OV - exec - Function kineticEnergy._omp

	Function: kineticEnergy._omp_fn.0	Module: exec	Source: timestep.c:107-116	Coverage (incl. loops): 0.13% \| (excl. loops): 0.00%

/home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c: 107 - 116

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

107:    #pragma omp parallel for reduction(+:kenergy)

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

109:    {

110:       for (int iOff=MAXATOMS*iBox,ii=0; ii<s->boxes->nAtoms[iBox]; ii++,iOff++)

111:       {

112:          int iSpecies = s->atoms->iSpecies[iOff];

113:          real_t invMass = 0.5/s->species[iSpecies].mass;

114:          kenergy += ( s->atoms->p[iOff][0] * s->atoms->p[iOff][0] +

115:          s->atoms->p[iOff][1] * s->atoms->p[iOff][1] +

116:          s->atoms->p[iOff][2] * s->atoms->p[iOff][2] )*invMass;

0x40edb0 PUSH	%RBP

0x40edb1 MOV	%RSP,%RBP

0x40edb4 PUSH	%R14

0x40edb6 PUSH	%R13

0x40edb8 PUSH	%R12

0x40edba PUSH	%RBX

0x40edbb MOV	(%RDI),%RBX

0x40edbe MOV	%RDI,%R12

0x40edc1 CALL	403060 <omp_get_num_threads@plt>

0x40edc6 MOV	%EAX,%R13D

0x40edc9 CALL	403160 <omp_get_thread_num@plt>

0x40edce MOV	%EAX,%ECX

0x40edd0 MOV	0x18(%RBX),%R14

0x40edd4 MOV	0xc(%R14),%EAX

0x40edd8 CLTD

0x40edd9 IDIV	%R13D

0x40eddc CMP	%EDX,%ECX

0x40edde JL	40f043

0x40ede4 IMUL	%EAX,%ECX

0x40ede7 VXORPD	%XMM0,%XMM0,%XMM0

0x40edeb ADD	%ECX,%EDX

0x40eded LEA	(%RAX,%RDX,1),%ESI

0x40edf0 CMP	%ESI,%EDX

0x40edf2 JGE	40f01b

0x40edf8 MOV	0x78(%R14),%RDI

0x40edfc MOVSXD	%EDX,%R10

0x40edff MOV	%EAX,%R11D

0x40ee02 VMOVSD	0x1cbe(%RIP),%XMM3

0x40ee0a ADD	%R10,%R11

0x40ee0d LEA	(%R10,%R10,2),%R13

0x40ee11 MOV	%R10,%R8

0x40ee14 SAL	$0x9,%R13

0x40ee18 SAL	$0x6,%R8

0x40ee1c LEA	(%RDI,%R10,4),%R9

0x40ee20 LEA	(%R11,%R11,2),%R10

0x40ee24 SAL	$0x9,%R10

0x40ee28 NOPL	(%RAX,%RAX,1)

(102) 0x40ee30 MOVSXD	(%R9),%RCX

(102) 0x40ee33 TEST	%ECX,%ECX

(102) 0x40ee35 JLE	40f003

(102) 0x40ee3b MOV	0x20(%RBX),%R14

(102) 0x40ee3f ADD	%R8,%RCX

(102) 0x40ee42 MOVSXD	%R8D,%RAX

(102) 0x40ee45 MOV	0x28(%RBX),%RDI

(102) 0x40ee49 MOV	0x10(%R14),%RSI

(102) 0x40ee4d LEA	(%RSI,%RCX,4),%R11

(102) 0x40ee51 LEA	(%RSI,%RAX,4),%RDX

(102) 0x40ee55 MOV	0x20(%R14),%RAX

(102) 0x40ee59 MOV	%R11,%RCX

(102) 0x40ee5c SUB	%RDX,%RCX

(102) 0x40ee5f SUB	$0x4,%RCX

(102) 0x40ee63 SHR	$0x2,%RCX

(102) 0x40ee67 INC	%RCX

(102) 0x40ee6a ADD	%R13,%RAX

(102) 0x40ee6d AND	$0x3,%ECX

(102) 0x40ee70 JE	40ef30

(102) 0x40ee76 CMP	$0x1,%RCX

(102) 0x40ee7a JE	40eef0

(102) 0x40ee7c CMP	$0x2,%RCX

(102) 0x40ee80 JE	40eeb9

(102) 0x40ee82 VMOVSD	(%RAX),%XMM1

(102) 0x40ee86 VMOVSD	0x8(%RAX),%XMM2

(102) 0x40ee8b VMULSD	%XMM2,%XMM2,%XMM4

(102) 0x40ee8f VFMADD231SD	%XMM1,%XMM1,%XMM4

(102) 0x40ee94 MOVSXD	(%RDX),%R14

(102) 0x40ee97 VMOVSD	0x10(%RAX),%XMM5

(102) 0x40ee9c ADD	$0x4,%RDX

(102) 0x40eea0 ADD	$0x18,%RAX

(102) 0x40eea4 VFMADD132SD	%XMM5,%XMM4,%XMM5

(102) 0x40eea9 SAL	$0x4,%R14

(102) 0x40eead VDIVSD	0x8(%RDI,%R14,1),%XMM3,%XMM6

(102) 0x40eeb4 VFMADD231SD	%XMM5,%XMM6,%XMM0

(102) 0x40eeb9 VMOVSD	(%RAX),%XMM7

(102) 0x40eebd VMOVSD	0x8(%RAX),%XMM8

(102) 0x40eec2 VMULSD	%XMM8,%XMM8,%XMM9

(102) 0x40eec7 VFMADD231SD	%XMM7,%XMM7,%XMM9

(102) 0x40eecc MOVSXD	(%RDX),%RSI

(102) 0x40eecf VMOVSD	0x10(%RAX),%XMM10

(102) 0x40eed4 ADD	$0x4,%RDX

(102) 0x40eed8 ADD	$0x18,%RAX

(102) 0x40eedc VFMADD132SD	%XMM10,%XMM9,%XMM10

(102) 0x40eee1 SAL	$0x4,%RSI

(102) 0x40eee5 VDIVSD	0x8(%RDI,%RSI,1),%XMM3,%XMM11

(102) 0x40eeeb VFMADD231SD	%XMM10,%XMM11,%XMM0

(102) 0x40eef0 VMOVSD	(%RAX),%XMM12

(102) 0x40eef4 VMOVSD	0x8(%RAX),%XMM13

(102) 0x40eef9 VMULSD	%XMM13,%XMM13,%XMM14

(102) 0x40eefe VFMADD231SD	%XMM12,%XMM12,%XMM14

(102) 0x40ef03 MOVSXD	(%RDX),%RCX

(102) 0x40ef06 VMOVSD	0x10(%RAX),%XMM15

(102) 0x40ef0b ADD	$0x4,%RDX

(102) 0x40ef0f ADD	$0x18,%RAX

(102) 0x40ef13 VFMADD132SD	%XMM15,%XMM14,%XMM15

(102) 0x40ef18 SAL	$0x4,%RCX

(102) 0x40ef1c VDIVSD	0x8(%RDI,%RCX,1),%XMM3,%XMM1

(102) 0x40ef22 VFMADD231SD	%XMM15,%XMM1,%XMM0

(102) 0x40ef27 CMP	%RDX,%R11

(102) 0x40ef2a JE	40f003

  (103) 0x40ef30 VMOVSD	(%RAX),%XMM4

  (103) 0x40ef34 VMOVSD	0x8(%RAX),%XMM2

  (103) 0x40ef39 VMULSD	%XMM2,%XMM2,%XMM5

  (103) 0x40ef3d VFMADD231SD	%XMM4,%XMM4,%XMM5

  (103) 0x40ef42 MOVSXD	(%RDX),%R14

  (103) 0x40ef45 VMOVSD	0x10(%RAX),%XMM6

  (103) 0x40ef4a VMOVSD	0x20(%RAX),%XMM8

  (103) 0x40ef4f VMULSD	%XMM8,%XMM8,%XMM9

  (103) 0x40ef54 MOVSXD	0x4(%RDX),%RSI

  (103) 0x40ef58 VMOVSD	0x28(%RAX),%XMM10

  (103) 0x40ef5d VMOVSD	0x30(%RAX),%XMM12

  (103) 0x40ef62 VMOVSD	0x38(%RAX),%XMM13

  (103) 0x40ef67 VMULSD	%XMM13,%XMM13,%XMM14

  (103) 0x40ef6c VFMADD231SD	%XMM12,%XMM12,%XMM14

  (103) 0x40ef71 MOVSXD	0x8(%RDX),%RCX

  (103) 0x40ef75 VMOVSD	0x40(%RAX),%XMM15

  (103) 0x40ef7a VFMADD132SD	%XMM6,%XMM5,%XMM6

  (103) 0x40ef7f VMOVSD	0x48(%RAX),%XMM2

  (103) 0x40ef84 VMOVSD	0x50(%RAX),%XMM1

  (103) 0x40ef89 VMULSD	%XMM1,%XMM1,%XMM5

  (103) 0x40ef8d VFMADD231SD	%XMM2,%XMM2,%XMM5

  (103) 0x40ef92 ADD	$0x10,%RDX

  (103) 0x40ef96 ADD	$0x60,%RAX

  (103) 0x40ef9a SAL	$0x4,%R14

  (103) 0x40ef9e VFMADD132SD	%XMM15,%XMM14,%XMM15

  (103) 0x40efa3 VDIVSD	0x8(%RDI,%R14,1),%XMM3,%XMM7

  (103) 0x40efaa VFMADD132SD	%XMM6,%XMM0,%XMM7

  (103) 0x40efaf VMOVSD	-0x48(%RAX),%XMM0

  (103) 0x40efb4 VFMADD231SD	%XMM0,%XMM0,%XMM9

  (103) 0x40efb9 SAL	$0x4,%RSI

  (103) 0x40efbd VMOVSD	-0x8(%RAX),%XMM0

  (103) 0x40efc2 VFMADD132SD	%XMM0,%XMM5,%XMM0

  (103) 0x40efc7 VDIVSD	0x8(%RDI,%RSI,1),%XMM3,%XMM11

  (103) 0x40efcd MOVSXD	-0x4(%RDX),%R14

  (103) 0x40efd1 VFMADD132SD	%XMM10,%XMM9,%XMM10

  (103) 0x40efd6 SAL	$0x4,%RCX

  (103) 0x40efda VDIVSD	0x8(%RDI,%RCX,1),%XMM3,%XMM4

  (103) 0x40efe0 VFMADD132SD	%XMM10,%XMM7,%XMM11

  (103) 0x40efe5 SAL	$0x4,%R14

  (103) 0x40efe9 VDIVSD	0x8(%RDI,%R14,1),%XMM3,%XMM6

  (103) 0x40eff0 VFMADD132SD	%XMM15,%XMM11,%XMM4

  (103) 0x40eff5 VFMADD132SD	%XMM6,%XMM4,%XMM0

  (103) 0x40effa CMP	%RDX,%R11

  (103) 0x40effd JNE	40ef30

(102) 0x40f003 ADD	$0x600,%R13

(102) 0x40f00a ADD	$0x4,%R9

(102) 0x40f00e ADD	$0x40,%R8

(102) 0x40f012 CMP	%R13,%R10

(102) 0x40f015 JNE	40ee30

0x40f01b MOV	0x8(%R12),%RAX

0x40f020 LEA	0x8(%R12),%RBX

(101) 0x40f025 VMOVQ	%RAX,%XMM3

(101) 0x40f02a VADDSD	%XMM3,%XMM0,%XMM7

(101) 0x40f02e VMOVQ	%XMM7,%R12

(101) 0x40f033 LOCK CMPXCHG	%R12,(%RBX)

(101) 0x40f038 JNE	40f025

0x40f03a POP	%RBX

0x40f03b POP	%R12

0x40f03d POP	%R13

0x40f03f POP	%R14

0x40f041 POP	%RBP

0x40f042 RET

0x40f043 INC	%EAX

0x40f045 XOR	%EDX,%EDX

0x40f047 JMP	40ede4

0x40f04c NOPL	(%RAX)

Coverage (%)	Name	Source Location	Module
►97.29+	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.86+	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0
○	kineticEnergy	timestep.c:123	exec
○	timestep	timestep.c:60	exec
○	main	CoMD.c:125	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_10

Source file and lines	timestep.c:107-116
Module	exec

The function is defined in /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107-116.

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 6.13 to 0.88 cycles (7.00x speedup).

Workaround

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

Vectorization

Your function is probably not vectorized. Only 10% of vector register length is used (average across all SSE/AVX instructions). By vectorizing your function, you can lower the cost of an iteration from 6.13 to 0.41 cycles (15.08x speedup).

Details

Store and arithmetical SSE/AVX 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)

Execution units bottlenecks

Found no such bottlenecks but see expert reports for more complex bottlenecks.

No data for this section

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.

omp_get_num_threads@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107
omp_get_thread_num@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

IDIV: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:108

Conversion instructions

Detected expensive conversion instructions.

Details

CLTD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:108

Workaround

Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This function is probably writing/copying data or processing integer elements.

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 44 bytes.

General properties

nb instructions	49
nb uops	49
loop length	160
used x86 registers	15
used mmx registers	0
used xmm registers	2
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	6.13 cycles
front end	6.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	3.83	3.83	3.83	3.83	3.83	3.83	2.00	2.00	2.00	2.00	0.00	0.00	0.00	0.00	0.00	0.00
cycles	3.83	3.83	3.83	3.83	3.83	3.83	2.00	2.00	2.00	2.00	0.00	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

P0: ALU
P1: ALU
P2: ALU
P3: ALU, BRU
P4: ALU, BRU
P5: ALU, BRU
P6 (512 bits): store address, load
P7 (512 bits): store address, load
P8 (512 bits): store address, load
P9 (512 bits): store address, load
P10 (512 bits): VPU, DIV/SQRT
P11 (512 bits): VPU, DIV/SQRT
P12 (512 bits): VPU
P13 (512 bits): VPU
P14 (512 bits): FP store data
P15 (512 bits): FP store data

Cycles executing div or sqrt instructions

6.00

Cycles summary

Front-end	6.13
Dispatch	3.83
DIV/SQRT	6.00
Overall L1	6.13

Vectorization ratios

INT

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)
other	0%

all	50%
load	0%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	100%

INT+FP

all	7%
load	0%
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	0%
other	10%

Vector efficiency ratios

INT

all	8%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	6%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	8%

all	18%
load	12%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	10%
load	12%
store	NA (no store vectorizable/vectorized instructions)
mul	6%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	6%
other	10%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 6.13 cycles. At this rate:

2% of peak load performance is reached (7.18 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the function is: 40edb0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
PUSH %R14	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV (%RDI),%RBX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV %RDI,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
CALL 403060 <omp_get_num_threads@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0.50	N/A
MOV %EAX,%R13D	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (6.3%)
CALL 403160 <omp_get_thread_num@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0.50	N/A
MOV %EAX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (6.3%)
MOV 0x18(%RBX),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc(%R14),%EAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
CLTD																				scal (6.3%)
IDIV %R13D	2	0	0	0	2	0	0	0	0	0	0	0	0	0	0	0	0	9-14	6	scal (6.3%)
CMP %EDX,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JL 40f043 <kineticEnergy._omp_fn.0+0x293>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
IMUL %EAX,%ECX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (6.3%)
VXORPD %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
ADD %ECX,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA (%RAX,%RDX,1),%ESI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %ESI,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JGE 40f01b <kineticEnergy._omp_fn.0+0x26b>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
MOV 0x78(%R14),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOVSXD %EDX,%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOV %EAX,%R11D	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
VMOVSD 0x1cbe(%RIP),%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
ADD %R10,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA (%R10,%R10,2),%R13	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV %R10,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
SAL $0x9,%R13	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
SAL $0x6,%R8	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
LEA (%RDI,%R10,4),%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LEA (%R11,%R11,2),%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
SAL $0x9,%R10	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
NOPL (%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
MOV 0x8(%R12),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
LEA 0x8(%R12),%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
POP %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R14	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
RET	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
INC %EAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.25	N/A
JMP 40ede4 <kineticEnergy._omp_fn.0+0x34>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	1	N/A
NOPL (%RAX)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run gcc_10

Source file and lines	timestep.c:107-116
Module	exec

The function is defined in /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107-116.

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 6.13 to 0.88 cycles (7.00x speedup).

Workaround

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

Vectorization

Details

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)

Execution units bottlenecks

Found no such bottlenecks but see expert reports for more complex bottlenecks.

No data for this section

CALL instructions

Detected function call instructions.

Details

omp_get_num_threads@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107
omp_get_thread_num@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:107

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

IDIV: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:108

Conversion instructions

Detected expensive conversion instructions.

Details

CLTD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:108

Workaround

Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This function is probably writing/copying data or processing integer elements.

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 44 bytes.

General properties

nb instructions	49
nb uops	49
loop length	160
used x86 registers	15
used mmx registers	0
used xmm registers	2
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	6.13 cycles
front end	6.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	3.83	3.83	3.83	3.83	3.83	3.83	2.00	2.00	2.00	2.00	0.00	0.00	0.00	0.00	0.00	0.00
cycles	3.83	3.83	3.83	3.83	3.83	3.83	2.00	2.00	2.00	2.00	0.00	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

P0: ALU
P1: ALU
P2: ALU
P3: ALU, BRU
P4: ALU, BRU
P5: ALU, BRU
P6 (512 bits): store address, load
P7 (512 bits): store address, load
P8 (512 bits): store address, load
P9 (512 bits): store address, load
P10 (512 bits): VPU, DIV/SQRT
P11 (512 bits): VPU, DIV/SQRT
P12 (512 bits): VPU
P13 (512 bits): VPU
P14 (512 bits): FP store data
P15 (512 bits): FP store data

Cycles executing div or sqrt instructions

6.00

Cycles summary

Front-end	6.13
Dispatch	3.83
DIV/SQRT	6.00
Overall L1	6.13

Vectorization ratios

INT

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)
other	0%

all	50%
load	0%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	100%

INT+FP

all	7%
load	0%
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	0%
other	10%

Vector efficiency ratios

INT

all	8%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	6%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	8%

all	18%
load	12%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	10%
load	12%
store	NA (no store vectorizable/vectorized instructions)
mul	6%
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	6%
other	10%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 6.13 cycles. At this rate:

2% of peak load performance is reached (7.18 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the function is: 40edb0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
PUSH %R14	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV (%RDI),%RBX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV %RDI,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
CALL 403060 <omp_get_num_threads@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0.50	N/A
MOV %EAX,%R13D	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (6.3%)
CALL 403160 <omp_get_thread_num@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0.50	N/A
MOV %EAX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (6.3%)
MOV 0x18(%RBX),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc(%R14),%EAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
CLTD																				scal (6.3%)
IDIV %R13D	2	0	0	0	2	0	0	0	0	0	0	0	0	0	0	0	0	9-14	6	scal (6.3%)
CMP %EDX,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JL 40f043 <kineticEnergy._omp_fn.0+0x293>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
IMUL %EAX,%ECX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (6.3%)
VXORPD %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
ADD %ECX,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA (%RAX,%RDX,1),%ESI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %ESI,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JGE 40f01b <kineticEnergy._omp_fn.0+0x26b>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
MOV 0x78(%R14),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOVSXD %EDX,%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOV %EAX,%R11D	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
VMOVSD 0x1cbe(%RIP),%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
ADD %R10,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA (%R10,%R10,2),%R13	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV %R10,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
SAL $0x9,%R13	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
SAL $0x6,%R8	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	scal (12.5%)
LEA (%RDI,%R10,4),%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LEA (%R11,%R11,2),%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
SAL $0x9,%R10	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
NOPL (%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
MOV 0x8(%R12),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
LEA 0x8(%R12),%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
POP %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R14	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
RET	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
INC %EAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.25	N/A
JMP 40ede4 <kineticEnergy._omp_fn.0+0x34>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	1	N/A
NOPL (%RAX)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A

Name	Coverage (%)	Time (s)
▼kineticEnergy._omp_fn.0–	0.13	0.03
▼Loop 102 - timestep.c:110-116 - exec–	0.02	0.01
○Loop 103 - timestep.c:110-116 - exec	0.11	0.02
○Loop 101 - timestep.c:107-107 - exec	0.00	0.00

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

CALL instructions

Details

Complex instructions

Details

Conversion instructions

Details

Workaround

Type of elements and instruction set

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

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

Execution units bottlenecks

CALL instructions

Details

Complex instructions

Details

Conversion instructions

Details

Workaround

Type of elements and instruction set

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code