OV - exec - Function __nv_revert_kernel_module_revert_kernel__F1L41_1

	Function: __nv_revert_kernel_module_revert_kernel__F1L41_1_	Module: exec	Source: revert_kernel.f90-pp.f90:41-53	Coverage: 1.19%

/home/hbollore/qaas-runs/170-307-1706/intel/CloverLeafFC/build/armclang_5/CMakeFiles/clover_leaf.dir/CloverLeaf_ref/kernels/revert_kernel.f90-pp.f90: 41 - 53

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

41: # 41 "/home/hbollore/qaas-runs/170-307-1706/intel/CloverLeafFC/build/CloverLeafFC/CloverLeaf_ref/kernels/revert_kernel.f90"

42:     !$OMP PARALLEL

43: # 43 "/home/hbollore/qaas-runs/170-307-1706/intel/CloverLeafFC/build/CloverLeafFC/CloverLeaf_ref/kernels/revert_kernel.f90"

44:     !$OMP DO

45:     DO k=y_min,y_max

46:     !$OMP SIMD

47:       DO j=x_min,x_max

48:         density1(j,k)=density0(j,k)

49:         energy1(j,k)=energy0(j,k)

50:       ENDDO

51:     ENDDO

52:   !$OMP END DO

53: # 53 "/home/hbollore/qaas-runs/170-307-1706/intel/CloverLeafFC/build/CloverLeafFC/CloverLeaf_ref/kernels/revert_kernel.f90"

0x42e4f0 SUB	SP, SP, #128

0x42e4f4 STP	X29, X30, [SP, #32]

0x42e4f8 STP	X28, X27, [SP, #48]

0x42e4fc STP	X26, X25, [SP, #64]

0x42e500 STP	X24, X23, [SP, #80]

0x42e504 STP	X22, X21, [SP, #96]

0x42e508 STP	X20, X19, [SP, #112]

0x42e50c LDP	X8, X9, [X2]

0x42e510 LDR	W19, [X0]

0x42e514 ORR	X0, XZR, XZR

0x42e518 ORR	X20, XZR, X2

0x42e51c ADD	X3, SP, #28

0x42e520 ADD	X4, SP, #24

0x42e524 ADD	X5, SP, #20

0x42e528 STR	WZR, [SP, #28]

0x42e52c ADD	X6, SP, #16

0x42e530 MOVZ	W2, #34

0x42e534 MOVZ	W7, #1

0x42e538 LDR	W21, [X9]

0x42e53c LDR	W8, [X8]

0x42e540 ORR	W1, WZR, W19

0x42e544 STP	W21, W8, [SP, #20]

0x42e548 MOVZ	W8, #1

0x42e54c STR	W8, [SP, #16]

0x42e550 STR	W8, [SP]

0x42e554 BL	404ae0

0x42e558 LDRSW	X11, [SP, #24]

0x42e55c CMP	W11, W21

0x42e560 B.GT	42e6e0

0x42e564 LDR	W8, [SP, #20]

0x42e568 SUBS	W15, W8, W11

0x42e56c B.MI	42e6e0

0x42e570 LDP	X9, X8, [X20, #16]

0x42e574 LDR	W12, [X8]

0x42e578 LDRSW	X8, [X9]

0x42e57c SUBS	W10, W12, W8

0x42e580 B.MI	42e6e0

0x42e584 LDR	X9, [X20, #56]

0x42e588 ADDS	W10, W10, #1

0x42e58c SUB	W12, W12, W8

0x42e590 CNTW	X16, ALL

0x42e594 ADD	W15, W15, #1

0x42e598 CSINC	W14, WZR, WZR, #0

0x42e59c PTRUE	P0.D, ALL

0x42e5a0 SUB	W17, W12, W14

0x42e5a4 LDR	X12, [X20, #32]

0x42e5a8 LDR	X9, [X9]

0x42e5ac ADD	W17, W17, #1

0x42e5b0 ADD	X17, X17, #1

0x42e5b4 MADD	X21, X9, X11, XZR

0x42e5b8 LDP	X13, X11, [X20, #96]

0x42e5bc UBFM	X18, X9, #61, #60

0x42e5c0 UBFM	X14, X21, #61, #60

0x42e5c4 ADD	X14, X14, X8,LSL #3

0x42e5c8 LDR	X22, [X13]

0x42e5cc ADD	X13, X21, X8

0x42e5d0 SUB	X1, X13, X22

0x42e5d4 SUB	X6, X14, X22,LSL #3

0x42e5d8 LDP	X14, X13, [X20, #112]

0x42e5dc SUB	X20, X21, X22

0x42e5e0 UBFM	X7, X1, #61, #60

0x42e5e4 ADD	X4, X12, X6

0x42e5e8 ADD	X5, X12, X7

0x42e5ec ADD	X0, X14, X6

0x42e5f0 ADD	X1, X14, X7

0x42e5f4 ADD	X2, X13, X6

0x42e5f8 ADD	X3, X13, X7

0x42e5fc ADD	X6, X11, X6

0x42e600 ADD	X7, X11, X7

0x42e604 B	42e634

(701) 0x42e608 ADD	X0, X0, X18

(701) 0x42e60c ADD	X1, X1, X18

(701) 0x42e610 SUBS	W15, W15, #1

(701) 0x42e614 ADD	X2, X2, X18

(701) 0x42e618 ADD	X3, X3, X18

(701) 0x42e61c ADD	X4, X4, X18

(701) 0x42e620 ADD	X5, X5, X18

(701) 0x42e624 ADD	X6, X6, X18

(701) 0x42e628 ADD	X7, X7, X18

(701) 0x42e62c ADD	X20, X20, X9

(701) 0x42e630 B.LE	42e6e0

(701) 0x42e634 ORR	X22, XZR, X8

(701) 0x42e638 ORR	W21, WZR, W10

(701) 0x42e63c CMP	X17, X16

(701) 0x42e640 B.CC	42e69c

(701) 0x42e644 UDIV	X21, X17, X16

(701) 0x42e648 ORR	X24, XZR, XZR

(701) 0x42e64c ADDVL	X26, X1, #1

(701) 0x42e650 ADDVL	X27, X3, #1

(701) 0x42e654 ADDVL	X28, X5, #1

(701) 0x42e658 ADDVL	X29, X7, #1

(701) 0x42e65c MADD	X25, X21, X16, XZR

(701) 0x42e660 SUB	X23, X17, X25

(701) 0x42e664 ADD	X22, X25, X8

(701) 0x42e668 SUB	W21, W10, W25

  (702) 0x42e66c LD1D	{Z0.D}, P0/Z, [X6, X24,LSL #3]

  (702) 0x42e670 LD1D	{Z1.D}, P0/Z, [X29, X24,LSL #3]

  (702) 0x42e674 ST1D	{Z0.D}, P0, [X4, X24,LSL #3]

  (702) 0x42e678 ST1D	{Z1.D}, P0, [X28, X24,LSL #3]

  (702) 0x42e67c LD1D	{Z0.D}, P0/Z, [X2, X24,LSL #3]

  (702) 0x42e680 LD1D	{Z1.D}, P0/Z, [X27, X24,LSL #3]

  (702) 0x42e684 ST1D	{Z0.D}, P0, [X0, X24,LSL #3]

  (702) 0x42e688 ST1D	{Z1.D}, P0, [X26, X24,LSL #3]

  (702) 0x42e68c ADD	X24, X24, X16

  (702) 0x42e690 CMP	X25, X24

  (702) 0x42e694 B.NE	42e66c

(701) 0x42e698 CBZ	X23, 42e608

(701) 0x42e69c ADD	X22, X22, X20

(701) 0x42e6a0 ADD	W21, W21, #1

(701) 0x42e6a4 UBFM	X25, X22, #61, #60

(701) 0x42e6a8 ADD	X22, X13, X25

(701) 0x42e6ac ADD	X23, X14, X25

(701) 0x42e6b0 ADD	X24, X12, X25

(701) 0x42e6b4 ADD	X25, X11, X25

(701) 0x42e6b8 HINT	#0

(701) 0x42e6bc HINT	#0

  (700) 0x42e6c0 LDR	D0, [X25], #8

  (700) 0x42e6c4 SUB	W21, W21, #1

  (700) 0x42e6c8 STR	D0, [X24], #8

  (700) 0x42e6cc CMP	W21, #1

  (700) 0x42e6d0 LDR	D0, [X22], #8

  (700) 0x42e6d4 STR	D0, [X23], #8

  (700) 0x42e6d8 B.HI	42e6c0

(701) 0x42e6dc B	42e608

0x42e6e0 ORR	X0, XZR, XZR

0x42e6e4 ORR	W1, WZR, W19

0x42e6e8 BL	404bf0

0x42e6ec ORR	X0, XZR, XZR

0x42e6f0 ORR	W1, WZR, W19

0x42e6f4 BL	4047f0

0x42e6f8 LDP	X20, X19, [SP, #112]

0x42e6fc LDP	X22, X21, [SP, #96]

0x42e700 LDP	X24, X23, [SP, #80]

0x42e704 LDP	X26, X25, [SP, #64]

0x42e708 LDP	X28, X27, [SP, #48]

0x42e70c LDP	X29, X30, [SP, #32]

0x42e710 ADD	SP, SP, #128

0x42e714 RET

0x42e718 HINT	#0

0x42e71c HINT	#0

Path /

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

Source file and lines	revert_kernel.f90-pp.f90:41-53
Module	exec

The function is defined in /home/hbollore/qaas-runs/170-307-1706/intel/CloverLeafFC/build/armclang_5/CMakeFiles/clover_leaf.dir/CloverLeaf_ref/kernels/revert_kernel.f90-pp.f90:41-53.

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 12.00 to 1.00 cycles (12.00x speedup).

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)
To reference allocatable arrays, use "allocatable" instead of "pointer" pointers or qualify them with the "contiguous" attribute (Fortran 2008)
For structures, limit to one indirection. For example, use a_b%c instead of a%b%c with a_b set to a%b before this function

Vectorization

Your function is not vectorized. Only 22% of vector register length is used (average across all VPU instructions). By vectorizing your function, you can lower the cost of an iteration from 12.00 to 10.33 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): do i a(i)%x = b(i)%x (slow, non stride 1) => do i a%x(i) = b%x(i) (fast, stride 1)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 12.00 to 1.50 cycles (8.00x 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 220 bytes. The binary function is storing 116 bytes.

General properties

nb instructions	86
loop length	344
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

10.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	4.00	4.00	12.00	12.00	12.00	12.00	0.00	0.00	0.00	0.00	10.33	10.33	10.33	5.00	5.00
cycles	4.00	4.00	12.00	12.00	12.00	12.00	0.00	0.00	0.00	0.00	10.33	10.33	10.33	5.00	5.00

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	10.50
Overall L1	12.00

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	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 12.00 cycles. At this rate:

19% of peak load performance is reached (18.33 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
20% of peak store performance is reached (9.67 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 42e4f0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput
SUB SP, SP, #128	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STP X29, X30, [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
STP X28, X27, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X26, X25, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X24, X23, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X22, X21, [SP, #96]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X20, X19, [SP, #112]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
LDP X8, X9, [X2]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDR W19, [X0]	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, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR X20, XZR, X2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X3, SP, #28	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X4, SP, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X5, SP, #20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR WZR, [SP, #28]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
ADD X6, SP, #16	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MOVZ W2, #34	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MOVZ W7, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR W21, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDR W8, [X8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STP W21, W8, [SP, #20]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
MOVZ W8, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR W8, [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
STR W8, [SP]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
BL 404ae0 <@plt_start@+0x320>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDRSW X11, [SP, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
CMP W11, W21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.GT 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR W8, [SP, #20]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SUBS W15, W8, W11	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.MI 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDP X9, X8, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDR W12, [X8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDRSW X8, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SUBS W10, W12, W8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.MI 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X9, [X20, #56]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADDS W10, W10, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
SUB W12, W12, W8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
CNTW X16, ALL	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
ADD W15, W15, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
CSINC W14, WZR, WZR, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
PTRUE P0.D, ALL	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
SUB W17, W12, W14	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X12, [X20, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDR X9, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD W17, W17, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X17, X17, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MADD X21, X9, X11, XZR	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
LDP X13, X11, [X20, #96]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
UBFM X18, X9, #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
UBFM X14, X21, #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
ADD X14, X14, X8,LSL #3	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, [X13]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD X13, X21, X8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
SUB X1, X13, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
SUB X6, X14, X22,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDP X14, X13, [X20, #112]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
SUB X20, X21, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
UBFM X7, X1, #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
ADD X4, X12, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X5, X12, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X0, X14, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X1, X14, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X2, X13, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X3, X13, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X6, X11, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X7, X11, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
B 42e634 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x144>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
ORR X0, XZR, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 404bf0 <@plt_start@+0x430>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
ORR X0, XZR, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 4047f0 <@plt_start@+0x30>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDP X20, X19, [SP, #112]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X22, X21, [SP, #96]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X24, X23, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X26, X25, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X28, X27, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X29, X30, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
ADD SP, SP, #128	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
RET	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

Source file and lines	revert_kernel.f90-pp.f90:41-53
Module	exec

gain
potential
hint
expert

Code clean check

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)
To reference allocatable arrays, use "allocatable" instead of "pointer" pointers or qualify them with the "contiguous" attribute (Fortran 2008)
For structures, limit to one indirection. For example, use a_b%c instead of a%b%c with a_b set to a%b before this function

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): do i a(i)%x = b(i)%x (slow, non stride 1) => do i a%x(i) = b%x(i) (fast, stride 1)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 12.00 to 1.50 cycles (8.00x 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 220 bytes. The binary function is storing 116 bytes.

General properties

nb instructions	86
loop length	344
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

10.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	4.00	4.00	12.00	12.00	12.00	12.00	0.00	0.00	0.00	0.00	10.33	10.33	10.33	5.00	5.00
cycles	4.00	4.00	12.00	12.00	12.00	12.00	0.00	0.00	0.00	0.00	10.33	10.33	10.33	5.00	5.00

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	10.50
Overall L1	12.00

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	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 12.00 cycles. At this rate:

19% of peak load performance is reached (18.33 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
20% of peak store performance is reached (9.67 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the function is: 42e4f0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput
SUB SP, SP, #128	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STP X29, X30, [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
STP X28, X27, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X26, X25, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X24, X23, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X22, X21, [SP, #96]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
STP X20, X19, [SP, #112]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
LDP X8, X9, [X2]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDR W19, [X0]	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, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR X20, XZR, X2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X3, SP, #28	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X4, SP, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X5, SP, #20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR WZR, [SP, #28]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
ADD X6, SP, #16	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MOVZ W2, #34	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MOVZ W7, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR W21, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDR W8, [X8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STP W21, W8, [SP, #20]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
MOVZ W8, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
STR W8, [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
STR W8, [SP]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50
BL 404ae0 <@plt_start@+0x320>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDRSW X11, [SP, #24]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
CMP W11, W21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.GT 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR W8, [SP, #20]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SUBS W15, W8, W11	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.MI 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDP X9, X8, [X20, #16]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDR W12, [X8]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDRSW X8, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
SUBS W10, W12, W8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
B.MI 42e6e0 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x1f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDR X9, [X20, #56]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADDS W10, W10, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33
SUB W12, W12, W8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
CNTW X16, ALL	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
ADD W15, W15, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
CSINC W14, WZR, WZR, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
PTRUE P0.D, ALL	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
SUB W17, W12, W14	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDR X12, [X20, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
LDR X9, [X9]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD W17, W17, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X17, X17, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
MADD X21, X9, X11, XZR	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	2	1
LDP X13, X11, [X20, #96]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
UBFM X18, X9, #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
UBFM X14, X21, #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
ADD X14, X14, X8,LSL #3	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, [X13]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33
ADD X13, X21, X8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
SUB X1, X13, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
SUB X6, X14, X22,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
LDP X14, X13, [X20, #112]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
SUB X20, X21, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
UBFM X7, X1, #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
ADD X4, X12, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X5, X12, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X0, X14, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X1, X14, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X2, X13, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X3, X13, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X6, X11, X6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ADD X7, X11, X7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
B 42e634 <__nv_revert_kernel_module_revert_kernel__F1L41_1_+0x144>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50
ORR X0, XZR, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 404bf0 <@plt_start@+0x430>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
ORR X0, XZR, XZR	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
ORR W1, WZR, W19	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
BL 4047f0 <@plt_start@+0x30>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
LDP X20, X19, [SP, #112]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X22, X21, [SP, #96]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X24, X23, [SP, #80]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X26, X25, [SP, #64]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X28, X27, [SP, #48]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
LDP X29, X30, [SP, #32]	1	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	1
ADD SP, SP, #128	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25
RET	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

Name	Coverage (%)	Time (s)
▼__nv_revert_kernel_module_revert_kernel__F1L41_1_–	1.19	0.24
▼Loop 701 - revert_kernel.f90-pp.f90:46-50 - exec–	0	0
○Loop 702 - revert_kernel.f90-pp.f90:47-48 - exec	1.19	0.24
○Loop 700 - revert_kernel.f90-pp.f90:47-49 - 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