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

	Function: qmcplusplus::SoaDistanceTableABOMPTarget<double, 3u, 40>::update(int)	Module: exec	Source: SoaDistanceTableABOMPTarget.h:371-377 [...]	Coverage: 0.01%

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h: 242 - 249

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

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

[...]

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

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h: 271 - 271

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

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

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h: 248 - 248

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

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

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algobase.h: 413 - 453

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

413: 	{ *__to = *__from; }

[...]

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

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

453: 	  else if (_Num == 1)

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h: 1131 - 1131

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

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

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algo.h: 683 - 709

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

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

[...]

709:       if (__n2 <= 0)

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h: 371 - 377

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

371:   inline void update(IndexType iat) override

372:   {

373:     ScopedTimer local_timer(update_timer_);

374:     std::copy_n(temp_r_.data(), num_sources_, distances_[iat].data());

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

376:       std::copy_n(temp_dr_.data(idim), num_sources_, displacements_[iat].data(idim));

377:   }

0x50f840 STP	X29, X30, [SP, #976]!

0x50f844 ADD	X29, SP, #0

0x50f848 STP	X19, X20, [SP, #16]

0x50f84c ORR	X20, XZR, X0

0x50f850 STP	X21, X22, [SP, #32]

0x50f854 ORR	W21, WZR, W1

0x50f858 LDR	X22, [X0, #456]

0x50f85c ORR	X0, XZR, X22

0x50f860 BL	4c43c0

0x50f864 MOVZ	W0, #40

0x50f868 LDR	X14, [X20, #16]

0x50f86c SMADDL	X21, W21, W0, XZR

0x50f870 CBZ	X14, 50f898

0x50f874 LDR	X4, [X20, #72]

0x50f878 UBFM	X2, X14, #61, #60

0x50f87c LDR	X1, [X20, #144]

0x50f880 ADD	X5, X4, X21

0x50f884 LDR	X0, [X5, #24]

0x50f888 CMP	X2, #8

0x50f88c B.LE	50f91c

0x50f890 BL	403a40

0x50f894 LDR	X14, [X20, #16]

0x50f898 MOVZ	X19, #0

(3884) 0x50f89c CBZ	X14, 50f8e0

(3884) 0x50f8a0 LDR	X1, [X20, #96]

(3884) 0x50f8a4 UBFM	X2, X14, #61, #60

(3884) 0x50f8a8 LDR	X3, [X20, #168]

(3884) 0x50f8ac LDR	X6, [X20, #184]

(3884) 0x50f8b0 ADD	X7, X1, X21

(3884) 0x50f8b4 LDR	X10, [X7, #8]

(3884) 0x50f8b8 MADD	X8, X19, X3, XZR

(3884) 0x50f8bc LDR	X11, [X7, #24]

(3884) 0x50f8c0 UBFM	X9, X8, #61, #60

(3884) 0x50f8c4 MADD	X12, X19, X10, XZR

(3884) 0x50f8c8 ADD	X1, X6, X9

(3884) 0x50f8cc UBFM	X13, X12, #61, #60

(3884) 0x50f8d0 ADD	X0, X11, X13

(3884) 0x50f8d4 CMP	X2, #8

(3884) 0x50f8d8 B.LE	50f90c

(3884) 0x50f8dc BL	403a40

  (3883) 0x50f8e0 ADD	X19, X19, #1

  (3883) 0x50f8e4 CMP	X19, #3

  (3883) 0x50f8e8 B.EQ	50f8f4

(3884) 0x50f8ec LDR	X14, [X20, #16]

(3884) 0x50f8f0 B	50f89c

  (3883) 0x50f8f4 ORR	X0, XZR, X22

  (3883) 0x50f8f8 BL	4c4660

  (3883) 0x50f8fc LDP	X19, X20, [SP, #16]

  (3883) 0x50f900 LDP	X21, X22, [SP, #32]

  (3883) 0x50f904 LDP	X29, X30, [SP], #48

  (3883) 0x50f908 RET

  (3883) 0x50f90c B.NE	50f8e0

  (3883) 0x50f910 LDR	D0, [X6, X9]

  (3883) 0x50f914 STR	D0, [X11, X13]

  (3883) 0x50f918 B	50f8e0

0x50f91c B.NE	50f898

0x50f920 LDR	D31, [X1]

0x50f924 MOVZ	X19, #0

0x50f928 STR	D31, [X0]

0x50f92c B	50f89c

0x50f930 HINT	#0

0x50f934 HINT	#0

0x50f938 HINT	#0

0x50f93c HINT	#0

Coverage (%)	Name	Source Location	Module

Path /

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

Source file and lines	SoaDistanceTableABOMPTarget.h:371-377
Module	exec

The function is defined in:

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:242,249
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:271
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:248
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algobase.h:413,451-453
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h:1131
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algo.h:683,709
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h:371-377

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 3.67 to 0.83 cycles (4.40x 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 21% of vector register length is used (average across all VPU instructions). By vectorizing your function, you can lower the cost of an iteration from 3.67 to 0.56 cycles (6.52x 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 3.67 to 1.00 cycles (3.67x 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 56 bytes. The binary function is storing 56 bytes.

General properties

nb instructions	32
loop length	128
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

3.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	3.00	3.00	3.50	3.50	3.50	3.50	0.50	0.50	0.00	0.00	3.67	3.67	3.67	1.50	1.50
cycles	3.00	3.00	3.50	3.50	3.50	3.50	0.50	0.50	0.00	0.00	3.67	3.67	3.67	1.50	1.50

Execution ports to units layout:

P0:
P1:
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

Cycles summary

Front-end	3.50
Overall L1	3.67

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 3.67 cycles. At this rate:

15% of peak load performance is reached (15.27 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
31% of peak store performance is reached (15.27 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 50f840

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput
STP X29, X30, [SP, #976]!	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
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
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
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
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
ORR W21, WZR, W1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X22, [X0, #456]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ORR X0, XZR, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 4c43c0 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
MOVZ W0, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X14, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SMADDL X21, W21, W0, XZR	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
CBZ X14, 50f898 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x58>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X4, [X20, #72]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
UBFM X2, X14, #61, #60	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X1, [X20, #144]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD X5, X4, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X0, [X5, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
CMP X2, #8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.LE 50f91c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0xdc>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
BL 403a40 <@plt_start@+0x50>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X14, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
MOVZ X19, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
B.NE 50f898 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x58>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR D31, [X1]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33
MOVZ X19, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR D31, [X0]	1	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	0	0	2	0.50
B 50f89c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x5c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
HINT #0
HINT #0
HINT #0
HINT #0

Source file and lines	SoaDistanceTableABOMPTarget.h:371-377
Module	exec

The function is defined in:

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:242,249
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:271
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:248
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algobase.h:413,451-453
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h:1131
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_algo.h:683,709
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h:371-377

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 3.67 to 0.83 cycles (4.40x speedup).

Workaround

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

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 3.67 to 1.00 cycles (3.67x 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 56 bytes. The binary function is storing 56 bytes.

General properties

nb instructions	32
loop length	128
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

3.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	3.00	3.00	3.50	3.50	3.50	3.50	0.50	0.50	0.00	0.00	3.67	3.67	3.67	1.50	1.50
cycles	3.00	3.00	3.50	3.50	3.50	3.50	0.50	0.50	0.00	0.00	3.67	3.67	3.67	1.50	1.50

Execution ports to units layout:

P0:
P1:
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

Cycles summary

Front-end	3.50
Overall L1	3.67

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 3.67 cycles. At this rate:

15% of peak load performance is reached (15.27 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
31% of peak store performance is reached (15.27 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 50f840

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput
STP X29, X30, [SP, #976]!	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
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
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
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
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
ORR W21, WZR, W1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X22, [X0, #456]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ORR X0, XZR, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 4c43c0 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
MOVZ W0, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X14, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SMADDL X21, W21, W0, XZR	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
CBZ X14, 50f898 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x58>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X4, [X20, #72]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
UBFM X2, X14, #61, #60	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X1, [X20, #144]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD X5, X4, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X0, [X5, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
CMP X2, #8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.LE 50f91c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0xdc>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
BL 403a40 <@plt_start@+0x50>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X14, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
MOVZ X19, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
B.NE 50f898 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x58>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR D31, [X1]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33
MOVZ X19, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR D31, [X0]	1	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	0	0	2	0.50
B 50f89c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE6updateEi+0x5c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
HINT #0
HINT #0
HINT #0
HINT #0

Name	Coverage (%)	Time (s)
▼qmcplusplus::SoaDistanceTableABOMPTarget::update(int)–	0.01	0.02
▼Loop 3884 - SoaDistanceTableABOMPTarget.h:375-377 - exec–	0	0.01
○Loop 3883 - stl_algobase.h:413-453 - exec	0	0

Report Configuration

Code clean check

Workaround

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

Code clean check

Workaround

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