OV - exec - Function sortAtomsById

	Function: sortAtomsById	Module: exec	Source: haloExchange.c:653-661	Coverage (incl. loops): 0.14% \| (excl. loops): 0.14%

/home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c: 653 - 661

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

653: {

654:    int aId = ((AtomMsg*) a)->gid;

655:    int bId = ((AtomMsg*) b)->gid;

656:    assert(aId != bId);

657:

658:    if (aId < bId)

659:       return -1;

660:    return 1;

661: }

0x407c00 MOV	(%RSI),%EAX

0x407c02 CMP	%EAX,(%RDI)

0x407c04 JE	407c11

0x407c06 SETGE	%DL

0x407c09 MOVZX	%DL,%ECX

0x407c0c LEA	-0x1(%RCX,%RCX,1),%EAX

0x407c10 RET

0x407c11 PUSH	%RBP

0x407c12 MOV	$0x410c30,%ECX

0x407c17 MOV	$0x290,%EDX

0x407c1c MOV	$0x410ae0,%ESI

0x407c21 MOV	%RSP,%RBP

0x407c24 MOV	$0x410ba7,%EDI

0x407c29 CALL	4030c0 <__assert_fail@plt>

0x407c2e XCHG	%AX,%AX

Coverage (%)	Name	Source Location	Module
►2.37+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►2.18+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►2.11+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►2.11+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.79+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.79+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.73+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.73+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.73+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.73+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.66+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.66+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.60+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.60+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.60+	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.60+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.60+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.54+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.54+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.47+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.41+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.41+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.34+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.28+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.28+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.22+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.15+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►1.02+	msort_with_tmp.part.0		libc.so.6
○	msort_with_tmp.part.0		libc.so.6
○	__GI___qsort_r		libc.so.6
○	sortAtomsInCell	haloExchange.c:633	exec
○	redistributeAtoms._omp_fn.0	timestep.c:154	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		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 0.14% of application time for run gcc_10

Source file and lines	haloExchange.c:653-661
Module	exec

The function is defined in /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:653-661.

The structure of this function is probably <if then [else] end>.

The presence of multiple execution paths is typically the main/first bottleneck.
Try to simplify control inside function: ideally, try to remove all conditional expressions, for example by (if applicable):

hoisting them (moving them outside the function)
turning them into conditional moves, MIN or MAX

Ex: if (x<0) x=0 => x = (x<0 ? 0 : x) (or MAX(0,x) after defining the corresponding macro)

gain
potential
hint
expert

Vectorization

Your function is not vectorized. Only 8% of vector register length is used (average across all SSE/AVX instructions).

Details

All SSE/AVX instructions are used in scalar version (process only one data element in vector registers).

No data for this section

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.

LEA: 0 occurrences

Conversion instructions

Detected expensive conversion instructions.

Details

MOVZX: 0 occurrences

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 8 bytes.

General properties

nb instructions	8.50
nb uops	9.50
loop length	26
used x86 registers	6
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE

micro-operation queue	1.19 cycles
front end	1.19 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	1.08	1.08	1.08	1.08	1.08	1.08	0.75	0.75	0.75	0.75	0.00	0.00	0.00	0.00	0.00	0.00
cycles	1.08	1.08	1.08	1.08	1.08	1.08	0.75	0.75	0.75	0.75	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

Cycles summary

Front-end	1.19
Dispatch	1.08
Overall L1	1.19

Vectorization ratios

all	0%
load	NA (no load vectorizable/vectorized instructions)
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	0%

Vector efficiency ratios

all	8%
load	NA (no load vectorizable/vectorized instructions)
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	8%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 1.19 cycles. At this rate:

2% of peak load performance is reached (6.91 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))

The code analyzed by CQA in that panel excludes loops and represents 0.14% of application time for run gcc_10

Source file and lines	haloExchange.c:653-661
Module	exec

hoisting them (moving them outside the function)
turning them into conditional moves, MIN or MAX

Ex: if (x<0) x=0 => x = (x<0 ? 0 : x) (or MAX(0,x) after defining the corresponding macro)

gain
potential
hint
expert

Vectorization

Your function is not vectorized.

Details

No SSE/AVX instruction: only general purpose registers are used and data elements are processed one at a time.

Execution units bottlenecks

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

No data for this section

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.

LEA: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:660

Conversion instructions

Detected expensive conversion instructions.

Details

MOVZX: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:660

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 8 bytes.

General properties

nb instructions	7
nb uops	8
loop length	17
used x86 registers	5
used mmx registers	0
used xmm registers	0
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	1.00 cycles
front end	1.00 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	1.00	1.00	1.00	1.00	1.00	1.00	0.75	0.75	0.75	0.75	0.00	0.00	0.00	0.00	0.00	0.00
cycles	1.00	1.00	1.00	1.00	1.00	1.00	0.75	0.75	0.75	0.75	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

Cycles summary

Front-end	1.00
Dispatch	1.00
Overall L1	1.00

Vectorization ratios

No vectorizable/vectorized instructions

Vector efficiency ratios

No vectorizable/vectorized instructions

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 1.00 cycles. At this rate:

3% of peak load performance is reached (8.00 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). By removing all these bottlenecks, you can lower the cost of an iteration from 1.00 to 0.75 cycles (1.33x speedup).

ASM code

In the binary file, the address of the function is: 407c00

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	Latency	Recip. throughput	Vectorization
MOV (%RSI),%EAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
CMP %EAX,(%RDI)	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	1	0.25	N/A
JE 407c11 <sortAtomsById+0x11>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	1	0.50-1	N/A
SETGE %DL	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOVZX %DL,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
LEA -0x1(%RCX,%RCX,1),%EAX	2	0	0	0.50	0.50	0	0	0.25	0.25	0.25	0.25	2	0.50	N/A
RET	1	0	0	0	0.33	0.33	0.33	0	0	0	0	1	0.50	N/A

The code analyzed by CQA in that panel excludes loops and represents 0.14% of application time for run gcc_10

Source file and lines	haloExchange.c:653-661
Module	exec

hoisting them (moving them outside the function)
turning them into conditional moves, MIN or MAX

Ex: if (x<0) x=0 => x = (x<0 ? 0 : x) (or MAX(0,x) after defining the corresponding macro)

gain
potential
hint
expert

Vectorization

Your function is not vectorized. Only 8% 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 1.38 to 0.09 cycles (16.00x speedup).

Details

All 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.

__assert_fail@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:656

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 8 bytes.

General properties

nb instructions	10
nb uops	11
loop length	35
used x86 registers	7
used mmx registers	0
used xmm registers	0
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	1.38 cycles
front end	1.38 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	1.17	1.17	1.17	1.17	1.17	1.17	0.75	0.75	0.75	0.75	0.00	0.00	0.00	0.00	0.00	0.00
cycles	1.17	1.17	1.17	1.17	1.17	1.17	0.75	0.75	0.75	0.75	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

Cycles summary

Front-end	1.38
Dispatch	1.17
Overall L1	1.38

Vectorization ratios

all	0%
load	NA (no load vectorizable/vectorized instructions)
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	0%

Vector efficiency ratios

all	8%
load	NA (no load vectorizable/vectorized instructions)
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	8%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 1.38 cycles. At this rate:

2% of peak load performance is reached (5.82 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

ASM code

In the binary file, the address of the function is: 407c00

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	Latency	Recip. throughput	Vectorization
MOV (%RSI),%EAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
CMP %EAX,(%RDI)	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	1	0.25	N/A
JE 407c11 <sortAtomsById+0x11>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	1	0.50-1	N/A
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV $0x410c30,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (6.3%)
MOV $0x290,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (6.3%)
MOV $0x410ae0,%ESI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV $0x410ba7,%EDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
CALL 4030c0 <__assert_fail@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0.50	N/A

Name	Coverage (%)	Time (s)
○sortAtomsById	0.14	0.04

Report Configuration

Vectorization

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

Vectorization

Details

Execution units bottlenecks

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

Vectorization

Details

Workaround

Execution units bottlenecks

CALL instructions

Details

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