OV - attention-clang-znver5-512 - Loop 53 - attention-clang-znver5-512

	Loop Id: 53	Module: attention-clang-znver5-512	Source: attention_v2.cpp:26-33	Coverage: 1.20%

0x811e MOV	0x90(%RSP),%R11

0x8126 MOV	0xc0(%RSP),%R15

0x812e MOV	0x160(%RSP),%RCX

0x8136 MOV	0x2a0(%RSP),%RDX

0x813e MOV	0x38(%RSP),%RSI

0x8143 VCVTSD2SS	%XMM0,%XMM0,%XMM0

0x8147 MOV	%R11,%R13

0x814a ADD	%R15D,%ECX

0x814d INC	%R15

0x8150 VMOVSS	%XMM0,(%RDX,%RCX,4)

0x8155 CMP	0x30(%RSP),%R15

0x815a JE	80cc

0x8160 VXORPS	%XMM0,%XMM0,%XMM0

0x8164 MOV	%R15,0xc0(%RSP)

0x816c CMP	$0x4,%R13D

0x8170 JB	8189

0x8172 MOV	%R15D,%ECX

0x8175 ADD	0x448(%RSP),%ECX

0x817c SETB	%CL

0x817f OR	0x40(%RSP),%CL

0x8183 JE	83d6

0x8189 XOR	%ECX,%ECX

0x818b MOV	0x80(%RSP),%RDX

0x8193 SUB	%ECX,%EDX

0x8195 AND	$0x7,%EDX

0x8198 JE	81fc

0x819a MOV	0x38(%RSP),%R9

0x819f MOV	0x90(%RSP),%R11

0x81a7 MOV	0xa0(%RSP),%R14

0x81af MOV	%RCX,%R8

0x81b2 MOV	%R9,%RSI

0x81b5 IMUL	%RCX,%RSI

0x81b9 ADD	0xc0(%RSP),%RSI

0x81c1 NOPW	%CS:(%RAX,%RAX,1)

  (55) 0x81d0 LEA	(%RAX,%R8,1),%EDI

  (55) 0x81d4 INC	%R8

  (55) 0x81d7 VMOVSS	(%R12,%RDI,4),%XMM1

  (55) 0x81dd MOV	%ESI,%EDI

  (55) 0x81df VMOVSS	(%R14,%RDI,4),%XMM2

  (55) 0x81e5 ADD	%R9,%RSI

  (55) 0x81e8 DEC	%RDX

  (55) 0x81eb VCVTSS2SD	%XMM1,%XMM1,%XMM1

  (55) 0x81ef VCVTSS2SD	%XMM2,%XMM2,%XMM2

  (55) 0x81f3 VFMADD231SD	%XMM2,%XMM1,%XMM0

  (55) 0x81f8 JNE	81d0

0x81fa JMP	820f

0x81fc MOV	0x90(%RSP),%R11

0x8204 MOV	0xa0(%RSP),%R14

0x820c MOV	%RCX,%R8

0x820f MOV	0x370(%RSP),%R15

0x8217 SUB	%R11,%RCX

0x821a CMP	$-0x8,%RCX

0x821e JA	8126

0x8224 MOV	0x38(%RSP),%RDI

0x8229 LEA	0x7(%R8),%RCX

0x822d LEA	0x4(%R8),%RBX

0x8231 LEA	0x3(%R8),%R13

0x8235 LEA	0x1(%R8),%RDX

0x8239 IMUL	%RDI,%RCX

0x823d MOV	%RDI,%RSI

0x8240 IMUL	%RDI,%RBX

0x8244 IMUL	%RDI,%R13

0x8248 IMUL	%RDI,%RDX

0x824c IMUL	%R8,%RSI

0x8250 MOV	%RCX,0x2c0(%RSP)

0x8258 LEA	0x6(%R8),%RCX

0x825c IMUL	%RDI,%RCX

0x8260 MOV	%RCX,0x200(%RSP)

0x8268 LEA	0x5(%R8),%RCX

0x826c IMUL	%RDI,%RCX

0x8270 MOV	%RCX,0x180(%RSP)

0x8278 LEA	0x2(%R8),%RCX

0x827c IMUL	%RDI,%RCX

0x8280 MOV	0xc0(%RSP),%RDI

0x8288 NOPL	(%RAX,%RAX,1)

  (54) 0x8290 LEA	(%RAX,%R8,1),%R9D

  (54) 0x8294 LEA	0x6(%RAX,%R8,1),%R10D

  (54) 0x8299 VMOVSS	(%R12,%R9,4),%XMM1

  (54) 0x829f LEA	(%RSI,%RDI,1),%R9D

  (54) 0x82a3 VMOVSS	(%R14,%R9,4),%XMM2

  (54) 0x82a9 LEA	0x1(%RAX,%R8,1),%R9D

  (54) 0x82ae VMOVSS	(%R12,%R10,4),%XMM13

  (54) 0x82b4 VMOVSS	(%R12,%R9,4),%XMM3

  (54) 0x82ba LEA	(%RDX,%RDI,1),%R9D

  (54) 0x82be VMOVSS	(%R14,%R9,4),%XMM4

  (54) 0x82c4 LEA	0x2(%RAX,%R8,1),%R9D

  (54) 0x82c9 VMOVSS	(%R12,%R9,4),%XMM5

  (54) 0x82cf LEA	(%RCX,%RDI,1),%R9D

  (54) 0x82d3 VMOVSS	(%R14,%R9,4),%XMM6

  (54) 0x82d9 LEA	0x3(%RAX,%R8,1),%R9D

  (54) 0x82de VCVTSS2SD	%XMM1,%XMM1,%XMM1

  (54) 0x82e2 VCVTSS2SD	%XMM2,%XMM2,%XMM2

  (54) 0x82e6 VMOVSS	(%R12,%R9,4),%XMM7

  (54) 0x82ec LEA	(%R13,%RDI,1),%R9D

  (54) 0x82f1 VMOVSS	(%R14,%R9,4),%XMM8

  (54) 0x82f7 LEA	0x4(%RAX,%R8,1),%R9D

  (54) 0x82fc VFMADD213SD	%XMM0,%XMM1,%XMM2

  (54) 0x8301 VCVTSS2SD	%XMM3,%XMM3,%XMM0

  (54) 0x8305 VMOVSS	(%R12,%R9,4),%XMM9

  (54) 0x830b LEA	(%RBX,%RDI,1),%R9D

  (54) 0x830f VMOVSS	(%R14,%R9,4),%XMM10

  (54) 0x8315 LEA	0x5(%RAX,%R8,1),%R9D

  (54) 0x831a VCVTSS2SD	%XMM4,%XMM4,%XMM1

  (54) 0x831e VMOVSS	(%R12,%R9,4),%XMM11

  (54) 0x8324 MOV	0x180(%RSP),%R9

  (54) 0x832c VFMADD213SD	%XMM2,%XMM0,%XMM1

  (54) 0x8331 VCVTSS2SD	%XMM5,%XMM5,%XMM0

  (54) 0x8335 LEA	(%R9,%RDI,1),%R9D

  (54) 0x8339 VCVTSS2SD	%XMM6,%XMM6,%XMM2

  (54) 0x833d VMOVSS	(%R14,%R9,4),%XMM12

  (54) 0x8343 MOV	0x200(%RSP),%R9

  (54) 0x834b VFMADD213SD	%XMM1,%XMM0,%XMM2

  (54) 0x8350 VCVTSS2SD	%XMM7,%XMM7,%XMM0

  (54) 0x8354 ADD	%EDI,%R9D

  (54) 0x8357 VCVTSS2SD	%XMM8,%XMM8,%XMM1

  (54) 0x835c VMOVSS	(%R14,%R9,4),%XMM14

  (54) 0x8362 LEA	0x7(%RAX,%R8,1),%R9D

  (54) 0x8367 ADD	$0x8,%R8

  (54) 0x836b VMOVSS	(%R12,%R9,4),%XMM15

  (54) 0x8371 MOV	0x2c0(%RSP),%R9

  (54) 0x8379 VFMADD213SD	%XMM2,%XMM0,%XMM1

  (54) 0x837e VCVTSS2SD	%XMM9,%XMM9,%XMM0

  (54) 0x8383 VCVTSS2SD	%XMM10,%XMM10,%XMM2

  (54) 0x8388 ADD	%EDI,%R9D

  (54) 0x838b ADD	%R15,%RDI

  (54) 0x838e VMOVSS	(%R14,%R9,4),%XMM16

  (54) 0x8395 VFMADD213SD	%XMM1,%XMM0,%XMM2

  (54) 0x839a VCVTSS2SD	%XMM11,%XMM11,%XMM0

  (54) 0x839f VCVTSS2SD	%XMM12,%XMM12,%XMM1

  (54) 0x83a4 VFMADD213SD	%XMM2,%XMM0,%XMM1

  (54) 0x83a9 VCVTSS2SD	%XMM13,%XMM13,%XMM0

  (54) 0x83ae VCVTSS2SD	%XMM14,%XMM14,%XMM2

  (54) 0x83b3 VFMADD213SD	%XMM1,%XMM0,%XMM2

  (54) 0x83b8 VCVTSS2SD	%XMM15,%XMM15,%XMM1

  (54) 0x83bd VCVTSS2SD	%XMM16,%XMM16,%XMM0

  (54) 0x83c3 VFMADD213SD	%XMM2,%XMM1,%XMM0

  (54) 0x83c8 CMP	%R8,%R11

  (54) 0x83cb JNE	8290

0x83d1 JMP	8126

0x83d6 CMP	$0x20,%R13D

0x83da JAE	83eb

0x83dc MOV	0xa0(%RSP),%RDI

0x83e4 XOR	%EDX,%EDX

0x83e6 JMP	84ca

0x83eb MOV	0x4c8(%RSP),%RSI

0x83f3 MOV	0xa0(%RSP),%RDI

0x83fb VXORPS	%XMM0,%XMM0,%XMM0

0x83ff XOR	%ECX,%ECX

0x8401 VXORPS	%XMM1,%XMM1,%XMM1

0x8405 VXORPS	%XMM2,%XMM2,%XMM2

0x8409 VXORPS	%XMM3,%XMM3,%XMM3

  (56) 0x840d LEA	(%RAX,%RCX,1),%EDX

  (56) 0x8410 VCVTPS2PD	(%R12,%RDX,4),%ZMM4

  (56) 0x8417 VCVTPS2PD	0x20(%R12,%RDX,4),%ZMM5

  (56) 0x841f VCVTPS2PD	0x40(%R12,%RDX,4),%ZMM6

  (56) 0x8427 VCVTPS2PD	0x60(%R12,%RDX,4),%ZMM7

  (56) 0x842f LEA	(%R15,%RCX,1),%EDX

  (56) 0x8433 ADD	$0x20,%RCX

  (56) 0x8437 VCVTPS2PD	(%RDI,%RDX,4),%ZMM8

  (56) 0x843e VCVTPS2PD	0x20(%RDI,%RDX,4),%ZMM9

  (56) 0x8446 VFMADD231PD	%ZMM8,%ZMM4,%ZMM0

  (56) 0x844c VFMADD231PD	%ZMM9,%ZMM5,%ZMM1

  (56) 0x8452 VCVTPS2PD	0x40(%RDI,%RDX,4),%ZMM8

  (56) 0x845a VCVTPS2PD	0x60(%RDI,%RDX,4),%ZMM5

  (56) 0x8462 VFMADD231PD	%ZMM8,%ZMM6,%ZMM2

  (56) 0x8468 VFMADD231PD	%ZMM5,%ZMM7,%ZMM3

  (56) 0x846e CMP	%RCX,%RSI

  (56) 0x8471 JNE	840d

0x8473 VADDPD	%ZMM0,%ZMM1,%ZMM0

0x8479 VADDPD	%ZMM2,%ZMM3,%ZMM2

0x847f VADDPD	%ZMM0,%ZMM2,%ZMM0

0x8485 VEXTRACTF64X4	$0x1,%ZMM0,%YMM1

0x848c VADDPD	%ZMM1,%ZMM0,%ZMM0

0x8492 VEXTRACTF128	$0x1,%YMM0,%XMM1

0x8498 VADDPD	%XMM1,%XMM0,%XMM0

0x849c VSHUFPD	$0x1,%XMM0,%XMM0,%XMM1

0x84a1 VADDSD	%XMM1,%XMM0,%XMM0

0x84a5 CMP	%ESI,%R13D

0x84a8 JE	811e

0x84ae TESTB	$0x1c,0x80(%RSP)

0x84b6 MOV	0xa0(%RSP),%RDI

0x84be MOV	%RSI,%RDX

0x84c1 MOV	%RSI,%RCX

0x84c4 JE	818b

0x84ca MOV	0x4d8(%RSP),%RSI

0x84d2 MOV	0xc0(%RSP),%R8

0x84da VMOVQ	%XMM0,%XMM0

  (57) 0x84de LEA	(%RAX,%RDX,1),%ECX

  (57) 0x84e1 VCVTPS2PD	(%R12,%RCX,4),%YMM1

  (57) 0x84e7 LEA	(%R8,%RDX,1),%ECX

  (57) 0x84eb ADD	$0x4,%RDX

  (57) 0x84ef VCVTPS2PD	(%RDI,%RCX,4),%YMM2

  (57) 0x84f4 VFMADD231PD	%YMM2,%YMM1,%YMM0

  (57) 0x84f9 CMP	%RDX,%RSI

  (57) 0x84fc JNE	84de

0x84fe VEXTRACTF128	$0x1,%YMM0,%XMM1

0x8504 MOV	0x90(%RSP),%R11

0x850c MOV	%RSI,%RCX

0x850f VADDPD	%XMM1,%XMM0,%XMM0

0x8513 VSHUFPD	$0x1,%XMM0,%XMM0,%XMM1

0x8518 VADDSD	%XMM1,%XMM0,%XMM0

0x851c CMP	%ESI,%R11D

0x851f JE	8126

0x8525 JMP	818b

/home/eoseret/llm-attention/attention_v2.cpp: 26 - 33

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

26:    for (unsigned int i = 0; i < M; ++i) {

27:       for (unsigned int j = 0; j < N; ++j) {  //vectorized

28:          double sum = 0.0;

29: #pragma clang loop vectorize(enable)

30:          for (unsigned int k = 0; k < K; ++k) { //vectorized

31:             sum += (double)A[i * K + k] * (double)B[k * N + j];

32:          }

33:          C[i * N + j] = alpha * static_cast<float>(sum);

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	2.92
CQA speedup if FP arith vectorized	1.03
CQA speedup if fully vectorized	1.38
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.69
Bottlenecks	micro-operation queue,
Function	main
Source	attention_v2.cpp:26-27,attention_v2.cpp:30-30,attention_v2.cpp:33-33
Source loop unroll info	NA
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	13.13
CQA cycles if no scalar integer	4.50
CQA cycles if FP arith vectorized	12.72
CQA cycles if fully vectorized	9.50
Front-end cycles	13.13
P0 cycles	7.50
P1 cycles	7.50
P2 cycles	7.50
P3 cycles	7.50
P4 cycles	7.50
P5 cycles	7.50
P6 cycles	7.75
P7 cycles	7.75
P8 cycles	7.75
P9 cycles	7.75
P10 cycles	3.00
P11 cycles	3.00
P12 cycles	4.50
P13 cycles	4.50
P14 cycles	0.50
P15 cycles	0.50
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	105.00
Nb uops	105.00
Nb loads	26.00
Nb stores	5.00
Nb stack references	16.00
FLOP/cycle	2.90
Nb FLOP add-sub	38.00
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	17.22
Bytes prefetched	0.00
Bytes loaded	190.00
Bytes stored	36.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	35.56
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	0.00
Vectorization ratio add_sub	66.67
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	52.63
Vector-efficiency ratio all	23.37
Vector-efficiency ratio load	11.59
Vector-efficiency ratio store	11.25
Vector-efficiency ratio mul	12.50
Vector-efficiency ratio add_sub	54.17
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	18.83

Metric	Value
CQA speedup if no scalar integer	2.92
CQA speedup if FP arith vectorized	1.03
CQA speedup if fully vectorized	1.38
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.69
Bottlenecks	micro-operation queue,
Function	main
Source	attention_v2.cpp:26-27,attention_v2.cpp:30-30,attention_v2.cpp:33-33
Source loop unroll info	NA
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	13.13
CQA cycles if no scalar integer	4.50
CQA cycles if FP arith vectorized	12.72
CQA cycles if fully vectorized	9.50
Front-end cycles	13.13
P0 cycles	7.50
P1 cycles	7.50
P2 cycles	7.50
P3 cycles	7.50
P4 cycles	7.50
P5 cycles	7.50
P6 cycles	7.75
P7 cycles	7.75
P8 cycles	7.75
P9 cycles	7.75
P10 cycles	3.00
P11 cycles	3.00
P12 cycles	4.50
P13 cycles	4.50
P14 cycles	0.50
P15 cycles	0.50
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	105.00
Nb uops	105.00
Nb loads	26.00
Nb stores	5.00
Nb stack references	16.00
FLOP/cycle	2.90
Nb FLOP add-sub	38.00
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	17.22
Bytes prefetched	0.00
Bytes loaded	190.00
Bytes stored	36.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	35.56
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	0.00
Vectorization ratio add_sub	66.67
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	52.63
Vector-efficiency ratio all	23.37
Vector-efficiency ratio load	11.59
Vector-efficiency ratio store	11.25
Vector-efficiency ratio mul	12.50
Vector-efficiency ratio add_sub	54.17
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	18.83

Path /

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

Function	main
Source file and lines	attention_v2.cpp:26-33
Module	attention-clang-znver5-512

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:26-33.

Analyzed code is defined in /home/eoseret/llm-attention/attention_v2.cpp:26-27,30,33.

Warnings:

Non-innermost loop: analyzing only self part (ignoring child loops).
Ignoring paths for analysis
Too many paths. Rerun with max-paths=23
RecMII not computed since number of paths is unknown or > max_paths
Streams not analyzed since number of paths is unknown or > max_paths

Try to simplify control and/or increase the maximum number of paths per function/loop through the 'max-paths-nb' option.

This loop has 23 execution paths.

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

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

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

gain
potential
hint
expert

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 13.13 to 4.50 cycles (2.92x speedup).

Workaround

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

Vectorization

Your loop is poorly vectorized. Only 23% of vector register length is used (average across all SSE/AVX instructions). By fully vectorizing your loop, you can lower the cost of an iteration from 13.13 to 9.50 cycles (1.38x speedup).

Details

35% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

0% of SSE/AVX loads are used in vector version.
0% of SSE/AVX stores are used in vector version.
66% of SSE/AVX addition or subtraction instructions are used in vector version.
0% of SSE/AVX multiply instructions are used in vector version.
52% of SSE/AVX 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 loop can use their full power.

Workaround

Try another compiler or update/tune your current one:
- recompile with ffast-math (included in Ofast) to extend loop vectorization to FP reductions.
Remove inter-iterations dependences from your loop and make it unit-stride:
- If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly: C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)
- If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

Execution units bottlenecks

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

No data for this section

Vector unaligned load/store instructions

Detected 3 suboptimal vector unaligned load/store instructions.

Details

VEXTRACTF128: 2 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:30
- /home/eoseret/llm-attention/attention_v2.cpp:30
VEXTRACTF64X4: 1 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:30

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.

Conversion instructions

Detected expensive conversion instructions.

Details

VCVTSD2SS (FP64 to FP32, scalar): 1 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:33

Workaround

Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements. Use double instead of single precision only when/where needed by numerical stability and avoid mixing precision. In C/C++, FP constants are double precision by default and must be suffixed by 'f' to make them single precision.

Type of elements and instruction set

2 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in scalar mode (one at a time). 2 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in vector mode (two at a time). 4 AVX-512 instructions are processing arithmetic or math operations on double 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 38 FP arithmetical operations:

38: addition or subtraction

The binary loop is loading 190 bytes (23 double precision FP elements). The binary loop is storing 36 bytes (4 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.17 FP operations per loaded or stored byte.

General properties

nb instructions	105
nb uops	105
loop length	539
used x86 registers	13
used mmx registers	0
used xmm registers	4
used ymm registers	2
used zmm registers	4
nb stack references	16

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	13.13 cycles
front end	13.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	7.50	7.50	7.50	7.50	7.50	7.50	7.75	7.75	7.75	7.75	3.00	3.00	4.50	4.50	0.50	0.50
cycles	7.50	7.50	7.50	7.50	7.50	7.50	7.75	7.75	7.75	7.75	3.00	3.00	4.50	4.50	0.50	0.50

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	13.13
Dispatch	7.75
Overall L1	13.13

Vectorization ratios

INT

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

all	80%
load	NA (no load vectorizable/vectorized instructions)
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	75%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	90%

INT+FP

all	35%
load	0%
store	0%
mul	0%
add-sub	66%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	52%

Vector efficiency ratios

INT

all	11%
load	11%
store	12%
mul	12%
add-sub	12%
fma	NA (no fma vectorizable/vectorized instructions)
other	8%

all	38%
load	NA (no load vectorizable/vectorized instructions)
store	6%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	59%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	26%

INT+FP

all	23%
load	11%
store	11%
mul	12%
add-sub	54%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	18%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 13.13 cycles. At this rate:

5% of peak load performance is reached (14.48 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
2% of peak store performance is reached (2.74 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). By removing all these bottlenecks, you can lower the cost of an iteration from 13.13 to 7.75 cycles (1.69x speedup).

ASM code

In the binary file, the address of the loop is: 8160

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc0(%RSP),%R15	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV 0x160(%RSP),%RCX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0x2a0(%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
MOV 0x38(%RSP),%RSI	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
VCVTSD2SS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	3	0.50	scal (12.5%)
MOV %R11,%R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	scal (12.5%)
ADD %R15D,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
INC %R15	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (12.5%)
VMOVSS %XMM0,(%RDX,%RCX,4)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	1	0.50	scal (6.3%)
CMP 0x30(%RSP),%R15	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 80cc <main+0x471c>	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
VXORPS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
MOV %R15,0xc0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
CMP $0x4,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JB 8189 <main+0x47d9>	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
MOV %R15D,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
ADD 0x448(%RSP),%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
SETB %CL	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
OR 0x40(%RSP),%CL	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
JE 83d6 <main+0x4a26>	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
XOR %ECX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
MOV 0x80(%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
SUB %ECX,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
AND $0x7,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
JE 81fc <main+0x484c>	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
MOV 0x38(%RSP),%R9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RCX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV %R9,%RSI	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
IMUL %RCX,%RSI	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
ADD 0xc0(%RSP),%RSI	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
NOPW %CS:(%RAX,%RAX,1)	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JMP 820f <main+0x485f>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RCX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV 0x370(%RSP),%R15	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
SUB %R11,%RCX	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
CMP $-0x8,%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (12.5%)
JA 8126 <main+0x4776>	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
MOV 0x38(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
LEA 0x7(%R8),%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA 0x4(%R8),%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA 0x3(%R8),%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
LEA 0x1(%R8),%RDX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RDI,%RSI	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
IMUL %RDI,%RBX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (12.5%)
IMUL %RDI,%R13	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (12.5%)
IMUL %RDI,%RDX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
IMUL %R8,%RSI	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x2c0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x6(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x200(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x5(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x180(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x2(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV 0xc0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
NOPL (%RAX,%RAX,1)	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JMP 8126 <main+0x4776>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
CMP $0x20,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JAE 83eb <main+0x4a3b>	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
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
JMP 84ca <main+0x4b1a>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
MOV 0x4c8(%RSP),%RSI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
VXORPS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
XOR %ECX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
VXORPS %XMM1,%XMM1,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VXORPS %XMM2,%XMM2,%XMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VXORPS %XMM3,%XMM3,%XMM3	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VADDPD %ZMM0,%ZMM1,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VADDPD %ZMM2,%ZMM3,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VADDPD %ZMM0,%ZMM2,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VEXTRACTF64X4 $0x1,%ZMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.25	vect (50.0%)
VADDPD %ZMM1,%ZMM0,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM1	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%)
VADDPD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (25.0%)
VSHUFPD $0x1,%XMM0,%XMM0,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1-2	0.25	vect (25.0%)
VADDSD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	scal (12.5%)
CMP %ESI,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JE 811e <main+0x476e>	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
TESTB $0x1c,0x80(%RSP)	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	scal (1.6%)
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RSI,%RDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV %RSI,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JE 818b <main+0x47db>	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
MOV 0x4d8(%RSP),%RSI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc0(%RSP),%R8	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
VMOVQ %XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	scal (12.5%)
VEXTRACTF128 $0x1,%YMM0,%XMM1	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%)
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV %RSI,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
VADDPD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (25.0%)
VSHUFPD $0x1,%XMM0,%XMM0,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1-2	0.25	vect (25.0%)
VADDSD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	scal (12.5%)
CMP %ESI,%R11D	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
JE 8126 <main+0x4776>	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
JMP 818b <main+0x47db>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A

Function	main
Source file and lines	attention_v2.cpp:26-33
Module	attention-clang-znver5-512

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:26-33.

Analyzed code is defined in /home/eoseret/llm-attention/attention_v2.cpp:26-27,30,33.

Warnings:

Non-innermost loop: analyzing only self part (ignoring child loops).
Ignoring paths for analysis
Too many paths. Rerun with max-paths=23
RecMII not computed since number of paths is unknown or > max_paths
Streams not analyzed since number of paths is unknown or > max_paths

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

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

gain
potential
hint
expert

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 13.13 to 4.50 cycles (2.92x speedup).

Workaround

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

Vectorization

Details

35% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

0% of SSE/AVX loads are used in vector version.
0% of SSE/AVX stores are used in vector version.
66% of SSE/AVX addition or subtraction instructions are used in vector version.
0% of SSE/AVX multiply instructions are used in vector version.
52% of SSE/AVX 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 loop can use their full power.

Workaround

Try another compiler or update/tune your current one:
- recompile with ffast-math (included in Ofast) to extend loop vectorization to FP reductions.
Remove inter-iterations dependences from your loop and make it unit-stride:
- If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly: C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)
- If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

Execution units bottlenecks

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

No data for this section

Vector unaligned load/store instructions

Detected 3 suboptimal vector unaligned load/store instructions.

Details

VEXTRACTF128: 2 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:30
- /home/eoseret/llm-attention/attention_v2.cpp:30
VEXTRACTF64X4: 1 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:30

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.

Conversion instructions

Detected expensive conversion instructions.

Details

VCVTSD2SS (FP64 to FP32, scalar): 1 occurrences►
- /home/eoseret/llm-attention/attention_v2.cpp:33

Workaround

Type of elements and instruction set

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

The binary loop is composed of 38 FP arithmetical operations:

38: addition or subtraction

The binary loop is loading 190 bytes (23 double precision FP elements). The binary loop is storing 36 bytes (4 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.17 FP operations per loaded or stored byte.

General properties

nb instructions	105
nb uops	105
loop length	539
used x86 registers	13
used mmx registers	0
used xmm registers	4
used ymm registers	2
used zmm registers	4
nb stack references	16

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	13.13 cycles
front end	13.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	7.50	7.50	7.50	7.50	7.50	7.50	7.75	7.75	7.75	7.75	3.00	3.00	4.50	4.50	0.50	0.50
cycles	7.50	7.50	7.50	7.50	7.50	7.50	7.75	7.75	7.75	7.75	3.00	3.00	4.50	4.50	0.50	0.50

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	13.13
Dispatch	7.75
Overall L1	13.13

Vectorization ratios

INT

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

all	80%
load	NA (no load vectorizable/vectorized instructions)
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	75%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	90%

INT+FP

all	35%
load	0%
store	0%
mul	0%
add-sub	66%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	52%

Vector efficiency ratios

INT

all	11%
load	11%
store	12%
mul	12%
add-sub	12%
fma	NA (no fma vectorizable/vectorized instructions)
other	8%

all	38%
load	NA (no load vectorizable/vectorized instructions)
store	6%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	59%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	26%

INT+FP

all	23%
load	11%
store	11%
mul	12%
add-sub	54%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	18%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 13.13 cycles. At this rate:

5% of peak load performance is reached (14.48 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
2% of peak store performance is reached (2.74 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

ASM code

In the binary file, the address of the loop is: 8160

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc0(%RSP),%R15	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV 0x160(%RSP),%RCX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0x2a0(%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
MOV 0x38(%RSP),%RSI	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
VCVTSD2SS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	3	0.50	scal (12.5%)
MOV %R11,%R13	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	scal (12.5%)
ADD %R15D,%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
INC %R15	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (12.5%)
VMOVSS %XMM0,(%RDX,%RCX,4)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	1	0.50	scal (6.3%)
CMP 0x30(%RSP),%R15	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 80cc <main+0x471c>	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
VXORPS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
MOV %R15,0xc0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
CMP $0x4,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JB 8189 <main+0x47d9>	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
MOV %R15D,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
ADD 0x448(%RSP),%ECX	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
SETB %CL	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
OR 0x40(%RSP),%CL	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
JE 83d6 <main+0x4a26>	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
XOR %ECX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
MOV 0x80(%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
SUB %ECX,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
AND $0x7,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
JE 81fc <main+0x484c>	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
MOV 0x38(%RSP),%R9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RCX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV %R9,%RSI	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
IMUL %RCX,%RSI	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
ADD 0xc0(%RSP),%RSI	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	N/A
NOPW %CS:(%RAX,%RAX,1)	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JMP 820f <main+0x485f>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%R14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RCX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV 0x370(%RSP),%R15	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
SUB %R11,%RCX	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
CMP $-0x8,%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (12.5%)
JA 8126 <main+0x4776>	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
MOV 0x38(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
LEA 0x7(%R8),%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA 0x4(%R8),%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
LEA 0x3(%R8),%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
LEA 0x1(%R8),%RDX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RDI,%RSI	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
IMUL %RDI,%RBX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (12.5%)
IMUL %RDI,%R13	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	scal (12.5%)
IMUL %RDI,%RDX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
IMUL %R8,%RSI	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x2c0(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x6(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x200(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x5(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV %RCX,0x180(%RSP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	4	0.50	scal (12.5%)
LEA 0x2(%R8),%RCX	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
IMUL %RDI,%RCX	1	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0	0	3	0.33	N/A
MOV 0xc0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
NOPL (%RAX,%RAX,1)	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JMP 8126 <main+0x4776>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
CMP $0x20,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JAE 83eb <main+0x4a3b>	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
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
JMP 84ca <main+0x4b1a>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A
MOV 0x4c8(%RSP),%RSI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
VXORPS %XMM0,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
XOR %ECX,%ECX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
VXORPS %XMM1,%XMM1,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VXORPS %XMM2,%XMM2,%XMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VXORPS %XMM3,%XMM3,%XMM3	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
VADDPD %ZMM0,%ZMM1,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VADDPD %ZMM2,%ZMM3,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VADDPD %ZMM0,%ZMM2,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VEXTRACTF64X4 $0x1,%ZMM0,%YMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.25	vect (50.0%)
VADDPD %ZMM1,%ZMM0,%ZMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (100.0%)
VEXTRACTF128 $0x1,%YMM0,%XMM1	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%)
VADDPD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (25.0%)
VSHUFPD $0x1,%XMM0,%XMM0,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1-2	0.25	vect (25.0%)
VADDSD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	scal (12.5%)
CMP %ESI,%R13D	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	scal (6.3%)
JE 811e <main+0x476e>	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
TESTB $0x1c,0x80(%RSP)	1	0.17	0.17	0.17	0.17	0.17	0.17	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25	scal (1.6%)
MOV 0xa0(%RSP),%RDI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	scal (12.5%)
MOV %RSI,%RDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
MOV %RSI,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
JE 818b <main+0x47db>	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
MOV 0x4d8(%RSP),%RSI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV 0xc0(%RSP),%R8	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
VMOVQ %XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	scal (12.5%)
VEXTRACTF128 $0x1,%YMM0,%XMM1	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%)
MOV 0x90(%RSP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.25	N/A
MOV %RSI,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.13	N/A
VADDPD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	vect (25.0%)
VSHUFPD $0x1,%XMM0,%XMM0,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1-2	0.25	vect (25.0%)
VADDSD %XMM1,%XMM0,%XMM0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	2	0.50	scal (12.5%)
CMP %ESI,%R11D	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
JE 8126 <main+0x4776>	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
JMP 818b <main+0x47db>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	0	0.50	N/A

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

Vector unaligned load/store instructions

Details

Workaround

Conversion 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

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

Vector unaligned load/store instructions

Details

Workaround

Conversion 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