OV - attention-gcc-native - Function std::mersenne_twister_engine<unsigned long, 32ul, 624ul, 397ul, 31ul, 2567483615ul, 11ul, 4294967295ul, 7ul, 2636928640ul, 15ul, 4022730752ul, 18ul, 1812433253ul>::_M_gen

	Function: std::mersenne_twister_engine<unsigned long, 32ul, 624ul, 397ul, 31ul, 2567483615ul, 11ul, ...	Module: attention-gcc-native	Source: random.tcc:404-425	Coverage (incl. loops): 0.29% \| (excl. loops): 0.00%

/usr/include/c++/14/bits/random.tcc: 404 - 425

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

404:       for (size_t __k = 0; __k < (__n - __m); ++__k)

405:         {

406: 	  _UIntType __y = ((_M_x[__k] & __upper_mask)

407: 			   | (_M_x[__k + 1] & __lower_mask));

408: 	  _M_x[__k] = (_M_x[__k + __m] ^ (__y >> 1)

409: 		       ^ ((__y & 0x01) ? __a : 0));

410:         }

411:

412:       for (size_t __k = (__n - __m); __k < (__n - 1); ++__k)

413: 	{

414: 	  _UIntType __y = ((_M_x[__k] & __upper_mask)

415: 			   | (_M_x[__k + 1] & __lower_mask));

416: 	  _M_x[__k] = (_M_x[__k + (__m - __n)] ^ (__y >> 1)

417: 		       ^ ((__y & 0x01) ? __a : 0));

418: 	}

419:

420:       _UIntType __y = ((_M_x[__n - 1] & __upper_mask)

421: 		       | (_M_x[0] & __lower_mask));

422:       _M_x[__n - 1] = (_M_x[__m - 1] ^ (__y >> 1)

423: 		       ^ ((__y & 0x01) ? __a : 0));

424:       _M_p = 0;

425:     }

0x411ee0 ADRP	X3,

0x411ee4 ADRP	X1,

0x411ee8 ADD	X2, X0, #1816

0x411eec LDR	Q29, [X3, #656]

0x411ef0 ADRP	X3,

0x411ef4 LDR	Q28, [X3, #672]

0x411ef8 LDR	Q31, [X1, #624]

0x411efc ADRP	X1,

0x411f00 LDR	Q30, [X1, #640]

0x411f04 ADD	X1, X0, #8

(41) 0x411f08 ADD	X1, X1, #16

(41) 0x411f0c LDUR	Q0, [X1, #488]

(41) 0x411f10 LDUR	Q26, [X1, #496]

(41) 0x411f14 LDR	Q25, [X1, #3152]

(41) 0x411f18 AND	V0.16B, V0.16B, V31.16B

(41) 0x411f1c AND	V26.16B, V26.16B, V30.16B

(41) 0x411f20 ORR	V26.16B, V0.16B, V26.16B

(41) 0x411f24 USHR	V27.2D, V26.2D, #1

(41) 0x411f28 AND	V26.16B, V26.16B, V29.16B

(41) 0x411f2c MUL	Z26.D, Z26.D, Z28.D

(41) 0x411f30 EOR3	V26.16B, V26.16B, V25.16B, V27.16B

(41) 0x411f34 STUR	Q26, [X1, #488]

(41) 0x411f38 CMP	X2, X1

(41) 0x411f3c B.NE	411f08

0x411f40 LDR	X4, [X0, #1808]

0x411f44 MOVZ	W6, #45279

0x411f48 ADRP	X8,

0x411f4c MOVK	W6, #39176

0x411f50 MOVZ	X1, #0

0x411f54 LDR	X5, [X0, #1816]

0x411f58 ADD	X3, X0, #1824

0x411f5c LDR	X7, [X0, #4984]

0x411f60 LDR	Q24, [X8, #624]

0x411f64 BFM	X4, X5, #0, #30

0x411f68 ADRP	X5,

0x411f6c LDR	Q23, [X5, #640]

0x411f70 AND	X5, X4, #0x1

0x411f74 EOR	X4, X7, X4,LSR #1

0x411f78 ADRP	X7,

0x411f7c UMULL	X5, W5, W6

0x411f80 ADRP	X6,

0x411f84 LDR	Q22, [X7, #656]

0x411f88 LDR	Q21, [X6, #672]

0x411f8c EOR	X4, X4, X5

0x411f90 STR	X4, [X0, #1808]

0x411f94 HINT	#0

0x411f98 HINT	#0

0x411f9c HINT	#0

(42) 0x411fa0 LDR	Q1, [X2, X1]

(42) 0x411fa4 LDR	Q19, [X3, X1]

(42) 0x411fa8 LDR	Q18, [X0, X1]

(42) 0x411fac AND	V1.16B, V1.16B, V24.16B

(42) 0x411fb0 AND	V19.16B, V19.16B, V23.16B

(42) 0x411fb4 ORR	V19.16B, V1.16B, V19.16B

(42) 0x411fb8 USHR	V20.2D, V19.2D, #1

(42) 0x411fbc AND	V19.16B, V19.16B, V22.16B

(42) 0x411fc0 MUL	Z19.D, Z19.D, Z21.D

(42) 0x411fc4 EOR3	V19.16B, V19.16B, V18.16B, V20.16B

(42) 0x411fc8 STR	Q19, [X2, X1]

(42) 0x411fcc ADD	X1, X1, #16

(42) 0x411fd0 CMP	X1, #3168

(42) 0x411fd4 B.NE	411fa0

0x411fd8 LDR	X2, [X0]

0x411fdc MOVZ	W3, #45279

0x411fe0 MOVK	W3, #39176

0x411fe4 STR	XZR, [X0, #4992]

0x411fe8 LDR	X1, [X0, #4984]

0x411fec LDR	X4, [X0, #3168]

0x411ff0 BFM	X1, X2, #0, #30

0x411ff4 AND	X2, X1, #0x1

0x411ff8 EOR	X1, X4, X1,LSR #1

0x411ffc UMULL	X2, W2, W3

0x412000 EOR	X1, X1, X2

0x412004 STR	X1, [X0, #4984]

0x412008 RET

Coverage (%)	Name	Source Location	Module
►100.00+	__libc_start_call_main		libc.so.6
○	__libc_start_main		libc.so.6
○	_start	stl_vector.h:1920	attention-gcc-native

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 run_0

Source file and lines	random.tcc:404-425
Module	attention-gcc-native

The function is defined in /usr/include/c++/14/bits/random.tcc:404-425.

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 5.83 to 2.67 cycles (2.19x speedup).

Workaround

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

Vectorization

Your function is poorly vectorized. Only 70% of vector register length is used (average across all VPU instructions). By fully vectorizing your function, you can lower the cost of an iteration from 5.83 to 4.80 cycles (1.22x speedup).

Details

40% of VPU instructions are used in vector version (process two or more data elements in vector registers):

72% of VPU loads are used in vector version.
0% of VPU stores are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU multiply instructions are used in vector version.
0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully 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)

No data for this section

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 does not load or store any data.

General properties

nb instructions	47
nb uops	44
loop length	188
used w registers	4
used x registers	10
used b registers	0
used h registers	0
used s registers	0
used d registers	0
used q registers	8
used v registers	0
used z registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	5.50 cycles
front end	5.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	4.33	4.33	4.33	4.33	4.33	4.33	0.00	0.00	0.00	0.00	5.83	5.50	5.67	1.50	1.50
cycles	0.50	0.50	4.33	4.33	4.33	4.33	4.33	4.33	0.00	0.00	0.00	0.00	5.83	5.50	5.67	1.50	1.50

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	5.50
Dispatch	5.83
Overall L1	5.83

Vectorization ratios

all	40%
load	72%
store	0%
mul	0%
add-sub	0%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Vector efficiency ratios

all	70%
load	86%
store	50%
mul	50%
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 5.83 cycles.

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the function is: 411ee0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
ADRP X3, <420ee0>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADRP X1, <420ee4>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD X2, X0, #1816	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
LDR Q29, [X3, #656]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADRP X3, <420ef0>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q28, [X3, #672]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
LDR Q31, [X1, #624]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADRP X1, <420efc>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q30, [X1, #640]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADD X1, X0, #8	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X4, [X0, #1808]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
MOVZ W6, #45279	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADRP X8, <420f48>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOVK W6, #39176	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVZ X1, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X5, [X0, #1816]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
ADD X3, X0, #1824	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X7, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR Q24, [X8, #624]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
BFM X4, X5, #0, #30	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	scal (50.0%)
ADRP X5, <420f68>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q23, [X5, #640]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
AND X5, X4, #0x1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
EOR X4, X7, X4,LSR #1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADRP X7, <420f78>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
UMULL X5, W5, W6	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	N/A
ADRP X6, <420f80>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q22, [X7, #656]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
LDR Q21, [X6, #672]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
EOR X4, X4, X5	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
STR X4, [X0, #1808]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
HINT #0																					N/A
HINT #0																					N/A
HINT #0																					N/A
LDR X2, [X0]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
MOVZ W3, #45279	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVK W3, #39176	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STR XZR, [X0, #4992]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDR X1, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR X4, [X0, #3168]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
BFM X1, X2, #0, #30	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	N/A
AND X2, X1, #0x1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
EOR X1, X4, X1,LSR #1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
UMULL X2, W2, W3	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	scal (50.0%)
EOR X1, X1, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STR X1, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
RET	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run run_0

Source file and lines	random.tcc:404-425
Module	attention-gcc-native

The function is defined in /usr/include/c++/14/bits/random.tcc:404-425.

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 5.83 to 2.67 cycles (2.19x speedup).

Workaround

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

Vectorization

Details

40% of VPU instructions are used in vector version (process two or more data elements in vector registers):

72% of VPU loads are used in vector version.
0% of VPU stores are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU multiply instructions are used in vector version.
0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully 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)

No data for this section

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 does not load or store any data.

General properties

nb instructions	47
nb uops	44
loop length	188
used w registers	4
used x registers	10
used b registers	0
used h registers	0
used s registers	0
used d registers	0
used q registers	8
used v registers	0
used z registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	5.50 cycles
front end	5.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	4.33	4.33	4.33	4.33	4.33	4.33	0.00	0.00	0.00	0.00	5.83	5.50	5.67	1.50	1.50
cycles	0.50	0.50	4.33	4.33	4.33	4.33	4.33	4.33	0.00	0.00	0.00	0.00	5.83	5.50	5.67	1.50	1.50

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	5.50
Dispatch	5.83
Overall L1	5.83

Vectorization ratios

all	40%
load	72%
store	0%
mul	0%
add-sub	0%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Vector efficiency ratios

all	70%
load	86%
store	50%
mul	50%
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 5.83 cycles.

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the function is: 411ee0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
ADRP X3, <420ee0>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADRP X1, <420ee4>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD X2, X0, #1816	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
LDR Q29, [X3, #656]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADRP X3, <420ef0>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q28, [X3, #672]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
LDR Q31, [X1, #624]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADRP X1, <420efc>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q30, [X1, #640]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
ADD X1, X0, #8	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X4, [X0, #1808]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
MOVZ W6, #45279	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADRP X8, <420f48>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOVK W6, #39176	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVZ X1, #0	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X5, [X0, #1816]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
ADD X3, X0, #1824	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LDR X7, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR Q24, [X8, #624]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
BFM X4, X5, #0, #30	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	scal (50.0%)
ADRP X5, <420f68>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q23, [X5, #640]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
AND X5, X4, #0x1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
EOR X4, X7, X4,LSR #1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADRP X7, <420f78>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
UMULL X5, W5, W6	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	N/A
ADRP X6, <420f80>	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.25	N/A
LDR Q22, [X7, #656]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
LDR Q21, [X6, #672]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	6	0.33	vect (100.0%)
EOR X4, X4, X5	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
STR X4, [X0, #1808]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
HINT #0																					N/A
HINT #0																					N/A
HINT #0																					N/A
LDR X2, [X0]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
MOVZ W3, #45279	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVK W3, #39176	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STR XZR, [X0, #4992]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
LDR X1, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	N/A
LDR X4, [X0, #3168]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	0	0	4	0.33	scal (50.0%)
BFM X1, X2, #0, #30	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	N/A
AND X2, X1, #0x1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
EOR X1, X4, X1,LSR #1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
UMULL X2, W2, W3	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	0	0	0	0	0	2	0.50	scal (50.0%)
EOR X1, X1, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
STR X1, [X0, #4984]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0.50	0.50	1	0.50	scal (50.0%)
RET	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Name	Coverage (%)	Time (s)
▼std::mersenne_twister_engine::_M_gen_rand()–	0.29	0.02
○Loop 42 - random.tcc:412-417 - attention-gcc-native	0.17	0.01
○Loop 41 - random.tcc:404-409 - attention-gcc-native	0.12	0.01

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

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

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