OV - exec - Function qmcplusplus::SoaDistanceTableAAOMPTarget<double, 3u, 40>::update(int)

	Function: qmcplusplus::SoaDistanceTableAAOMPTarget<double, 3u, 40>::update(int)	Module: exec	Source: SoaDistanceTableAAOMPTarget.h:429-445 [...]	Coverage (incl. loops): 9.11% \| (excl. loops): 0.00%

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h: 242 - 249

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

242:   ScopeGuard(TIMER& t) : timer(t) { timer.start(); }

[...]

249:   ~ScopeGuard() { timer.stop(); }

/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_vector.h: 1131 - 1131

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

1131: 	return *(this->_M_impl._M_start + __n);

/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algo.h: 683 - 709

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

683:     { return std::copy(__first, __first + __n, __result); }

[...]

709:       if (__n2 <= 0)

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h: 223 - 248

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

223:     return X[i];

[...]

248:   inline pointer data() { return X; }

/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algobase.h: 413 - 452

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

413: 	{ *__to = *__from; }

[...]

451: 	  if (__builtin_expect(_Num > 1, true))

452: 	    __builtin_memmove(__result, __first, sizeof(_Tp) * _Num);

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h: 231 - 271

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

231:   inline const AoSElement_t operator[](size_t i) const { return AoSElement_t(myData + i, nGhosts); }

[...]

244:         *(_base + M * i) = rhs[i];

[...]

263:   inline Accessor operator()(size_t i) { return Accessor(myData + i, nGhosts); }

[...]

271:   inline T* restrict data(size_t i) { return myData + i * nGhosts; }

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h: 429 - 445

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

429:   inline void update(IndexType iat) override

430:   {

431:     ScopedTimer local_timer(update_timer_);

[...]

437:     for (int idim = 0; idim < D; ++idim)

438:       std::copy_n(temp_dr_.data(idim), nupdate, displacements_[iat].data(idim));

439:     //copy column

440:     for (size_t i = iat + 1; i < num_targets_; ++i)

441:     {

442:       distances_[i][iat]     = temp_r_[i];

443:       displacements_[i](iat) = -temp_dr_[i];

444:     }

445:   }

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h: 182 - 182

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

182:       inverse[d] = -X[d];

0x5000a0 STP	X29, X30, [SP, #944]!

0x5000a4 ADD	X29, SP, #0

0x5000a8 STP	X19, X20, [SP, #16]

0x5000ac ORR	W19, WZR, W1

0x5000b0 SBFM	X20, X1, #61, #31

0x5000b4 STP	X21, X22, [SP, #32]

0x5000b8 STP	X23, X24, [SP, #48]

0x5000bc ORR	X23, XZR, X0

0x5000c0 STP	X25, X26, [SP, #64]

0x5000c4 MOVZ	W25, #40

0x5000c8 SMADDL	X25, W1, W25, XZR

0x5000cc LDR	X21, [X0, #648]

0x5000d0 ORR	X0, XZR, X21

0x5000d4 BL	4d5540

0x5000d8 CMP	W19, #0

0x5000dc B.LE	500148

0x5000e0 LDR	X0, [X23, #72]

0x5000e4 LDR	X1, [X23, #144]

0x5000e8 ADD	X2, X0, X25

0x5000ec LDR	X0, [X2, #24]

0x5000f0 CMP	X20, #8

0x5000f4 B.EQ	500298

0x5000f8 ORR	X2, XZR, X20

0x5000fc ADD	X24, X23, #160

0x500100 ADD	X26, X23, #96

0x500104 MOVZ	X22, #0

0x500108 BL	403d70

(3293) 0x50010c LDR	X15, [X26]

(3293) 0x500110 ORR	X2, XZR, X20

(3293) 0x500114 LDR	X12, [X24, #8]

(3293) 0x500118 LDR	X13, [X24, #24]

(3293) 0x50011c ADD	X16, X15, X25

(3293) 0x500120 LDR	X17, [X16, #8]

(3293) 0x500124 MADD	X14, X22, X12, XZR

(3293) 0x500128 LDR	X18, [X16, #24]

(3293) 0x50012c ADD	X1, X13, X14,LSL #3

(3293) 0x500130 MADD	X30, X22, X17, XZR

(3293) 0x500134 ADD	X22, X22, #1

(3293) 0x500138 ADD	X0, X18, X30,LSL #3

(3293) 0x50013c BL	403d70

(3293) 0x500140 CMP	X22, #3

(3293) 0x500144 B.NE	50010c

(3291) 0x500148 LDR	X24, [X23, #24]

(3291) 0x50014c ADD	W19, W19, #1

(3291) 0x500150 SBFM	X1, X19, #0, #31

(3291) 0x500154 CMP	X1, X24

(3291) 0x500158 B.CS	500278

(3291) 0x50015c LDR	X26, [X23, #72]

(3291) 0x500160 SUB	X0, X24, X1

(3291) 0x500164 ADD	X9, X25, #64

(3291) 0x500168 LDR	X11, [X23, #96]

(3291) 0x50016c LDR	X14, [X23, #144]

(3291) 0x500170 LDRSW	X2, [X23, #168]

(3291) 0x500174 SUB	X10, X11, #16

(3291) 0x500178 LDR	X23, [X23, #184]

(3291) 0x50017c ADD	X12, X23, X2,LSL #3

(3291) 0x500180 ADD	X6, X23, X2,LSL #4

(3291) 0x500184 TBZ	W0, #0, 5001e0

(3291) 0x500188 LDR	D3, [X14, X1,LSL #3]

(3291) 0x50018c LDR	X5, [X26, X9]

(3291) 0x500190 LDR	X4, [X10, X9]

(3291) 0x500194 LDR	X7, [X11, X9]

(3291) 0x500198 ADD	X9, X25, #104

(3291) 0x50019c STR	D3, [X5, X20]

(3291) 0x5001a0 LDR	D29, [X23, X1,LSL #3]

(3291) 0x5001a4 ADD	X25, X20, X4,LSL #3

(3291) 0x5001a8 LDR	D4, [X12, X1,LSL #3]

(3291) 0x5001ac ADD	X3, X7, X4,LSL #3

(3291) 0x5001b0 LDR	D30, [X6, X1,LSL #3]

(3291) 0x5001b4 ADD	X1, X1, #1

(3291) 0x5001b8 FNEG	D5, D29

(3291) 0x5001bc FNEG	D6, D4

(3291) 0x5001c0 FNEG	D7, D30

(3291) 0x5001c4 STR	D5, [X7, X20]

(3291) 0x5001c8 STR	D6, [X7, X25]

(3291) 0x5001cc STR	D7, [X3, X25]

(3291) 0x5001d0 CMP	X1, X24

(3291) 0x5001d4 B.EQ	500278

(3291) 0x5001d8 HINT	#0

(3291) 0x5001dc HINT	#0

  (3292) 0x5001e0 LDR	D16, [X14, X1,LSL #3]

  (3292) 0x5001e4 ADD	X15, X9, #40

  (3292) 0x5001e8 ADD	X13, X1, #1

  (3292) 0x5001ec LDR	X8, [X26, X9]

  (3292) 0x5001f0 LDR	X16, [X10, X9]

  (3292) 0x5001f4 LDR	X17, [X11, X9]

  (3292) 0x5001f8 ADD	X9, X9, #80

  (3292) 0x5001fc STR	D16, [X8, X20]

  (3292) 0x500200 LDR	D23, [X23, X1,LSL #3]

  (3292) 0x500204 ADD	X18, X20, X16,LSL #3

  (3292) 0x500208 LDR	D25, [X12, X1,LSL #3]

  (3292) 0x50020c ADD	X30, X17, X16,LSL #3

  (3292) 0x500210 LDR	D24, [X6, X1,LSL #3]

  (3292) 0x500214 ADD	X1, X1, #2

  (3292) 0x500218 FNEG	D17, D23

  (3292) 0x50021c LDR	X19, [X10, X15]

  (3292) 0x500220 FNEG	D18, D25

  (3292) 0x500224 LDR	X22, [X26, X15]

  (3292) 0x500228 FNEG	D19, D24

  (3292) 0x50022c STR	D17, [X17, X20]

  (3292) 0x500230 LDR	X0, [X11, X15]

  (3292) 0x500234 ADD	X2, X20, X19,LSL #3

  (3292) 0x500238 STR	D18, [X17, X18]

  (3292) 0x50023c STR	D19, [X30, X18]

  (3292) 0x500240 LDR	D20, [X14, X13,LSL #3]

  (3292) 0x500244 ADD	X5, X0, X19,LSL #3

  (3292) 0x500248 STR	D20, [X22, X20]

  (3292) 0x50024c LDR	D26, [X23, X13,LSL #3]

  (3292) 0x500250 LDR	D28, [X12, X13,LSL #3]

  (3292) 0x500254 LDR	D27, [X6, X13,LSL #3]

  (3292) 0x500258 FNEG	D21, D26

  (3292) 0x50025c FNEG	D22, D28

  (3292) 0x500260 FNEG	D31, D27

  (3292) 0x500264 STR	D21, [X0, X20]

  (3292) 0x500268 STR	D22, [X0, X2]

  (3292) 0x50026c STR	D31, [X5, X2]

  (3292) 0x500270 CMP	X1, X24

  (3292) 0x500274 B.NE	5001e0

(3291) 0x500278 ORR	X0, XZR, X21

(3291) 0x50027c BL	4d57e0

(3291) 0x500280 LDP	X19, X20, [SP, #16]

(3291) 0x500284 LDP	X21, X22, [SP, #32]

(3291) 0x500288 LDP	X23, X24, [SP, #48]

(3291) 0x50028c LDP	X25, X26, [SP, #64]

(3291) 0x500290 LDP	X29, X30, [SP], #80

(3291) 0x500294 RET

(3291) 0x500298 LDR	D31, [X1]

(3291) 0x50029c LDR	X5, [X23, #96]

(3291) 0x5002a0 LDR	X3, [X23, #184]

(3291) 0x5002a4 LDR	X1, [X23, #168]

(3291) 0x5002a8 ADD	X7, X5, X25

(3291) 0x5002ac STR	D31, [X0]

(3291) 0x5002b0 LDR	X8, [X7, #24]

(3291) 0x5002b4 LDR	D0, [X3]

(3291) 0x5002b8 UBFM	X4, X1, #61, #60

(3291) 0x5002bc ADD	X6, X3, X1,LSL #3

(3291) 0x5002c0 LDR	X9, [X7, #8]

(3291) 0x5002c4 STR	D0, [X8]

(3291) 0x5002c8 LDR	D1, [X3, X4]

(3291) 0x5002cc UBFM	X10, X9, #61, #60

(3291) 0x5002d0 ADD	X11, X8, X9,LSL #3

(3291) 0x5002d4 STR	D1, [X8, X10]

(3291) 0x5002d8 LDR	D2, [X6, X4]

(3291) 0x5002dc STR	D2, [X11, X10]

(3291) 0x5002e0 B	500148

0x5002e4 HINT	#0

0x5002e8 HINT	#0

0x5002ec HINT	#0

0x5002f0 HINT	#0

0x5002f4 HINT	#0

0x5002f8 HINT	#0

0x5002fc HINT	#0

Coverage (%)	Name	Source Location	Module
►98.94+	qmcplusplus::ParticleSet::acce[...]	ParticleSet.cpp:389	exec
○	main._omp_fn.1	miniqmc.cpp:453	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	thread_start		libc.so.6

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_0

Source file and lines	SoaDistanceTableAAOMPTarget.h:429-445
Module	exec

The function is defined in:

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:242,249
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_vector.h:1131
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algo.h:683,709
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:223,248
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algobase.h:413,451-452
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:231,244,263,271
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h:429-431,437-445
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h:182

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Vectorization

Your function is not vectorized. Only 62% of vector register length is used (average across all VPU instructions). By vectorizing your function, you can lower the cost of an iteration from 3.38 to 2.90 cycles (1.16x speedup).

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a 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 3.38 to 1.00 cycles (3.37x speedup).

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

The binary function does not contain any FP arithmetical operations. The binary function is loading 32 bytes. The binary function is storing 80 bytes.

General properties

nb instructions	34
loop length	136
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

3.38 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	2.00	2.00	2.92	2.92	2.83	2.83	2.75	2.75	0.00	0.00	0.00	0.00	3.17	2.83	3.00	2.50	2.50
cycles	2.00	2.00	2.92	2.92	2.83	2.83	2.75	2.75	0.00	0.00	0.00	0.00	3.17	2.83	3.00	2.50	2.50

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6: ALU
P7: ALU
P8 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P9 (128 bits): VPU, ALU, FP store data
P10 (128 bits): VPU, ALU, DIV/SQRT
P11 (128 bits): ALU, VPU
P12 (256 bits): store address, load
P13 (256 bits): store address, load
P14 (256 bits): load
P15 (64 bits): store data
P16 (64 bits): store data

Cycles executing div or sqrt instructions

Cycles summary

Front-end	3.38
Overall L1	3.38

Vectorization ratios

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 3.38 cycles. At this rate:

9% of peak load performance is reached (9.48 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
49% of peak store performance is reached (23.70 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 5000a0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
STP X29, X30, [SP, #944]!	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ADD X29, SP, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
STP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ORR W19, WZR, W1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (25.0%)
SBFM X20, X1, #61, #31	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (100.0%)
STP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
STP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ORR X23, XZR, X0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
MOVZ W25, #40	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (25.0%)
SMADDL X25, W1, W25, XZR	1	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (25.0%)
LDR X21, [X0, #648]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
ORR X0, XZR, X21	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
BL 4d5540 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
CMP W19, #0	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (25.0%)
B.LE 500148 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0xa8>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR X0, [X23, #72]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR X1, [X23, #144]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
ADD X2, X0, X25	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X0, [X2, #24]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
CMP X20, #8	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (50.0%)
B.EQ 500298 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0x1f8>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ORR X2, XZR, X20	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADD X24, X23, #160	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X26, X23, #96	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVZ X22, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
BL 403d70 <@plt_start@+0x80>	1	0.50	0.50	0.50	0.50	0	0	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
HINT #0																					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_0

Source file and lines	SoaDistanceTableAAOMPTarget.h:429-445
Module	exec

The function is defined in:

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:242,249
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_vector.h:1131
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algo.h:683,709
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:223,248
/opt/arm/gcc-14.2.0_AmazonLinux-2023/include/c++/14.2.0/bits/stl_algobase.h:413,451-452
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:231,244,263,271
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h:429-431,437-445
/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h:182

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Vectorization

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a 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 3.38 to 1.00 cycles (3.37x speedup).

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

The binary function does not contain any FP arithmetical operations. The binary function is loading 32 bytes. The binary function is storing 80 bytes.

General properties

nb instructions	34
loop length	136
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

3.38 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	2.00	2.00	2.92	2.92	2.83	2.83	2.75	2.75	0.00	0.00	0.00	0.00	3.17	2.83	3.00	2.50	2.50
cycles	2.00	2.00	2.92	2.92	2.83	2.83	2.75	2.75	0.00	0.00	0.00	0.00	3.17	2.83	3.00	2.50	2.50

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6: ALU
P7: ALU
P8 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P9 (128 bits): VPU, ALU, FP store data
P10 (128 bits): VPU, ALU, DIV/SQRT
P11 (128 bits): ALU, VPU
P12 (256 bits): store address, load
P13 (256 bits): store address, load
P14 (256 bits): load
P15 (64 bits): store data
P16 (64 bits): store data

Cycles executing div or sqrt instructions

Cycles summary

Front-end	3.38
Overall L1	3.38

Vectorization ratios

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 3.38 cycles. At this rate:

9% of peak load performance is reached (9.48 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
49% of peak store performance is reached (23.70 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 5000a0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
STP X29, X30, [SP, #944]!	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ADD X29, SP, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
STP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ORR W19, WZR, W1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (25.0%)
SBFM X20, X1, #61, #31	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (100.0%)
STP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
STP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
ORR X23, XZR, X0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STP X25, X26, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (100.0%)
MOVZ W25, #40	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (25.0%)
SMADDL X25, W1, W25, XZR	1	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1	scal (25.0%)
LDR X21, [X0, #648]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
ORR X0, XZR, X21	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
BL 4d5540 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
CMP W19, #0	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (25.0%)
B.LE 500148 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0xa8>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
LDR X0, [X23, #72]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR X1, [X23, #144]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
ADD X2, X0, X25	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X0, [X2, #24]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
CMP X20, #8	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	scal (50.0%)
B.EQ 500298 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0x1f8>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
ORR X2, XZR, X20	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADD X24, X23, #160	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X26, X23, #96	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVZ X22, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
BL 403d70 <@plt_start@+0x80>	1	0.50	0.50	0.50	0.50	0	0	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
HINT #0																					N/A
HINT #0																					N/A
HINT #0																					N/A
HINT #0																					N/A

Name	Coverage (%)	Time (s)
▼qmcplusplus::SoaDistanceTableAAOMPTarget::update(int)–	9.11	12.61
▼Loop 3291 - stl_algobase.h:413-413 - exec–	0.00	0.01
○Loop 3292 - SoaDistanceTableAAOMPTarget.h:440-442 - exec	9.10	12.47
○Loop 3293 - SoaDistanceTableAAOMPTarget.h:437-437 - exec	0.00	0.01

Report Configuration

Vectorization

Details

Workaround

Expensive FP math instructions/calls

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code

Vectorization

Details

Workaround

Expensive FP math instructions/calls

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code