC_LASX_XVFCVT
UNSPEC_LASX_XVFCVTH
UNSPEC_LASX_XVFCVTL
@ -44,17 +42,6 @@
UNSPEC_LASX_XVFRINT
UNSPEC_LASX_XVFRSQRT
UNSPEC_LASX_XVFRSQRTE
UNSPEC_LASX_XVFCMP_SAF
UNSPEC_LASX_XVFCMP_SEQ
UNSPEC_LASX_XVFCMP_SLE
UNSPEC_LASX_XVFCMP_SLT
UNSPEC_LASX_XVFCMP_SNE
UNSPEC_LASX_XVFCMP_SOR
UNSPEC_LASX_XVFCMP_SUEQ
UNSPEC_LASX_XVFCMP_SULE
UNSPEC_LASX_XVFCMP_SULT
UNSPEC_LASX_XVFCMP_SUN
UNSPEC_LASX_XVFCMP_SUNE
UNSPEC_LASX_XVFTINT_U
UNSPEC_LASX_XVCLO
UNSPEC_LASX_XVSAT_S
@ -1481,69 +1468,6 @@
[(set_attr "type" "simd_fclass")
(set_attr "mode" "<MODE>")])
(define_insn "lasx_xvfcmp_caf_<flasxfmt>"
[(set (match_operand:<VIMODE256> 0 "register_operand" "=f")
(unspec:<VIMODE256> [(match_operand:FLASX 1 "register_operand" "f")
(match_operand:FLASX 2 "register_operand" "f")]
UNSPEC_LASX_XVFCMP_CAF))]
"ISA_HAS_LASX"
"xvfcmp.caf.<flasxfmt>\t%u0,%u1,%u2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_insn "lasx_xvfcmp_cune_<FLASX:flasxfmt>"
[(set (match_operand:<VIMODE256> 0 "register_operand" "=f")
(unspec:<VIMODE256> [(match_operand:FLASX 1 "register_operand" "f")
(match_operand:FLASX 2 "register_operand" "f")]
UNSPEC_LASX_XVFCMP_CUNE))]
"ISA_HAS_LASX"
"xvfcmp.cune.<FLASX:flasxfmt>\t%u0,%u1,%u2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_int_iterator FSC256_UNS [UNSPEC_LASX_XVFCMP_SAF UNSPEC_LASX_XVFCMP_SUN
UNSPEC_LASX_XVFCMP_SOR UNSPEC_LASX_XVFCMP_SEQ
UNSPEC_LASX_XVFCMP_SNE UNSPEC_LASX_XVFCMP_SUEQ
UNSPEC_LASX_XVFCMP_SUNE UNSPEC_LASX_XVFCMP_SULE
UNSPEC_LASX_XVFCMP_SULT UNSPEC_LASX_XVFCMP_SLE
UNSPEC_LASX_XVFCMP_SLT])
(define_int_attr fsc256
[(UNSPEC_LASX_XVFCMP_SAF "saf")
(UNSPEC_LASX_XVFCMP_SUN "sun")
(UNSPEC_LASX_XVFCMP_SOR "sor")
(UNSPEC_LASX_XVFCMP_SEQ "seq")
(UNSPEC_LASX_XVFCMP_SNE "sne")
(UNSPEC_LASX_XVFCMP_SUEQ "sueq")
(UNSPEC_LASX_XVFCMP_SUNE "sune")
(UNSPEC_LASX_XVFCMP_SULE "sule")
(UNSPEC_LASX_XVFCMP_SULT "sult")
(UNSPEC_LASX_XVFCMP_SLE "sle")
(UNSPEC_LASX_XVFCMP_SLT "slt")])
(define_insn "lasx_xvfcmp_<vfcond:fcc>_<FLASX:flasxfmt>"
[(set (match_operand:<VIMODE256> 0 "register_operand" "=f")
(vfcond:<VIMODE256> (match_operand:FLASX 1 "register_operand" "f")
(match_operand:FLASX 2 "register_operand" "f")))]
"ISA_HAS_LASX"
"xvfcmp.<vfcond:fcc>.<FLASX:flasxfmt>\t%u0,%u1,%u2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_insn "lasx_xvfcmp_<fsc256>_<FLASX:flasxfmt>"
[(set (match_operand:<VIMODE256> 0 "register_operand" "=f")
(unspec:<VIMODE256> [(match_operand:FLASX 1 "register_operand" "f")
(match_operand:FLASX 2 "register_operand" "f")]
FSC256_UNS))]
"ISA_HAS_LASX"
"xvfcmp.<fsc256>.<FLASX:flasxfmt>\t%u0,%u1,%u2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_mode_attr fint256
[(V8SF "v8si")
(V4DF "v4di")])

View File

@ -11161,7 +11161,6 @@ static void
loongarch_expand_lsx_cmp (rtx dest, enum rtx_code cond, rtx op0, rtx op1)
{
machine_mode cmp_mode = GET_MODE (op0);
int unspec = -1;
bool negate = false;
switch (cmp_mode)
@ -11203,66 +11202,9 @@ loongarch_expand_lsx_cmp (rtx dest, enum rtx_code cond, rtx op0, rtx op1)
case E_V4SFmode:
case E_V2DFmode:
switch (cond)
{
case UNORDERED:
case ORDERED:
case EQ:
case NE:
case UNEQ:
case UNLE:
case UNLT:
break;
case LTGT: cond = NE; break;
case UNGE: cond = UNLE; std::swap (op0, op1); break;
case UNGT: cond = UNLT; std::swap (op0, op1); break;
case LE: unspec = UNSPEC_LSX_VFCMP_SLE; break;
case LT: unspec = UNSPEC_LSX_VFCMP_SLT; break;
case GE: unspec = UNSPEC_LSX_VFCMP_SLE; std::swap (op0, op1); break;
case GT: unspec = UNSPEC_LSX_VFCMP_SLT; std::swap (op0, op1); break;
default:
gcc_unreachable ();
}
if (unspec < 0)
loongarch_emit_binary (cond, dest, op0, op1);
else
{
rtx x = gen_rtx_UNSPEC (GET_MODE (dest),
gen_rtvec (2, op0, op1), unspec);
emit_insn (gen_rtx_SET (dest, x));
}
break;
case E_V8SFmode:
case E_V4DFmode:
switch (cond)
{
case UNORDERED:
case ORDERED:
case EQ:
case NE:
case UNEQ:
case UNLE:
case UNLT:
break;
case LTGT: cond = NE; break;
case UNGE: cond = UNLE; std::swap (op0, op1); break;
case UNGT: cond = UNLT; std::swap (op0, op1); break;
case LE: unspec = UNSPEC_LASX_XVFCMP_SLE; break;
case LT: unspec = UNSPEC_LASX_XVFCMP_SLT; break;
case GE: unspec = UNSPEC_LASX_XVFCMP_SLE; std::swap (op0, op1); break;
case GT: unspec = UNSPEC_LASX_XVFCMP_SLT; std::swap (op0, op1); break;
default:
gcc_unreachable ();
}
if (unspec < 0)
loongarch_emit_binary (cond, dest, op0, op1);
else
{
rtx x = gen_rtx_UNSPEC (GET_MODE (dest),
gen_rtvec (2, op0, op1), unspec);
emit_insn (gen_rtx_SET (dest, x));
}
loongarch_emit_binary (cond, dest, op0, op1);
break;
default:

View File

@ -34,9 +34,7 @@
UNSPEC_LSX_VBITSETI
UNSPEC_LSX_BRANCH_V
UNSPEC_LSX_BRANCH
UNSPEC_LSX_VFCMP_CAF
UNSPEC_LSX_VFCLASS
UNSPEC_LSX_VFCMP_CUNE
UNSPEC_LSX_VFCVT
UNSPEC_LSX_VFCVTH
UNSPEC_LSX_VFCVTL
@ -46,17 +44,6 @@
UNSPEC_LSX_VFRINT
UNSPEC_LSX_VFRSQRT
UNSPEC_LSX_VFRSQRTE
UNSPEC_LSX_VFCMP_SAF
UNSPEC_LSX_VFCMP_SEQ
UNSPEC_LSX_VFCMP_SLE
UNSPEC_LSX_VFCMP_SLT
UNSPEC_LSX_VFCMP_SNE
UNSPEC_LSX_VFCMP_SOR
UNSPEC_LSX_VFCMP_SUEQ
UNSPEC_LSX_VFCMP_SULE
UNSPEC_LSX_VFCMP_SULT
UNSPEC_LSX_VFCMP_SUN
UNSPEC_LSX_VFCMP_SUNE
UNSPEC_LSX_VFTINT_U
UNSPEC_LSX_VSAT_S
UNSPEC_LSX_VSAT_U
@ -1377,76 +1364,6 @@
[(set_attr "type" "simd_fclass")
(set_attr "mode" "<MODE>")])
(define_insn "lsx_vfcmp_caf_<flsxfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(unspec:<VIMODE> [(match_operand:FLSX 1 "register_operand" "f")
(match_operand:FLSX 2 "register_operand" "f")]
UNSPEC_LSX_VFCMP_CAF))]
"ISA_HAS_LSX"
"vfcmp.caf.<flsxfmt>\t%w0,%w1,%w2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_insn "lsx_vfcmp_cune_<FLSX:flsxfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(unspec:<VIMODE> [(match_operand:FLSX 1 "register_operand" "f")
(match_operand:FLSX 2 "register_operand" "f")]
UNSPEC_LSX_VFCMP_CUNE))]
"ISA_HAS_LSX"
"vfcmp.cune.<FLSX:flsxfmt>\t%w0,%w1,%w2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_code_iterator vfcond [unordered ordered eq ne le lt uneq unle unlt])
(define_code_attr fcc
[(unordered "cun")
(ordered "cor")
(eq "ceq")
(ne "cne")
(uneq "cueq")
(unle "cule")
(unlt "cult")
(le "cle")
(lt "clt")])
(define_int_iterator FSC_UNS [UNSPEC_LSX_VFCMP_SAF UNSPEC_LSX_VFCMP_SUN UNSPEC_LSX_VFCMP_SOR
UNSPEC_LSX_VFCMP_SEQ UNSPEC_LSX_VFCMP_SNE UNSPEC_LSX_VFCMP_SUEQ
UNSPEC_LSX_VFCMP_SUNE UNSPEC_LSX_VFCMP_SULE UNSPEC_LSX_VFCMP_SULT
UNSPEC_LSX_VFCMP_SLE UNSPEC_LSX_VFCMP_SLT])
(define_int_attr fsc
[(UNSPEC_LSX_VFCMP_SAF "saf")
(UNSPEC_LSX_VFCMP_SUN "sun")
(UNSPEC_LSX_VFCMP_SOR "sor")
(UNSPEC_LSX_VFCMP_SEQ "seq")
(UNSPEC_LSX_VFCMP_SNE "sne")
(UNSPEC_LSX_VFCMP_SUEQ "sueq")
(UNSPEC_LSX_VFCMP_SUNE "sune")
(UNSPEC_LSX_VFCMP_SULE "sule")
(UNSPEC_LSX_VFCMP_SULT "sult")
(UNSPEC_LSX_VFCMP_SLE "sle")
(UNSPEC_LSX_VFCMP_SLT "slt")])
(define_insn "lsx_vfcmp_<vfcond:fcc>_<FLSX:flsxfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(vfcond:<VIMODE> (match_operand:FLSX 1 "register_operand" "f")
(match_operand:FLSX 2 "register_operand" "f")))]
"ISA_HAS_LSX"
"vfcmp.<vfcond:fcc>.<FLSX:flsxfmt>\t%w0,%w1,%w2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_insn "lsx_vfcmp_<fsc>_<FLSX:flsxfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(unspec:<VIMODE> [(match_operand:FLSX 1 "register_operand" "f")
(match_operand:FLSX 2 "register_operand" "f")]
FSC_UNS))]
"ISA_HAS_LSX"
"vfcmp.<fsc>.<FLSX:flsxfmt>\t%w0,%w1,%w2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_mode_attr fint
[(V4SF "v4si")
(V2DF "v2di")])

View File

@ -279,6 +279,124 @@
[(set_attr "type" "simd_int_arith")
(set_attr "mode" "<MODE>")])
;; <x>vfcmp.*.{s/d} with defined RTX code
;; There are no fcmp.{sugt/suge/cgt/cge}.{s/d} menmonics in GAS, so we have
;; to reverse the operands ourselves :(.
(define_code_iterator fcond_simd [unordered uneq unlt unle eq lt le
ordered ltgt ne])
(define_insn "<simd_isa>_<x>vfcmp_<fcond>_<simdfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(fcond_simd:<VIMODE>
(match_operand:FVEC 1 "register_operand" "f")
(match_operand:FVEC 2 "register_operand" "f")))]
""
"<x>vfcmp.<fcond>.<simdfmt>\t%<wu>0,%<wu>1,%<wu>2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
;; There are no fcmp.{sge/sgt/cuge/cugt}.{s/d} menmonics in GAS, so we have
;; to reverse the operands ourselves.
(define_code_iterator fcond_simd_rev [ge gt unge ungt])
(define_code_attr fcond_rev_asm
[(ge "sle")
(gt "slt")
(unge "cule")
(ungt "cult")])
(define_insn "<simd_isa>_<x>vfcmp_<fcond>_<simdfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(fcond_simd_rev:<VIMODE>
(match_operand:FVEC 1 "register_operand" "f")
(match_operand:FVEC 2 "register_operand" "f")))]
""
"<x>vfcmp.<fcond_rev_asm>.<simdfmt>\t%<wu>0,%<wu>2,%<wu>1";
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
;; <x>vfcmp.*.{s/d} without defined RTX code, but with defined RTX code for
;; its inverse. Again, there are no fcmp.{sugt/suge/cgt/cge}.{s/d}
;; menmonics in GAS, so we have to reverse the operands ourselves.
(define_code_iterator fcond_inv [ge gt unge ungt])
(define_code_iterator fcond_inv_rev [le lt unle unlt])
(define_code_attr fcond_inv
[(ge "sult")
(gt "sule")
(unge "clt")
(ungt "cle")
(le "sugt")
(lt "suge")
(unle "cgt")
(unlt "cge")])
(define_code_attr fcond_inv_rev_asm
[(le "sult")
(lt "sule")
(unle "clt")
(unlt "cle")])
(define_insn "<simd_isa>_<x>vfcmp_<fcond_inv>_<simdfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(not:<VIMODE>
(fcond_inv:<VIMODE>
(match_operand:FVEC 1 "register_operand" "f")
(match_operand:FVEC 2 "register_operand" "f"))))]
""
"<x>vfcmp.<fcond_inv>.<simdfmt>\t%<wu>0,%<wu>1,%<wu>2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
(define_insn "<simd_isa>_<x>vfcmp_<fcond_inv>_<simdfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(not:<VIMODE>
(fcond_inv_rev:<VIMODE>
(match_operand:FVEC 1 "register_operand" "f")
(match_operand:FVEC 2 "register_operand" "f"))))]
""
"<x>vfcmp.<fcond_inv_rev_asm>.<simdfmt>\t%<wu>0,%<wu>2,%<wu>1"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
;; <x>vfcmp.*.{s/d} instructions only as instrinsics
(define_c_enum "unspec"
[UNSPEC_SIMD_FCMP_CAF
UNSPEC_SIMD_FCMP_SAF
UNSPEC_SIMD_FCMP_SEQ
UNSPEC_SIMD_FCMP_SUN
UNSPEC_SIMD_FCMP_SUEQ
UNSPEC_SIMD_FCMP_CNE
UNSPEC_SIMD_FCMP_SOR
UNSPEC_SIMD_FCMP_SUNE])
(define_int_iterator SIMD_FCMP
[UNSPEC_SIMD_FCMP_CAF
UNSPEC_SIMD_FCMP_SAF
UNSPEC_SIMD_FCMP_SEQ
UNSPEC_SIMD_FCMP_SUN
UNSPEC_SIMD_FCMP_SUEQ
UNSPEC_SIMD_FCMP_CNE
UNSPEC_SIMD_FCMP_SOR
UNSPEC_SIMD_FCMP_SUNE])
(define_int_attr fcond_unspec
[(UNSPEC_SIMD_FCMP_CAF "caf")
(UNSPEC_SIMD_FCMP_SAF "saf")
(UNSPEC_SIMD_FCMP_SEQ "seq")
(UNSPEC_SIMD_FCMP_SUN "sun")
(UNSPEC_SIMD_FCMP_SUEQ "sueq")
(UNSPEC_SIMD_FCMP_CNE "cne")
(UNSPEC_SIMD_FCMP_SOR "sor")
(UNSPEC_SIMD_FCMP_SUNE "sune")])
(define_insn "<simd_isa>_<x>vfcmp_<fcond_unspec>_<simdfmt>"
[(set (match_operand:<VIMODE> 0 "register_operand" "=f")
(unspec:<VIMODE> [(match_operand:FVEC 1 "register_operand" "f")
(match_operand:FVEC 2 "register_operand" "f")]
SIMD_FCMP))]
""
"<x>vfcmp.<fcond_unspec>.<simdfmt>\t%<wu>0,%<wu>1,%<wu>2"
[(set_attr "type" "simd_fcmp")
(set_attr "mode" "<MODE>")])
; The LoongArch SX Instructions.
(include "lsx.md")

View File

@ -69,8 +69,8 @@ TEST_CMP (nugt)
/* { dg-final { scan-assembler-times {\txvfcmp\.ceq\.s} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.ceq\.d} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.cne\.s} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.cne\.d} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.cune\.s} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.cune\.d} 3 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.slt\.s} 6 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.slt\.d} 6 } } */
/* { dg-final { scan-assembler-times {\txvfcmp\.sle\.s} 6 } } */

View File

@ -69,8 +69,8 @@ TEST_CMP (nugt)
/* { dg-final { scan-assembler-times {\tvfcmp\.ceq\.s} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.ceq\.d} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.cne\.s} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.cne\.d} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.cune\.s} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.cune\.d} 3 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.slt\.s} 6 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.slt\.d} 6 } } */
/* { dg-final { scan-assembler-times {\tvfcmp\.sle\.s} 6 } } */

View File

@ -0,0 +1,28 @@
/* { dg-do compile } */
/* { dg-options "-O2 -mlsx -ffixed-f0 -ffixed-f1 -ffixed-f2 -fno-vect-cost-model" } */
#define F double
#define I long long
#include "vfcmp-f.c"
/* { dg-final { scan-assembler "compare_quiet_equal:.*\tvfcmp\\.ceq\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_equal:.*\tvfcmp\\.cune\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_not_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater:.*\tvfcmp\\.slt\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_equal:.*\tvfcmp\\.sle\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less:.*\tvfcmp\\.slt\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_equal:.*\tvfcmp\\.sle\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_greater:.*\tvfcmp\\.sule\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_unordered:.*\tvfcmp\\.sult\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_less:.*\tvfcmp\\.sule\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_not_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_unordered:.*\tvfcmp\\.sult\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less:.*\tvfcmp\\.clt\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_equal:.*\tvfcmp\\.cle\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater:.*\tvfcmp\\.clt\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_equal:.*\tvfcmp\\.cle\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_less:.*\tvfcmp\\.cule\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_not_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_unordered:.*\tvfcmp\\.cult\\.d\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_greater:.*\tvfcmp\\.cule\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_unordered:.*\tvfcmp\\.cult\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_unordered:.*\tvfcmp\\.cun\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_ordered:.*\tvfcmp\\.cor\\.d\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_ordered\n" } } */

View File

@ -0,0 +1,178 @@
/* Test mapping IEC 60559 operations to SIMD instructions.
For details read C23 Annex F.3 and LoongArch Vol. 1 section 3.2.2.1. */
/* { dg-do compile } */
/* { dg-options "-O2 -mlsx -ffixed-f0 -ffixed-f1 -ffixed-f2 -fno-vect-cost-model" } */
#ifndef F
#define F float
#endif
#ifndef I
#define I int
#endif
#ifndef VL
#define VL 16
#endif
typedef F VF __attribute__ ((vector_size (VL)));
typedef I VI __attribute__ ((vector_size (VL)));
register VF a asm ("f0");
register VF b asm ("f1");
register VI c asm ("f2");
void
compare_quiet_equal (void)
{
c = (a == b);
}
void
compare_quiet_not_equal (void)
{
c = (a != b);
}
void
compare_signaling_greater (void)
{
c = (a > b);
}
void
compare_signaling_greater_equal (void)
{
c = (a >= b);
}
void
compare_signaling_less (void)
{
c = (a < b);
}
void
compare_signaling_less_equal (void)
{
c = (a <= b);
}
void
compare_signaling_not_greater (void)
{
c = ~(a > b);
}
void
compare_signaling_less_unordered (void)
{
c = ~(a >= b);
}
void
compare_signaling_not_less (void)
{
c = ~(a < b);
}
void
compare_signaling_greater_unordered (void)
{
c = ~(a <= b);
}
void
compare_quiet_less (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isless (a[i], b[i]) ? -1 : 0;
}
void
compare_quiet_less_equal (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_islessequal (a[i], b[i]) ? -1 : 0;
}
void
compare_quiet_greater (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isgreater (a[i], b[i]) ? -1 : 0;
}
void
compare_quiet_greater_equal (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isgreaterequal (a[i], b[i]) ? -1 : 0;
}
void
compare_quiet_not_less (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isless (a[i], b[i]) ? 0 : -1;
}
void
compare_quiet_greater_unordered (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_islessequal (a[i], b[i]) ? 0 : -1;
}
void
compare_quiet_not_greater (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isgreater (a[i], b[i]) ? 0 : -1;
}
void
compare_quiet_less_unordered (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isgreaterequal (a[i], b[i]) ? 0 : -1;
}
void
compare_quiet_unordered (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isunordered (a[i], b[i]) ? -1 : 0;
}
void
compare_quiet_ordered (void)
{
for (int i = 0; i < sizeof (c) / sizeof (c[0]); i++)
c[i] = __builtin_isunordered (a[i], b[i]) ? 0 : -1;
}
/* The "-<function_name>" matches the .size directive after the function
body, so we can ensure the instruction is in the correct function. */
/* { dg-final { scan-assembler "compare_quiet_equal:.*\tvfcmp\\.ceq\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_equal:.*\tvfcmp\\.cune\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_not_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater:.*\tvfcmp\\.slt\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_equal:.*\tvfcmp\\.sle\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less:.*\tvfcmp\\.slt\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_equal:.*\tvfcmp\\.sle\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_greater:.*\tvfcmp\\.sule\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_unordered:.*\tvfcmp\\.sult\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_signaling_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_less:.*\tvfcmp\\.sule\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_not_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_unordered:.*\tvfcmp\\.sult\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_signaling_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less:.*\tvfcmp\\.clt\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_equal:.*\tvfcmp\\.cle\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater:.*\tvfcmp\\.clt\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_equal:.*\tvfcmp\\.cle\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_less:.*\tvfcmp\\.cule\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_not_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_unordered:.*\tvfcmp\\.cult\\.s\t\\\$vr2,\\\$vr1,\\\$vr0.*-compare_quiet_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_greater:.*\tvfcmp\\.cule\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_unordered:.*\tvfcmp\\.cult\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_unordered:.*\tvfcmp\\.cun\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_ordered:.*\tvfcmp\\.cor\\.s\t\\\$vr2,\\\$vr0,\\\$vr1.*-compare_quiet_ordered\n" } } */

View File

@ -0,0 +1,29 @@
/* { dg-do compile } */
/* { dg-options "-O2 -mlasx -ffixed-f0 -ffixed-f1 -ffixed-f2 -fno-vect-cost-model" } */
#define F double
#define I long long
#define VL 32
#include "vfcmp-f.c"
/* { dg-final { scan-assembler "compare_quiet_equal:.*\txvfcmp\\.ceq\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_equal:.*\txvfcmp\\.cune\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_not_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater:.*\txvfcmp\\.slt\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_equal:.*\txvfcmp\\.sle\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less:.*\txvfcmp\\.slt\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_equal:.*\txvfcmp\\.sle\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_greater:.*\txvfcmp\\.sule\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_unordered:.*\txvfcmp\\.sult\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_less:.*\txvfcmp\\.sule\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_not_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_unordered:.*\txvfcmp\\.sult\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less:.*\txvfcmp\\.clt\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_equal:.*\txvfcmp\\.cle\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater:.*\txvfcmp\\.clt\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_equal:.*\txvfcmp\\.cle\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_less:.*\txvfcmp\\.cule\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_not_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_unordered:.*\txvfcmp\\.cult\\.d\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_greater:.*\txvfcmp\\.cule\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_unordered:.*\txvfcmp\\.cult\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_unordered:.*\txvfcmp\\.cun\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_ordered:.*\txvfcmp\\.cor\\.d\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_ordered\n" } } */

View File

@ -0,0 +1,27 @@
/* { dg-do compile } */
/* { dg-options "-O2 -mlasx -ffixed-f0 -ffixed-f1 -ffixed-f2" } */
#define VL 32
#include "vfcmp-f.c"
/* { dg-final { scan-assembler "compare_quiet_equal:.*\txvfcmp\\.ceq\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_equal:.*\txvfcmp\\.cune\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_not_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater:.*\txvfcmp\\.slt\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_equal:.*\txvfcmp\\.sle\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less:.*\txvfcmp\\.slt\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_equal:.*\txvfcmp\\.sle\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_greater:.*\txvfcmp\\.sule\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_signaling_less_unordered:.*\txvfcmp\\.sult\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_signaling_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_signaling_not_less:.*\txvfcmp\\.sule\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_not_less\n" } } */
/* { dg-final { scan-assembler "compare_signaling_greater_unordered:.*\txvfcmp\\.sult\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_signaling_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less:.*\txvfcmp\\.clt\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_equal:.*\txvfcmp\\.cle\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater:.*\txvfcmp\\.clt\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_equal:.*\txvfcmp\\.cle\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater_equal\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_less:.*\txvfcmp\\.cule\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_not_less\n" } } */
/* { dg-final { scan-assembler "compare_quiet_greater_unordered:.*\txvfcmp\\.cult\\.s\t\\\$xr2,\\\$xr1,\\\$xr0.*-compare_quiet_greater_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_not_greater:.*\txvfcmp\\.cule\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_not_greater\n" } } */
/* { dg-final { scan-assembler "compare_quiet_less_unordered:.*\txvfcmp\\.cult\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_less_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_unordered:.*\txvfcmp\\.cun\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_unordered\n" } } */
/* { dg-final { scan-assembler "compare_quiet_ordered:.*\txvfcmp\\.cor\\.s\t\\\$xr2,\\\$xr0,\\\$xr1.*-compare_quiet_ordered\n" } } */