OV - exec - Function ljForce._omp

	Function: ljForce._omp_fn.1	Module: exec	Source: ljForce.c:172-216 [...]	Coverage (incl. loops): 90.81% \| (excl. loops): 0.00%

/home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c: 172 - 216

--------------------------------------------------------------------------------

172:    #pragma omp parallel for reduction(+:ePot)

173:    for (int iBox=0; iBox<s->boxes->nLocalBoxes; iBox++)

174:    {

175:       int nIBox = s->boxes->nAtoms[iBox];

176:

177:       // loop over neighbors of iBox

178:       for (int jTmp=0; jTmp<nNbrBoxes; jTmp++)

179:       {

180:          int jBox = s->boxes->nbrBoxes[iBox][jTmp];

181:

182:          assert(jBox>=0);

183:

184:          int nJBox = s->boxes->nAtoms[jBox];

185:

186:          // loop over atoms in iBox

187:          for (int iOff=MAXATOMS*iBox; iOff<(iBox*MAXATOMS+nIBox); iOff++)

188:          {

189:

190:             // loop over atoms in jBox

191:             for (int jOff=jBox*MAXATOMS; jOff<(jBox*MAXATOMS+nJBox); jOff++)

[...]

197:                   dr[m] = s->atoms->r[iOff][m]-s->atoms->r[jOff][m];

198:                   r2+=dr[m]*dr[m];

199:                }

200:

201:                if ( r2 <= rCut2 && r2 > 0.0)

202:                {

203:

204:                   // Important note:

205:                   // from this point on r actually refers to 1.0/r

206:                   r2 = 1.0/r2;

207:                   real_t r6 = s6 * (r2*r2*r2);

208:                   real_t eLocal = r6 * (r6 - 1.0) - eShift;

209:                   s->atoms->U[iOff] += 0.5*eLocal;

210:                   ePot += 0.5*eLocal;

211:

212:                   // different formulation to avoid sqrt computation

213:                   real_t fr = - 4.0*epsilon*r6*r2*(12.0*r6 - 6.0);

214:                   for (int m=0; m<3; m++)

215:                   {

216:                      s->atoms->f[iOff][m] -= dr[m]*fr;

0x417f0c STP	X29, X30, [SP, #928]!

0x417f10 ADD	X29, SP, #0

0x417f14 STP	X19, X20, [SP, #16]

0x417f18 ORR	X20, XZR, X0

0x417f1c STP	X21, X22, [SP, #32]

0x417f20 LDR	X19, [X0]

0x417f24 STR	D8, [SP, #80]

0x417f28 BL	410120

0x417f2c LDR	D8, [X20, #8]

0x417f30 ORR	W21, WZR, W0

0x417f34 BL	410140

0x417f38 LDR	X22, [X19, #24]

0x417f3c ORR	W13, WZR, W0

0x417f40 LDR	W30, [X20, #48]

0x417f44 LDP	D7, D21, [X20, #16]

0x417f48 LDR	W0, [X22, #12]

0x417f4c LDR	D20, [X20, #32]

0x417f50 SDIV	W15, W0, W21

0x417f54 MSUB	W1, W15, W21, W0

0x417f58 CMP	W13, W1

0x417f5c B.LT	418154

0x417f60 MADD	W2, W15, W13, W1

0x417f64 ADD	W15, W15, W2

0x417f68 CMP	W2, W15

0x417f6c B.GE	41816c

0x417f70 FMOV	D0, #-4.0000000

0x417f74 LDR	X21, [X22, #120]

0x417f78 MOVI	D16, #0

0x417f7c FMOV	D18, #1.0000000

0x417f80 FMOV	D17, #0.5000000

0x417f84 SBFM	X16, X2, #0, #31

0x417f88 FMOV	D23, #12.0000000

0x417f8c FMOV	D22, #-6.0000000

0x417f90 UBFM	W13, W2, #26, #25

0x417f94 STP	X23, X24, [SP, #48]

0x417f98 MOVZ	W23, #24

0x417f9c FMUL	D19, D8, D0

0x417fa0 STP	X25, X26, [SP, #64]

(90) 0x417fa4 LDR	W17, [X21, X16,LSL #2]

(90) 0x417fa8 CMP	W30, #0

(90) 0x417fac B.LE	418090

(90) 0x417fb0 UBFM	W18, W16, #26, #25

(90) 0x417fb4 SUB	W3, W17, #1

(90) 0x417fb8 LDR	X6, [X22, #128]

(90) 0x417fbc SBFM	X12, X18, #0, #31

(90) 0x417fc0 ADD	W17, W17, W13

(90) 0x417fc4 ADD	X4, X12, #1

(90) 0x417fc8 SBFM	X25, X18, #61, #31

(90) 0x417fcc ADD	X5, X4, X3

(90) 0x417fd0 MOVZ	X14, #0

(90) 0x417fd4 UBFM	X12, X5, #61, #60

(90) 0x417fd8 LDR	X24, [X6, X16,LSL #3]

  (92) 0x417fdc LDR	W9, [X24, X14,LSL #2]

  (92) 0x417fe0 TBNZ	W9, #31, 418174

  (92) 0x417fe4 LDR	W11, [X21, W9,SXTW #2]

  (92) 0x417fe8 CMP	W17, W18

  (92) 0x417fec B.LE	418084

  (92) 0x417ff0 UBFM	W9, W9, #26, #25

  (92) 0x417ff4 SUB	W26, W11, #1

  (92) 0x417ff8 ADD	X0, X26, W9,SXTW

  (92) 0x417ffc ADD	X1, X0, X0,LSL #1

  (92) 0x418000 ADD	W11, W11, W9

  (92) 0x418004 ORR	X4, XZR, X25

  (92) 0x418008 SMULL	X26, W9, W23

  (92) 0x41800c UBFM	X10, X1, #61, #60

    (93) 0x418010 CMP	W9, W11

    (93) 0x418014 B.GE	418078

    (93) 0x418018 LDR	X8, [X19, #32]

    (93) 0x41801c ADD	X3, X4, X4,LSL #1

    (93) 0x418020 LDR	X5, [X8, #24]

    (93) 0x418024 ADD	X6, X5, X3

    (93) 0x418028 LDR	D30, [X5, X3]

    (93) 0x41802c ADD	X2, X5, #24

    (93) 0x418030 ADD	X2, X2, X10

    (93) 0x418034 ADD	X0, X5, X26

    (93) 0x418038 LDP	D5, D31, [X6, #8]

    (93) 0x41803c HINT	#0

      (94) 0x418040 LDP	D2, D4, [X0]

      (94) 0x418044 LDR	D1, [X0, #16]

      (94) 0x418048 FSUB	D25, D5, S4

      (94) 0x41804c FSUB	D26, D30, S2

      (94) 0x418050 FSUB	D27, D31, S1

      (94) 0x418054 FMUL	D24, D25, D25

      (94) 0x418058 FMADD	D28, D26, D26, D24

      (94) 0x41805c FMADD	D29, D27, D27, D28

      (94) 0x418060 FCMPE	D29, #0

      (94) 0x418064 FCCMPE	D7, D29, #1, #12

      (94) 0x418068 B.GE	4180e0

      (94) 0x41806c ADD	X0, X0, #24

      (94) 0x418070 CMP	X0, X2

      (94) 0x418074 B.NE	418040

    (93) 0x418078 ADD	X4, X4, #8

    (93) 0x41807c CMP	X12, X4

    (93) 0x418080 B.NE	418010

  (92) 0x418084 ADD	X14, X14, #1

  (92) 0x418088 CMP	W30, W14

  (92) 0x41808c B.GT	417fdc

(90) 0x418090 ADD	X16, X16, #1

(90) 0x418094 ADD	W13, W13, #64

(90) 0x418098 CMP	W15, W16

(90) 0x41809c B.GT	417fa4

0x4180a0 LDP	X23, X24, [SP, #48]

0x4180a4 LDP	X25, X26, [SP, #64]

0x4180a8 ADD	X20, X20, #40

0x4180ac LDR	X30, [X20]

(91) 0x4180b0 ORR	X19, XZR, X30

(91) 0x4180b4 FMOV	D7, X30

(91) 0x4180b8 FADD	D21, D16, D7

(91) 0x4180bc FMOV	X22, D21

(91) 0x4180c0 CASAL	X30, X22, [X20]

(91) 0x4180c4 CMP	X19, X30

(91) 0x4180c8 B.NE	4180b0

0x4180cc LDP	X19, X20, [SP, #16]

0x4180d0 LDP	X21, X22, [SP, #32]

0x4180d4 LDR	D8, [SP, #80]

0x4180d8 LDP	X29, X30, [SP], #96

0x4180dc RET

      (94) 0x4180e0 FDIV	D6, D18, D29

      (94) 0x4180e4 LDP	X5, X7, [X8, #40]

      (94) 0x4180e8 ADD	X0, X0, #24

      (94) 0x4180ec LDR	D24, [X7, X4]

      (94) 0x4180f0 ADD	X1, X5, X3

      (94) 0x4180f4 FMUL	D3, D21, D6

      (94) 0x4180f8 FMUL	D5, D6, D6

      (94) 0x4180fc FMUL	D30, D19, D6

      (94) 0x418100 FMUL	D31, D3, D5

      (94) 0x418104 FMADD	D8, D31, D23, D22

      (94) 0x418108 FSUB	D0, D31, S18

      (94) 0x41810c FNMSUB	D2, D31, D0, D20

      (94) 0x418110 FMUL	D4, D31, D8

      (94) 0x418114 FMADD	D28, D2, D17, D24

      (94) 0x418118 FMADD	D16, D2, D17, D16

      (94) 0x41811c FMUL	D1, D4, D30

      (94) 0x418120 STR	D28, [X7, X4]

      (94) 0x418124 LDR	D29, [X5, X3]

      (94) 0x418128 FMSUB	D26, D1, D26, D29

      (94) 0x41812c STR	D26, [X5, X3]

      (94) 0x418130 LDP	D6, D3, [X1, #8]

      (94) 0x418134 FMSUB	D25, D1, D25, D6

      (94) 0x418138 FMSUB	D27, D1, D27, D3

      (94) 0x41813c STP	D25, D27, [X1, #8]

      (94) 0x418140 CMP	X0, X2

      (94) 0x418144 B.EQ	418078

      (94) 0x418148 LDP	D30, D5, [X6]

      (94) 0x41814c LDR	D31, [X6, #16]

      (94) 0x418150 B	418040

0x418154 ADD	W15, W15, #1

0x418158 MOVZ	W1, #0

0x41815c MADD	W2, W15, W13, W1

0x418160 ADD	W15, W15, W2

0x418164 CMP	W2, W15

0x418168 B.LT	417f70

0x41816c MOVI	D16, #0

0x418170 B	4180a8

0x418174 ADRP	X7,

0x418178 ADRP	X8,

0x41817c ADRP	X10,

0x418180 ADD	X3, X7, #3488

0x418184 ADD	X1, X8, #3544

0x418188 ADD	X0, X10, #3640

0x41818c MOVZ	W2, #182

0x418190 BL	4101c0

0x418194 HINT	#0

0x418198 HINT	#0

0x41819c HINT	#0

Coverage (%)	Name	Source Location	Module
►98.58+	omp_fulfill_event		libgomp.so.1.0.0
○	start_thread		libc.so.6
○	thread_start		libc.so.6
►1.40+	GOMP_parallel		libgomp.so.1.0.0
○	ljForce	ljForce.c:172	exec
○	timestep	timestep.c:51	exec
○	main	CoMD.c:125	exec
○	__libc_start_call_main		libc.so.6
○	__libc_start_main		libc.so.6
○	_start	CoMD.c:266	exec

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

Path /

Average path: Display a virtual path defined by average values of all real paths

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run gcc_4

Source file and lines	ljForce.c:172-216
Module	exec

The function is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172-191,197-216.

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 12.50 to 1.83 cycles (6.82x speedup).

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Vectorization

Your function is not vectorized. Only 26% of vector register length is used (average across all VPU instructions). By fully vectorizing your function, you can lower the cost of an iteration from 12.50 to 4.80 cycles (2.60x speedup).

Details

2% of VPU instructions are used in vector version (process two or more data elements in vector registers):

10% of VPU loads are used in vector version.
0% of VPU stores are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU multiply instructions are used in vector version.
0% of VPU fused multiply-add instructions are used in vector version.
0% of VPU divide and square root instructions are used in vector version.
0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully vectorized function can use their full power.

Workaround

Try another compiler or update/tune your current one
Make array accesses unit-stride:
- If your function streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 12.50 to 7.75 cycles (1.61x speedup).

CALL instructions

Detected function call instructions.

Details

Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.

unknown: 3 occurrences►
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:182

Matching between your function (in the source code) and the binary function

The binary function is composed of 1 FP arithmetical operations:

1: multiply

The binary function is loading 6 bytes.

Arithmetic intensity

Arithmetic intensity is 0.17 FP operations per loaded or stored byte.

General properties

nb instructions	66
nb uops	63
loop length	264
used w registers	9
used x registers	19
used b registers	0
used h registers	0
used s registers	1
used d registers	11
used q registers	0
used v registers	0
used z registers	0
nb stack references	12

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	7.88 cycles
front end	7.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	4.00	4.00	7.50	7.50	7.50	7.50	2.25	2.25	2.25	2.25	7.00	7.00	7.00	2.50	2.50
cycles	4.00	4.00	7.50	7.50	7.50	7.50	2.25	2.25	2.25	2.25	7.00	7.00	7.00	2.50	2.50

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P7 (128 bits): VPU, ALU, FP store data
P8 (128 bits): VPU, ALU, DIV/SQRT
P9 (128 bits): ALU, VPU
P10 (256 bits): store address, load
P11 (256 bits): store address, load
P12 (256 bits): load
P13 (64 bits): store data
P14 (64 bits): store data

Cycles executing div or sqrt instructions

5.00-12.50

Cycles summary

Front-end	7.88
Dispatch	7.50
DIV/SQRT	5.00-12.50
Overall L1	7.88-12.50

Vectorization ratios

INT

all	2%
load	10%
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	0%
other	0%

all	0%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	0%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

INT+FP

all	2%
load	10%
store	0%
mul	0%
add-sub	0%
fma	0%
div/sqrt	0%
other	0%

Vector efficiency ratios

INT

all	27%
load	33%
store	45%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	20%
fma	12%
other	21%

all	25%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	25%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	25%

INT+FP

all	26%
load	33%
store	45%
mul	25%
add-sub	20%
fma	12%
div/sqrt	12%
other	22%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 12.50 cycles. At this rate:

0% of peak load performance is reached (0.48 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the function is: 417f0c

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput	Vectorization
STP X29, X30, [SP, #928]!	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
ADD X29, SP, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
STP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
ORR X20, XZR, X0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
STP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDR X19, [X0]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
STR D8, [SP, #80]	1	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	0	0	2	0.50	scal (25.0%)
BL 410120 <@plt_start@+0x100>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR D8, [X20, #8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
ORR W21, WZR, W0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
BL 410140 <@plt_start@+0x120>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR X22, [X19, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
ORR W13, WZR, W0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
LDR W30, [X20, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (12.5%)
LDP D7, D21, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (50.0%)
LDR W0, [X22, #12]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR D20, [X20, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
SDIV W15, W0, W21	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	5-12	5-12.50	scal (12.5%)
MSUB W1, W15, W21, W0	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
CMP W13, W1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.LT 418154 <ljForce._omp_fn.1+0x248>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MADD W2, W15, W13, W1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
ADD W15, W15, W2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
CMP W2, W15	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.GE 41816c <ljForce._omp_fn.1+0x260>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
FMOV D0, #-4.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
LDR X21, [X22, #120]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (25.0%)
MOVI D16, #0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D18, #1.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D17, #0.5000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
SBFM X16, X2, #0, #31	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (100.0%)
FMOV D23, #12.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D22, #-6.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
UBFM W13, W2, #26, #25	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
STP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
MOVZ W23, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
FMUL D19, D8, D0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	3	0.25	scal (25.0%)
STP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
LDP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
ADD X20, X20, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR X30, [X20]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (25.0%)
LDP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	N/A
LDP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	N/A
LDR D8, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
LDP X29, X30, [SP], #96	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
RET	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ADD W15, W15, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
MOVZ W1, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
MADD W2, W15, W13, W1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
ADD W15, W15, W2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
CMP W2, W15	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.LT 417f70 <ljForce._omp_fn.1+0x64>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOVI D16, #0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
B 4180a8 <ljForce._omp_fn.1+0x19c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ADRP X7, <421174>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADRP X8, <420178>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADRP X10, <42017c>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD X3, X7, #3488	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADD X1, X8, #3544	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADD X0, X10, #3640	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOVZ W2, #182	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
BL 4101c0 <@plt_start@+0x1a0>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
HINT #0																			N/A
HINT #0																			N/A
HINT #0																			N/A

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run gcc_4

Source file and lines	ljForce.c:172-216
Module	exec

The function is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172-191,197-216.

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 12.50 to 1.83 cycles (6.82x speedup).

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Vectorization

Details

2% of VPU instructions are used in vector version (process two or more data elements in vector registers):

10% of VPU loads are used in vector version.
0% of VPU stores are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU multiply instructions are used in vector version.
0% of VPU fused multiply-add instructions are used in vector version.
0% of VPU divide and square root instructions are used in vector version.
0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully vectorized function can use their full power.

Workaround

Try another compiler or update/tune your current one
Make array accesses unit-stride:
- If your function streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 12.50 to 7.75 cycles (1.61x speedup).

CALL instructions

Detected function call instructions.

Details

unknown: 3 occurrences►
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:172
- /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:182

Matching between your function (in the source code) and the binary function

The binary function is composed of 1 FP arithmetical operations:

1: multiply

The binary function is loading 6 bytes.

Arithmetic intensity

Arithmetic intensity is 0.17 FP operations per loaded or stored byte.

General properties

nb instructions	66
nb uops	63
loop length	264
used w registers	9
used x registers	19
used b registers	0
used h registers	0
used s registers	1
used d registers	11
used q registers	0
used v registers	0
used z registers	0
nb stack references	12

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	7.88 cycles
front end	7.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	4.00	4.00	7.50	7.50	7.50	7.50	2.25	2.25	2.25	2.25	7.00	7.00	7.00	2.50	2.50
cycles	4.00	4.00	7.50	7.50	7.50	7.50	2.25	2.25	2.25	2.25	7.00	7.00	7.00	2.50	2.50

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P7 (128 bits): VPU, ALU, FP store data
P8 (128 bits): VPU, ALU, DIV/SQRT
P9 (128 bits): ALU, VPU
P10 (256 bits): store address, load
P11 (256 bits): store address, load
P12 (256 bits): load
P13 (64 bits): store data
P14 (64 bits): store data

Cycles executing div or sqrt instructions

5.00-12.50

Cycles summary

Front-end	7.88
Dispatch	7.50
DIV/SQRT	5.00-12.50
Overall L1	7.88-12.50

Vectorization ratios

INT

all	2%
load	10%
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	0%
other	0%

all	0%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	0%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

INT+FP

all	2%
load	10%
store	0%
mul	0%
add-sub	0%
fma	0%
div/sqrt	0%
other	0%

Vector efficiency ratios

INT

all	27%
load	33%
store	45%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	20%
fma	12%
other	21%

all	25%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	25%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	25%

INT+FP

all	26%
load	33%
store	45%
mul	25%
add-sub	20%
fma	12%
div/sqrt	12%
other	22%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 12.50 cycles. At this rate:

0% of peak load performance is reached (0.48 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the function is: 417f0c

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput	Vectorization
STP X29, X30, [SP, #928]!	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
ADD X29, SP, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
STP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
ORR X20, XZR, X0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
STP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDR X19, [X0]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
STR D8, [SP, #80]	1	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	0	0	2	0.50	scal (25.0%)
BL 410120 <@plt_start@+0x100>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR D8, [X20, #8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
ORR W21, WZR, W0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
BL 410140 <@plt_start@+0x120>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR X22, [X19, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
ORR W13, WZR, W0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
LDR W30, [X20, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (12.5%)
LDP D7, D21, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (50.0%)
LDR W0, [X22, #12]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR D20, [X20, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
SDIV W15, W0, W21	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	5-12	5-12.50	scal (12.5%)
MSUB W1, W15, W21, W0	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
CMP W13, W1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.LT 418154 <ljForce._omp_fn.1+0x248>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MADD W2, W15, W13, W1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
ADD W15, W15, W2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
CMP W2, W15	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.GE 41816c <ljForce._omp_fn.1+0x260>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
FMOV D0, #-4.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
LDR X21, [X22, #120]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (25.0%)
MOVI D16, #0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D18, #1.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D17, #0.5000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
SBFM X16, X2, #0, #31	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (100.0%)
FMOV D23, #12.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
FMOV D22, #-6.0000000	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
UBFM W13, W2, #26, #25	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
STP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
MOVZ W23, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
FMUL D19, D8, D0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	3	0.25	scal (25.0%)
STP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
LDP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
ADD X20, X20, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR X30, [X20]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (25.0%)
LDP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	N/A
LDP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	N/A
LDR D8, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	scal (25.0%)
LDP X29, X30, [SP], #96	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1	scal (50.0%)
RET	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ADD W15, W15, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
MOVZ W1, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
MADD W2, W15, W13, W1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (12.5%)
ADD W15, W15, W2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
CMP W2, W15	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
B.LT 417f70 <ljForce._omp_fn.1+0x64>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOVI D16, #0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (25.0%)
B 4180a8 <ljForce._omp_fn.1+0x19c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ADRP X7, <421174>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADRP X8, <420178>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADRP X10, <42017c>	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD X3, X7, #3488	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADD X1, X8, #3544	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (25.0%)
ADD X0, X10, #3640	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOVZ W2, #182	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
BL 4101c0 <@plt_start@+0x1a0>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
HINT #0																			N/A
HINT #0																			N/A
HINT #0																			N/A

Name	Coverage (%)	Time (s)
▼ljForce._omp_fn.1–	90.81	13.40
▼Loop 90 - ljForce.c:175-216 - exec–	0.03	0.01
▼Loop 92 - ljForce.c:178-216 - exec–	0.40	0.04
▼Loop 93 - ljForce.c:187-216 - exec–	3.00	0.33
○Loop 94 - ljForce.c:191-216 - exec	87.38	9.74
○Loop 91 - ljForce.c:172-172 - exec	0.00	0.00

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Expensive FP math instructions/calls

CALL instructions

Details

Matching between your function (in the source code) and the binary function

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code

Code clean check

Workaround

Vectorization

Details

Workaround

Expensive FP math instructions/calls

CALL instructions

Details

Matching between your function (in the source code) and the binary function

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code