OSchip
/
llvm-project
mirror of https://github.com/THU-DSP-LAB/llvm-project.git
13
0
Fork 6
Code Issues Packages Projects Releases Wiki Activity

[X86][Atom] Fix (U)COMISS/SD uops, latency and throughput 

Both ports are required, for reg and mem variants - we can also use the WriteFComX class directly and remove the unnecessary InstRW overrides. Matches what Intel AoM / Agner / InstLatX64 report as well.
This commit is contained in:
Simon Pilgrim 2021-09-18 21:29:33 +01:00
parent b7ec8f3dcb
commit e381d8b243
3 changed files with 21 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
llvm/lib/Target/X86/X86ScheduleAtom.td
View File

@ -249,13 +249,13 @@ defm : AtomWriteResPair<WriteFCmp64X, [AtomPort0,AtomPort1], [AtomPort0,A
defm : X86WriteResPairUnsupported<WriteFCmp64Y>;
defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
defm : AtomWriteResPair<WriteFCom, [AtomPort0], [AtomPort0], 5, 5, [5], [5]>;
defm : AtomWriteResPair<WriteFComX, [AtomPort0], [AtomPort0], 5, 5, [5], [5]>;
defm : AtomWriteResPair<WriteFComX, [AtomPort0,AtomPort1], [AtomPort0,AtomPort1], 9, 10, [9,9],[10,10], 4, 5>;
defm : AtomWriteResPair<WriteFMul, [AtomPort0], [AtomPort0], 4, 4, [2], [2]>;
defm : AtomWriteResPair<WriteFMulX, [AtomPort0], [AtomPort0], 5, 5, [2], [2]>;
defm : X86WriteResPairUnsupported<WriteFMulY>;
defm : X86WriteResPairUnsupported<WriteFMulZ>;
defm : AtomWriteResPair<WriteFMul64, [AtomPort0], [AtomPort0], 5, 5, [2], [2]>;
defm : AtomWriteResPair<WriteFMul64X, [AtomPort0,AtomPort1], [AtomPort0,AtomPort1], 9, 10, [9,9], [10,10]>;
defm : AtomWriteResPair<WriteFMul64X, [AtomPort0,AtomPort1], [AtomPort0,AtomPort1], 9, 10, [9,9],[10,10]>;
defm : X86WriteResPairUnsupported<WriteFMul64Y>;
defm : X86WriteResPairUnsupported<WriteFMul64Z>;
defm : AtomWriteResPair<WriteFRcp, [AtomPort0], [AtomPort0], 4, 4, [4], [4]>;
@ -648,7 +648,6 @@ def : InstRW<[AtomWrite01_9], (instrs POPA16, POPA32,
SHLD64rri8, SHRD64rri8,
CMPXCHG8rr)>;
def : InstRW<[AtomWrite01_9], (instregex "(U)?COM_FI", "TST_F",
"(U)?COMIS(D|S)rr",
"CVT(T)?SS2SI64rr(_Int)?")>;
def AtomWrite01_10 : SchedWriteRes<[AtomPort01]> {
@ -656,8 +655,7 @@ def AtomWrite01_10 : SchedWriteRes<[AtomPort01]> {
let ResourceCycles = [10];
}
def : SchedAlias<WriteFLDC, AtomWrite01_10>;
def : InstRW<[AtomWrite01_10], (instregex "(U)?COMIS(D|S)rm",
"CVT(T)?SS2SI64rm(_Int)?")>;
def : InstRW<[AtomWrite01_10], (instregex "CVT(T)?SS2SI64rm(_Int)?")>;
def AtomWrite01_11 : SchedWriteRes<[AtomPort01]> {
let Latency = 11;

18
llvm/test/tools/llvm-mca/X86/Atom/resources-sse1.s
View File

@ -206,8 +206,8 @@ xorps (%rax), %xmm2
# CHECK-NEXT: 1 7 6.00 * cmpeqps (%rax), %xmm2
# CHECK-NEXT: 1 5 1.00 cmpeqss %xmm0, %xmm2
# CHECK-NEXT: 1 5 1.00 * cmpeqss (%rax), %xmm2
# CHECK-NEXT: 1 9 4.50 comiss %xmm0, %xmm1
# CHECK-NEXT: 1 10 5.00 * comiss (%rax), %xmm1
# CHECK-NEXT: 4 9 9.00 comiss %xmm0, %xmm1
# CHECK-NEXT: 5 10 10.00 * comiss (%rax), %xmm1
# CHECK-NEXT: 1 5 5.00 cvtpi2ps %mm0, %xmm2
# CHECK-NEXT: 1 5 5.00 * cvtpi2ps (%rax), %xmm2
# CHECK-NEXT: 1 5 5.00 cvtps2pi %xmm0, %mm2
@ -310,8 +310,8 @@ xorps (%rax), %xmm2
# CHECK-NEXT: 1 5 1.00 * subps (%rax), %xmm2
# CHECK-NEXT: 1 5 1.00 subss %xmm0, %xmm2
# CHECK-NEXT: 1 5 1.00 * subss (%rax), %xmm2
# CHECK-NEXT: 1 9 4.50 ucomiss %xmm0, %xmm1
# CHECK-NEXT: 1 10 5.00 * ucomiss (%rax), %xmm1
# CHECK-NEXT: 4 9 9.00 ucomiss %xmm0, %xmm1
# CHECK-NEXT: 5 10 10.00 * ucomiss (%rax), %xmm1
# CHECK-NEXT: 1 1 1.00 unpckhps %xmm0, %xmm2
# CHECK-NEXT: 1 1 1.00 * unpckhps (%rax), %xmm2
# CHECK-NEXT: 1 1 1.00 unpcklps %xmm0, %xmm2
@ -325,7 +325,7 @@ xorps (%rax), %xmm2
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1]
# CHECK-NEXT: 456.00 411.00
# CHECK-NEXT: 475.00 430.00
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] Instructions:
@ -341,8 +341,8 @@ xorps (%rax), %xmm2
# CHECK-NEXT: 6.00 6.00 cmpeqps (%rax), %xmm2
# CHECK-NEXT: - 1.00 cmpeqss %xmm0, %xmm2
# CHECK-NEXT: 1.00 1.00 cmpeqss (%rax), %xmm2
# CHECK-NEXT: 4.50 4.50 comiss %xmm0, %xmm1
# CHECK-NEXT: 5.00 5.00 comiss (%rax), %xmm1
# CHECK-NEXT: 9.00 9.00 comiss %xmm0, %xmm1
# CHECK-NEXT: 10.00 10.00 comiss (%rax), %xmm1
# CHECK-NEXT: - 5.00 cvtpi2ps %mm0, %xmm2
# CHECK-NEXT: - 5.00 cvtpi2ps (%rax), %xmm2
# CHECK-NEXT: - 5.00 cvtps2pi %xmm0, %mm2
@ -445,8 +445,8 @@ xorps (%rax), %xmm2
# CHECK-NEXT: 1.00 1.00 subps (%rax), %xmm2
# CHECK-NEXT: - 1.00 subss %xmm0, %xmm2
# CHECK-NEXT: 1.00 1.00 subss (%rax), %xmm2
# CHECK-NEXT: 4.50 4.50 ucomiss %xmm0, %xmm1
# CHECK-NEXT: 5.00 5.00 ucomiss (%rax), %xmm1
# CHECK-NEXT: 9.00 9.00 ucomiss %xmm0, %xmm1
# CHECK-NEXT: 10.00 10.00 ucomiss (%rax), %xmm1
# CHECK-NEXT: 1.00 - unpckhps %xmm0, %xmm2
# CHECK-NEXT: 1.00 - unpckhps (%rax), %xmm2
# CHECK-NEXT: 1.00 - unpcklps %xmm0, %xmm2

18
llvm/test/tools/llvm-mca/X86/Atom/resources-sse2.s
View File

@ -420,8 +420,8 @@ xorpd (%rax), %xmm2
# CHECK-NEXT: 1 7 6.00 * cmpeqpd (%rax), %xmm2
# CHECK-NEXT: 1 5 1.00 cmpeqsd %xmm0, %xmm2
# CHECK-NEXT: 1 5 1.00 * cmpeqsd (%rax), %xmm2
# CHECK-NEXT: 1 9 4.50 comisd %xmm0, %xmm1
# CHECK-NEXT: 1 10 5.00 * comisd (%rax), %xmm1
# CHECK-NEXT: 4 9 9.00 comisd %xmm0, %xmm1
# CHECK-NEXT: 5 10 10.00 * comisd (%rax), %xmm1
# CHECK-NEXT: 1 7 6.00 cvtdq2pd %xmm0, %xmm2
# CHECK-NEXT: 1 8 7.00 * cvtdq2pd (%rax), %xmm2
# CHECK-NEXT: 1 6 5.00 cvtdq2ps %xmm0, %xmm2
@ -666,8 +666,8 @@ xorpd (%rax), %xmm2
# CHECK-NEXT: 1 7 6.00 * subpd (%rax), %xmm2
# CHECK-NEXT: 1 5 1.00 subsd %xmm0, %xmm2
# CHECK-NEXT: 1 5 1.00 * subsd (%rax), %xmm2
# CHECK-NEXT: 1 9 4.50 ucomisd %xmm0, %xmm1
# CHECK-NEXT: 1 10 5.00 * ucomisd (%rax), %xmm1
# CHECK-NEXT: 4 9 9.00 ucomisd %xmm0, %xmm1
# CHECK-NEXT: 5 10 10.00 * ucomisd (%rax), %xmm1
# CHECK-NEXT: 1 1 1.00 unpckhpd %xmm0, %xmm2
# CHECK-NEXT: 1 1 1.00 * unpckhpd (%rax), %xmm2
# CHECK-NEXT: 1 1 1.00 unpcklpd %xmm0, %xmm2
@ -681,7 +681,7 @@ xorpd (%rax), %xmm2
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1]
# CHECK-NEXT: 913.00 773.00
# CHECK-NEXT: 932.00 792.00
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] Instructions:
@ -698,8 +698,8 @@ xorpd (%rax), %xmm2
# CHECK-NEXT: 6.00 6.00 cmpeqpd (%rax), %xmm2
# CHECK-NEXT: - 1.00 cmpeqsd %xmm0, %xmm2
# CHECK-NEXT: 1.00 1.00 cmpeqsd (%rax), %xmm2
# CHECK-NEXT: 4.50 4.50 comisd %xmm0, %xmm1
# CHECK-NEXT: 5.00 5.00 comisd (%rax), %xmm1
# CHECK-NEXT: 9.00 9.00 comisd %xmm0, %xmm1
# CHECK-NEXT: 10.00 10.00 comisd (%rax), %xmm1
# CHECK-NEXT: 6.00 6.00 cvtdq2pd %xmm0, %xmm2
# CHECK-NEXT: 7.00 7.00 cvtdq2pd (%rax), %xmm2
# CHECK-NEXT: 5.00 5.00 cvtdq2ps %xmm0, %xmm2
@ -944,8 +944,8 @@ xorpd (%rax), %xmm2
# CHECK-NEXT: 6.00 6.00 subpd (%rax), %xmm2
# CHECK-NEXT: - 1.00 subsd %xmm0, %xmm2
# CHECK-NEXT: 1.00 1.00 subsd (%rax), %xmm2
# CHECK-NEXT: 4.50 4.50 ucomisd %xmm0, %xmm1
# CHECK-NEXT: 5.00 5.00 ucomisd (%rax), %xmm1
# CHECK-NEXT: 9.00 9.00 ucomisd %xmm0, %xmm1
# CHECK-NEXT: 10.00 10.00 ucomisd (%rax), %xmm1
# CHECK-NEXT: 1.00 - unpckhpd %xmm0, %xmm2
# CHECK-NEXT: 1.00 - unpckhpd (%rax), %xmm2
# CHECK-NEXT: 1.00 - unpcklpd %xmm0, %xmm2