OV - attention-clang-znver5-256 - Loop 47 - attention-clang-znver5-256

	Loop Id: 47	Module: attention-clang-znver5-256	Source: attention_v2.cpp:52-53	Coverage: 0.28%

0x6043 VMOVAPS	%YMM1,0x460(%RSP)    [1]

0x604c VMOVUPS	-0x60(%RDX,%R13,4),%YMM0    [2]

0x6053 VMOVAPS	0x440(%RSP),%YMM4    [1]

0x605c VMOVUPS	-0x40(%RDX,%R13,4),%YMM1    [2]

0x6063 VMOVUPS	-0x20(%RDX,%R13,4),%YMM2    [2]

0x606a VMOVUPS	(%RDX,%R13,4),%YMM3    [2]

0x6070 VSUBPS	%YMM4,%YMM0,%YMM5

0x6074 VSUBPS	%YMM4,%YMM1,%YMM0

0x6078 VMOVAPS	%YMM0,0xa0(%RSP)    [1]

0x6081 VSUBPS	%YMM4,%YMM2,%YMM0

0x6085 VMOVAPS	%YMM5,0xe0(%RSP)    [1]

0x608e VMOVAPS	%YMM0,0x120(%RSP)    [1]

0x6097 VSUBPS	%YMM4,%YMM3,%YMM0

0x609b VMOVAPS	%YMM0,0x140(%RSP)    [1]

0x60a4 VEXTRACTF128	$0x1,%YMM5,%XMM0

0x60aa VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x60b0 VZEROUPPER

0x60b3 CALL	1160 <expf@plt>

0x60b8 VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x60c1 VMOVSHDUP	0x70(%RSP),%XMM0    [1]

0x60c7 CALL	1160 <expf@plt>

0x60cc VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x60d5 VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x60db VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x60e4 VPERMILPD	$0x1,0x70(%RSP),%XMM0    [1]

0x60ec CALL	1160 <expf@plt>

0x60f1 VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x60fa VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x6100 VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x6109 VPERMILPS	$-0x1,0x70(%RSP),%XMM0    [1]

0x6111 CALL	1160 <expf@plt>

0x6116 VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x611f VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x6125 VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x612b VMOVAPS	0xe0(%RSP),%YMM0    [1]

0x6134 VZEROUPPER

0x6137 CALL	1160 <expf@plt>

0x613c VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x6145 VMOVSHDUP	0xe0(%RSP),%XMM0    [1]

0x614e CALL	1160 <expf@plt>

0x6153 VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x615c VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x6162 VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x616b VPERMILPD	$0x1,0xe0(%RSP),%XMM0    [1]

0x6176 CALL	1160 <expf@plt>

0x617b VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x6184 VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x618a VMOVAPS	%XMM0,0x1d0(%RSP)    [1]

0x6193 VPERMILPS	$-0x1,0xe0(%RSP),%XMM0    [1]

0x619e CALL	1160 <expf@plt>

0x61a3 VMOVAPS	0x1d0(%RSP),%XMM1    [1]

0x61ac VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x61b2 VINSERTF128	$0x1,0x70(%RSP),%YMM0,%YMM0    [1]

0x61ba VMOVAPS	0x320(%RSP),%YMM1    [1]

0x61c3 VADDPS	%YMM1,%YMM0,%YMM1

0x61c7 VMOVAPS	0xa0(%RSP),%YMM0    [1]

0x61d0 VMOVAPS	%YMM1,0x320(%RSP)    [1]

0x61d9 VEXTRACTF128	$0x1,%YMM0,%XMM0

0x61df VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x61e8 VZEROUPPER

0x61eb CALL	1160 <expf@plt>

0x61f0 VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x61f6 VMOVSHDUP	0xe0(%RSP),%XMM0    [1]

0x61ff CALL	1160 <expf@plt>

0x6204 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x620a VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x6210 VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x6216 VPERMILPD	$0x1,0xe0(%RSP),%XMM0    [1]

0x6221 CALL	1160 <expf@plt>

0x6226 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x622c VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x6232 VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x6238 VPERMILPS	$-0x1,0xe0(%RSP),%XMM0    [1]

0x6243 CALL	1160 <expf@plt>

0x6248 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x624e VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x6254 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x625d VMOVAPS	0xa0(%RSP),%YMM0    [1]

0x6266 VZEROUPPER

0x6269 CALL	1160 <expf@plt>

0x626e VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x6274 VMOVSHDUP	0xa0(%RSP),%XMM0    [1]

0x627d CALL	1160 <expf@plt>

0x6282 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x6288 VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x628e VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x6294 VPERMILPD	$0x1,0xa0(%RSP),%XMM0    [1]

0x629f CALL	1160 <expf@plt>

0x62a4 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x62aa VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x62b0 VMOVAPS	%XMM0,0x70(%RSP)    [1]

0x62b6 VPERMILPS	$-0x1,0xa0(%RSP),%XMM0    [1]

0x62c1 CALL	1160 <expf@plt>

0x62c6 VMOVAPS	0x70(%RSP),%XMM1    [1]

0x62cc VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x62d2 VINSERTF128	$0x1,0xe0(%RSP),%YMM0,%YMM0    [1]

0x62dd VMOVAPS	0x3e0(%RSP),%YMM1    [1]

0x62e6 VADDPS	%YMM1,%YMM0,%YMM1

0x62ea VMOVAPS	0x120(%RSP),%YMM0    [1]

0x62f3 VMOVAPS	%YMM1,0x3e0(%RSP)    [1]

0x62fc VEXTRACTF128	$0x1,%YMM0,%XMM0

0x6302 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x630b VZEROUPPER

0x630e CALL	1160 <expf@plt>

0x6313 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x631c VMOVSHDUP	0xa0(%RSP),%XMM0    [1]

0x6325 CALL	1160 <expf@plt>

0x632a VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x6333 VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x6339 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x6342 VPERMILPD	$0x1,0xa0(%RSP),%XMM0    [1]

0x634d CALL	1160 <expf@plt>

0x6352 VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x635b VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x6361 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x636a VPERMILPS	$-0x1,0xa0(%RSP),%XMM0    [1]

0x6375 CALL	1160 <expf@plt>

0x637a VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x6383 VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x6389 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x6392 VMOVAPS	0x120(%RSP),%YMM0    [1]

0x639b VZEROUPPER

0x639e CALL	1160 <expf@plt>

0x63a3 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x63ac VMOVSHDUP	0x120(%RSP),%XMM0    [1]

0x63b5 CALL	1160 <expf@plt>

0x63ba VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x63c3 VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x63c9 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x63d2 VPERMILPD	$0x1,0x120(%RSP),%XMM0    [1]

0x63dd CALL	1160 <expf@plt>

0x63e2 VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x63eb VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x63f1 VMOVAPS	%XMM0,0xe0(%RSP)    [1]

0x63fa VPERMILPS	$-0x1,0x120(%RSP),%XMM0    [1]

0x6405 CALL	1160 <expf@plt>

0x640a VMOVAPS	0xe0(%RSP),%XMM1    [1]

0x6413 VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x6419 VINSERTF128	$0x1,0xa0(%RSP),%YMM0,%YMM0    [1]

0x6424 VMOVAPS	0x400(%RSP),%YMM1    [1]

0x642d VADDPS	%YMM1,%YMM0,%YMM1

0x6431 VMOVAPS	0x140(%RSP),%YMM0    [1]

0x643a VMOVAPS	%YMM1,0x400(%RSP)    [1]

0x6443 VEXTRACTF128	$0x1,%YMM0,%XMM0

0x6449 VMOVAPS	%XMM0,0x120(%RSP)    [1]

0x6452 VZEROUPPER

0x6455 CALL	1160 <expf@plt>

0x645a VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x6463 VMOVSHDUP	0x120(%RSP),%XMM0    [1]

0x646c CALL	1160 <expf@plt>

0x6471 VMOVAPS	0xa0(%RSP),%XMM1    [1]

0x647a VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x6480 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x6489 VPERMILPD	$0x1,0x120(%RSP),%XMM0    [1]

0x6494 CALL	1160 <expf@plt>

0x6499 VMOVAPS	0xa0(%RSP),%XMM1    [1]

0x64a2 VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x64a8 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x64b1 VPERMILPS	$-0x1,0x120(%RSP),%XMM0    [1]

0x64bc CALL	1160 <expf@plt>

0x64c1 VMOVAPS	0xa0(%RSP),%XMM1    [1]

0x64ca VINSERTPS	$0x30,%XMM0,%XMM1,%XMM0

0x64d0 VMOVAPS	%XMM0,0x120(%RSP)    [1]

0x64d9 VMOVAPS	0x140(%RSP),%YMM0    [1]

0x64e2 VZEROUPPER

0x64e5 CALL	1160 <expf@plt>

0x64ea VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x64f3 VMOVSHDUP	0x140(%RSP),%XMM0    [1]

0x64fc CALL	1160 <expf@plt>

0x6501 VMOVAPS	0xa0(%RSP),%XMM1    [1]

0x650a VINSERTPS	$0x10,%XMM0,%XMM1,%XMM0

0x6510 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x6519 VPERMILPD	$0x1,0x140(%RSP),%XMM0    [1]

0x6524 CALL	1160 <expf@plt>

0x6529 VMOVAPS	0xa0(%RSP),%XMM1    [1]

0x6532 VINSERTPS	$0x20,%XMM0,%XMM1,%XMM0

0x6538 VMOVAPS	%XMM0,0xa0(%RSP)    [1]

0x6541 VPERMILPS	$-0x1,0x140(%RSP),%XMM0    [1]

0x654c CALL	1160 <expf@plt>

0x6551 VMOVAPS	0xa0(%RSP),%XMM2    [1]

0x655a VMOVAPS	0x460(%RSP),%YMM1    [1]

0x6563 MOV	0x2e0(%RSP),%RDX    [1]

0x656b ADD	$0x20,%R13

0x656f VINSERTPS	$0x30,%XMM0,%XMM2,%XMM0

0x6575 VINSERTF128	$0x1,0x120(%RSP),%YMM0,%YMM0    [1]

0x6580 VADDPS	%YMM1,%YMM0,%YMM1

0x6584 CMP	%R13,%R14

0x6587 JNE	6043

/home/eoseret/llm-attention/attention_v2.cpp: 52 - 53

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

52:       for (int idx = 0; idx <= row; ++idx) // vectorised

53:          sum += expf(S_row[idx] - max_val);

Coverage (%)	Name	Source Location	Module

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 /

Metric	Value
CQA speedup if no scalar integer	1.01
CQA speedup if FP arith vectorized	1.01
CQA speedup if fully vectorized	3.42
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.00
Bottlenecks	micro-operation queue,
Function	main
Source	attention_v2.cpp:52-53
Source loop unroll info	unrolled by 32
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	32
CQA cycles	27.38
CQA cycles if no scalar integer	27.00
CQA cycles if FP arith vectorized	27.13
CQA cycles if fully vectorized	8.02
Front-end cycles	27.38
P0 cycles	0.33
P1 cycles	0.33
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.33
P5 cycles	0.33
P6 cycles	27.25
P7 cycles	27.25
P8 cycles	27.25
P9 cycles	27.25
P10 cycles	14.00
P11 cycles	14.00
P12 cycles	14.00
P13 cycles	14.00
P14 cycles	20.00
P15 cycles	20.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	188.00
Nb uops	219.00
Nb loads	69.00
Nb stores	40.00
Nb stack references	12.00
FLOP/cycle	2.34
Nb FLOP add-sub	64.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	77.44
Bytes prefetched	0.00
Bytes loaded	1352.00
Bytes stored	768.00
Stride 0	1.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	84.21
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	100.00
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	57.14
Vector-efficiency ratio all	26.64
Vector-efficiency ratio load	29.41
Vector-efficiency ratio store	30.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	50.00
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	16.96

Metric	Value
CQA speedup if no scalar integer	1.01
CQA speedup if FP arith vectorized	1.01
CQA speedup if fully vectorized	3.42
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.00
Bottlenecks	micro-operation queue,
Function	main
Source	attention_v2.cpp:52-53
Source loop unroll info	unrolled by 32
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	32
CQA cycles	27.38
CQA cycles if no scalar integer	27.00
CQA cycles if FP arith vectorized	27.13
CQA cycles if fully vectorized	8.02
Front-end cycles	27.38
P0 cycles	0.33
P1 cycles	0.33
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.33
P5 cycles	0.33
P6 cycles	27.25
P7 cycles	27.25
P8 cycles	27.25
P9 cycles	27.25
P10 cycles	14.00
P11 cycles	14.00
P12 cycles	14.00
P13 cycles	14.00
P14 cycles	20.00
P15 cycles	20.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	188.00
Nb uops	219.00
Nb loads	69.00
Nb stores	40.00
Nb stack references	12.00
FLOP/cycle	2.34
Nb FLOP add-sub	64.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	77.44
Bytes prefetched	0.00
Bytes loaded	1352.00
Bytes stored	768.00
Stride 0	1.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	84.21
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	100.00
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	57.14
Vector-efficiency ratio all	26.64
Vector-efficiency ratio load	29.41
Vector-efficiency ratio store	30.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	50.00
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	16.96

Path /

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

Function	main
Source file and lines	attention_v2.cpp:52-53
Module	attention-clang-znver5-256

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:52-53.

It is main loop of related source loop which is unrolled by 32 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is partially vectorized. Only 26% of vector register length is used (average across all SSE/AVX instructions).

Details

84% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

57% of SSE/AVX instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Workaround

Read the "512-bits vectorization" report at "Potential" confidence level.

Execution units bottlenecks

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

512-bits vectorization

On some x86 processors supporting 512-bits vectorization, compilers are often too conservative and limit vectorization to 256 bits. Performance can then be improved by enforcing 512-bits vectorization, especially with many vectorized and high trip count loops. 512-bits vectorization performance overhead (compared to 256-bits) is generally lower on newer processors.

Workaround

Recompile with -mprefer-vector-width=512

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.

expf@plt: 32 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53

Vector unaligned load/store instructions

Detected 8 suboptimal vector unaligned load/store instructions.

Details

VEXTRACTF128: 4 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
VINSERTF128: 4 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53

Workaround

Pass to your compiler a micro-architecture specialization option:
- use march=native
Use vector aligned instructions:
1. align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
2. inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Type of elements and instruction set

8 AVX instructions are processing arithmetic or math operations on single precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 64 FP arithmetical operations:

64: addition or subtraction

The binary loop is loading 1352 bytes (338 single precision FP elements). The binary loop is storing 768 bytes (192 single precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.03 FP operations per loaded or stored byte.

General properties

nb instructions	188
nb uops	219
loop length	1354
used x86 registers	4
used mmx registers	0
used xmm registers	3
used ymm registers	6
used zmm registers	0
nb stack references	12

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	27.38 cycles
front end	27.38 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.33	0.33	0.33	0.33	0.33	0.33	27.25	27.25	27.25	27.25	14.00	14.00	14.00	14.00	20.00	20.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	27.25	27.25	27.25	27.25	14.00	14.00	14.00	14.00	20.00	20.00

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	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	27.38
Dispatch	27.25
Data deps.	1.00
Overall L1	27.38

Vectorization ratios

INT

all	100%
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)
other	100%

all	83%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

INT+FP

all	84%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	57%

Vector efficiency ratios

INT

all	25%
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)
other	25%

all	26%
load	29%
store	30%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	15%

INT+FP

all	26%
load	29%
store	30%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	16%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 27.38 cycles. At this rate:

19% of peak load performance is reached (49.39 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
21% of peak store performance is reached (28.05 out of 128.00 bytes stored 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 loop is: 6043

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VMOVAPS %YMM1,0x460(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPS -0x60(%RDX,%R13,4),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS 0x440(%RSP),%YMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS -0x40(%RDX,%R13,4),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS -0x20(%RDX,%R13,4),%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS (%RDX,%R13,4),%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM0,%YMM5	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM1,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM2,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM5,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVAPS %YMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM3,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM0,0x140(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM5,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0xe0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0x70(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x320(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0xa0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x320(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0xa0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0xe0(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x3e0(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0x120(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x3e0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0x120(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0xa0(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x400(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0x140(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x400(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0x140(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VMOVAPS 0x460(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
MOV 0x2e0(%RSP),%RDX	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
ADD $0x20,%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	0.17	N/A
VINSERTPS $0x30,%XMM0,%XMM2,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0x120(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
CMP %R13,%R14	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
JNE 6043 <main+0x3273>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.33-0.50	N/A

Function	main
Source file and lines	attention_v2.cpp:52-53
Module	attention-clang-znver5-256

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:52-53.

It is main loop of related source loop which is unrolled by 32 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is partially vectorized. Only 26% of vector register length is used (average across all SSE/AVX instructions).

Details

84% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

57% of SSE/AVX instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Workaround

Read the "512-bits vectorization" report at "Potential" confidence level.

Execution units bottlenecks

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

512-bits vectorization

Workaround

Recompile with -mprefer-vector-width=512

CALL instructions

Detected function call instructions.

Details

expf@plt: 32 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53

Vector unaligned load/store instructions

Detected 8 suboptimal vector unaligned load/store instructions.

Details

VEXTRACTF128: 4 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
VINSERTF128: 4 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53
- /home/eoseret/llm-attention/attention_v2.cpp:53

Workaround

Pass to your compiler a micro-architecture specialization option:
- use march=native
Use vector aligned instructions:
1. align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
2. inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Type of elements and instruction set

8 AVX instructions are processing arithmetic or math operations on single precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 64 FP arithmetical operations:

64: addition or subtraction

The binary loop is loading 1352 bytes (338 single precision FP elements). The binary loop is storing 768 bytes (192 single precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.03 FP operations per loaded or stored byte.

General properties

nb instructions	188
nb uops	219
loop length	1354
used x86 registers	4
used mmx registers	0
used xmm registers	3
used ymm registers	6
used zmm registers	0
nb stack references	12

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	27.38 cycles
front end	27.38 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.33	0.33	0.33	0.33	0.33	0.33	27.25	27.25	27.25	27.25	14.00	14.00	14.00	14.00	20.00	20.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	27.25	27.25	27.25	27.25	14.00	14.00	14.00	14.00	20.00	20.00

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	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	27.38
Dispatch	27.25
Data deps.	1.00
Overall L1	27.38

Vectorization ratios

INT

all	100%
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)
other	100%

all	83%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

INT+FP

all	84%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	57%

Vector efficiency ratios

INT

all	25%
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)
other	25%

all	26%
load	29%
store	30%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	15%

INT+FP

all	26%
load	29%
store	30%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	50%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	16%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 27.38 cycles. At this rate:

19% of peak load performance is reached (49.39 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
21% of peak store performance is reached (28.05 out of 128.00 bytes stored 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 loop is: 6043

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VMOVAPS %YMM1,0x460(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPS -0x60(%RDX,%R13,4),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS 0x440(%RSP),%YMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS -0x40(%RDX,%R13,4),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS -0x20(%RDX,%R13,4),%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPS (%RDX,%R13,4),%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM0,%YMM5	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM1,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM2,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM5,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVAPS %YMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VSUBPS %YMM4,%YMM3,%YMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS %YMM0,0x140(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM5,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x70(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0xe0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x1d0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x1d0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0x70(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x320(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0xa0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x320(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xe0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0xa0(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x70(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0x70(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0xe0(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x3e0(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0x120(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x3e0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0xa0(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0x120(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xe0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xe0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0xa0(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VMOVAPS 0x400(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVAPS 0x140(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VMOVAPS %YMM1,0x400(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (50.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	vect (25.0%)
VMOVAPS %XMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x120(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x30,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0x120(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVAPS 0x140(%RSP),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
VZEROUPPER	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VMOVSHDUP 0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	vect (12.5%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x10,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPD $0x1,0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VINSERTPS $0x20,%XMM0,%XMM1,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VMOVAPS %XMM0,0xa0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	0.50	vect (25.0%)
VPERMILPS $-0x1,0x140(%RSP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0.25	0	0	1-2	0.50	vect (25.0%)
CALL 1160 <expf@plt>	2	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	N/A
VMOVAPS 0xa0(%RSP),%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (25.0%)
VMOVAPS 0x460(%RSP),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (50.0%)
MOV 0x2e0(%RSP),%RDX	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
ADD $0x20,%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	0.17	N/A
VINSERTPS $0x30,%XMM0,%XMM2,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	2	0.25	scal (6.3%)
VINSERTF128 $0x1,0x120(%RSP),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	2	0.50	vect (25.0%)
VADDPS %YMM1,%YMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (50.0%)
CMP %R13,%R14	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
JNE 6043 <main+0x3273>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.33-0.50	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

512-bits vectorization

Workaround

CALL instructions

Details

Vector unaligned load/store instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

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

512-bits vectorization

Workaround

CALL instructions

Details

Vector unaligned load/store instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code