OV - exec - Loop 3596

  (3595) 0x49f060 STR	D0, [X12, X20,LSL #3]

  (3595) 0x49f064 ORR	X20, XZR, X18

  (3595) 0x49f068 CMP	X18, X19

  (3595) 0x49f06c B.EQ	4a00dc

  (3595) 0x49f070 ADD	X18, X20, #1

  (3595) 0x49f074 ORR	X4, XZR, X14

  (3595) 0x49f078 MOVI	D0, #0

  (3595) 0x49f07c LDR	X14, [X8, X18,LSL #3]

  (3595) 0x49f080 SUBS	X1, X14, X4

  (3595) 0x49f084 B.LE	49f060

  (3595) 0x49f088 CMP	X1, #2

  (3595) 0x49f08c B.CS	49f0a0

  (3595) 0x49f090 ORR	X0, XZR, X4

  (3595) 0x49f094 B	49f0f4

0x49f0a0 AND	X2, X1, #8127

0x49f0a4 MOVI	D1, #0

0x49f0a8 ADD	X3, X13, X4,LSL #3

0x49f0ac ADD	X0, X4, X2

0x49f0b0 ADD	X4, X15, X4,LSL #3

0x49f0b4 ORR	X5, XZR, X2

0x49f0b8 HINT	#0

0x49f0bc HINT	#0

  (3594) 0x49f0c0 LDP	X6, X7, [X4, #1016]

  (3594) 0x49f0c4 LDP	D2, D3, [X3, #1016]

  (3594) 0x49f0c8 SUBS	X5, X5, #2

  (3594) 0x49f0cc ADD	X3, X3, #16

  (3594) 0x49f0d0 ADD	X4, X4, #16

  (3594) 0x49f0d4 LDR	D4, [X10, X6,LSL #3]

  (3594) 0x49f0d8 LDR	D5, [X10, X7,LSL #3]

  (3594) 0x49f0dc FMADD	D0, D4, D2, D0

  (3594) 0x49f0e0 FMADD	D1, D5, D3, D1

  (3594) 0x49f0e4 B.NE	49f0c0

0x49f0e8 FADD	D0, D1, D0

0x49f0ec CMP	X1, X2

0x49f0f0 B.EQ	49f060

  (3595) 0x49f0f4 SUB	W1, W14, W0

  (3595) 0x49f0f8 ORR	X2, XZR, X0

  (3595) 0x49f0fc ANDS	X1, X1, #4160

  (3595) 0x49f100 B.EQ	49f120

    (3598) 0x49f104 LDR	X3, [X11, X2,LSL #3]

    (3598) 0x49f108 LDR	D1, [X9, X2,LSL #3]

    (3598) 0x49f10c ADD	X2, X2, #1

    (3598) 0x49f110 SUBS	X1, X1, #1

    (3598) 0x49f114 LDR	D2, [X10, X3,LSL #3]

    (3598) 0x49f118 FMADD	D0, D2, D1, D0

    (3598) 0x49f11c B.NE	49f104

  (3595) 0x49f120 SUB	X0, X0, X14

  (3595) 0x49f124 CMN	X0, #4

  (3595) 0x49f128 B.HI	49f060

  (3595) 0x49f12c SUB	X0, X14, X2

  (3595) 0x49f130 ADD	X1, X16, X2,LSL #3

  (3595) 0x49f134 ADD	X2, X17, X2,LSL #3

  (3595) 0x49f138 HINT	#0

  (3595) 0x49f13c HINT	#0

    (3597) 0x49f140 LDP	X3, X4, [X1, #1008]

    (3597) 0x49f144 LDP	D2, D3, [X2, #1008]

    (3597) 0x49f148 SUBS	X0, X0, #4

    (3597) 0x49f14c LDR	D1, [X10, X3,LSL #3]

    (3597) 0x49f150 FMADD	D0, D1, D2, D0

    (3597) 0x49f154 LDR	D1, [X10, X4,LSL #3]

    (3597) 0x49f158 LDP	X3, X4, [X1], #32

    (3597) 0x49f15c LDR	D2, [X10, X3,LSL #3]

    (3597) 0x49f160 FMADD	D0, D1, D3, D0

    (3597) 0x49f164 LDP	D1, D3, [X2], #32

    (3597) 0x49f168 FMADD	D0, D2, D1, D0

    (3597) 0x49f16c LDR	D1, [X10, X4,LSL #3]

    (3597) 0x49f170 FMADD	D0, D1, D3, D0

    (3597) 0x49f174 B.NE	49f140

  (3595) 0x49f178 B	49f060

/home/hbollore/qaas/qaas-runs/174-161-6712/intel/AMG/build/AMG/AMG/seq_mv/csr_matvec.c: 256 - 263

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

256:             for (i = iBegin; i < iEnd; i++)

257:             {

258:                tempx = 0.0;

259:                for (jj = A_i[i]; jj < A_i[i+1]; jj++)

260:                {

261:                   tempx += A_data[jj] * x_data[A_j[jj]];

262:                }

263:                y_data[i] = tempx;

Coverage (%)	Name	Source Location	Module
○100.00	__kmp_invoke_microtask		libomp.so

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	1.00
CQA speedup if FP arith vectorized	1.29
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.04
Bottlenecks	micro-operation queue,
Function	hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18
Source	csr_matvec.c:259-259
Source loop unroll info	NA
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	1.13
CQA cycles if no scalar integer	1.13
CQA cycles if FP arith vectorized	0.88
CQA cycles if fully vectorized	0.56
Front-end cycles	1.13
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	1.08
P3 cycles	1.08
P4 cycles	1.00
P5 cycles	1.00
P6 cycles	0.92
P7 cycles	0.92
P8 cycles	0.50
P9 cycles	0.50
P10 cycles	0.50
P11 cycles	0.50
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
P15 cycles	0.00
P16 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	11.00
Nb uops	9.00
Nb loads	NA
Nb stores	0.00
Nb stack references	0.00
FLOP/cycle	0.89
Nb FLOP add-sub	1.00
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	0.00
Bytes prefetched	0.00
Bytes loaded	0.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	0.00
Vectorization ratio load	NA
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	0.00
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	NA
Vector-efficiency ratio store	NA
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	50.00

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.29
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.04
Bottlenecks	micro-operation queue,
Function	hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18
Source	csr_matvec.c:259-259
Source loop unroll info	NA
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	1.13
CQA cycles if no scalar integer	1.13
CQA cycles if FP arith vectorized	0.88
CQA cycles if fully vectorized	0.56
Front-end cycles	1.13
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	1.08
P3 cycles	1.08
P4 cycles	1.00
P5 cycles	1.00
P6 cycles	0.92
P7 cycles	0.92
P8 cycles	0.50
P9 cycles	0.50
P10 cycles	0.50
P11 cycles	0.50
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
P15 cycles	0.00
P16 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	11.00
Nb uops	9.00
Nb loads	NA
Nb stores	0.00
Nb stack references	0.00
FLOP/cycle	0.89
Nb FLOP add-sub	1.00
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	0.00
Bytes prefetched	0.00
Bytes loaded	0.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	0.00
Vectorization ratio load	NA
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	0.00
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	NA
Vector-efficiency ratio store	NA
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	50.00

Path /

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

Function	hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18
Source file and lines	csr_matvec.c:256-263
Module	exec

The loop is defined in /home/hbollore/qaas/qaas-runs/174-161-6712/intel/AMG/build/AMG/AMG/seq_mv/csr_matvec.c:256-263.

Analyzed code is defined in /home/hbollore/qaas/qaas-runs/174-161-6712/intel/AMG/build/AMG/AMG/seq_mv/csr_matvec.c:259.

Warnings:

Non-innermost loop: analyzing only self part (ignoring child loops).
Ignoring paths for analysis
Failed to get the number of paths
RecMII not computed since number of paths is unknown or > max_paths
Streams not analyzed since number of paths is unknown or > max_paths

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 2 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 1.13 to 0.56 cycles (2.00x speedup).

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one
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)

No data for this section

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

The binary loop is composed of 1 FP arithmetical operations:

1: addition or subtraction

The binary loop does not load or store any data.

General properties

nb instructions	11
loop length	44
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	1.08	1.08	1.00	1.00	0.92	0.92	0.50	0.50	0.50	0.50	0.00	0.00	0.00	0.00	0.00
cycles	0.50	0.50	1.08	1.08	1.00	1.00	0.92	0.92	0.50	0.50	0.50	0.50	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

NA

Cycles summary

Front-end	1.13
Overall L1	1.13

Vectorization ratios

INT

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

FP

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

INT+FP

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.13 cycles.

ASM code

In the binary file, the address of the loop is: 49f0a0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
AND X2, X1, #8127	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVI D1, #0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (50.0%)
ADD X3, X13, X4,LSL #3	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X0, X4, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X4, X15, X4,LSL #3	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ORR X5, XZR, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
HINT #0																					N/A
HINT #0																					N/A
FADD D0, D1, D0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (50.0%)
CMP X1, X2	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	N/A
B.EQ 49f060 <hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18+0x170>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Function	hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18
Source file and lines	csr_matvec.c:256-263
Module	exec

The loop is defined in /home/hbollore/qaas/qaas-runs/174-161-6712/intel/AMG/build/AMG/AMG/seq_mv/csr_matvec.c:256-263.

Analyzed code is defined in /home/hbollore/qaas/qaas-runs/174-161-6712/intel/AMG/build/AMG/AMG/seq_mv/csr_matvec.c:259.

Warnings:

Non-innermost loop: analyzing only self part (ignoring child loops).
Ignoring paths for analysis
Failed to get the number of paths
RecMII not computed since number of paths is unknown or > max_paths
Streams not analyzed since number of paths is unknown or > max_paths

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 2 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 1.13 to 0.56 cycles (2.00x speedup).

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one
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)

No data for this section

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

The binary loop is composed of 1 FP arithmetical operations:

1: addition or subtraction

The binary loop does not load or store any data.

General properties

nb instructions	11
loop length	44
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	1.08	1.08	1.00	1.00	0.92	0.92	0.50	0.50	0.50	0.50	0.00	0.00	0.00	0.00	0.00
cycles	0.50	0.50	1.08	1.08	1.00	1.00	0.92	0.92	0.50	0.50	0.50	0.50	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

NA

Cycles summary

Front-end	1.13
Overall L1	1.13

Vectorization ratios

INT

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

FP

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

INT+FP

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.13 cycles.

ASM code

In the binary file, the address of the loop is: 49f0a0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16	Latency	Recip. throughput	Vectorization
AND X2, X1, #8127	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOVI D1, #0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (50.0%)
ADD X3, X13, X4,LSL #3	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X0, X4, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ADD X4, X15, X4,LSL #3	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
ORR X5, XZR, X2	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	1	0.17	scal (50.0%)
HINT #0																					N/A
HINT #0																					N/A
FADD D0, D1, D0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	2	0.25	scal (50.0%)
CMP X1, X2	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	0	0	1	0.33	N/A
B.EQ 49f060 <hypre_CSRMatrixMatvecOutOfPlace.omp_outlined.18+0x170>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Report Configuration

Vectorization

Details

Workaround

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code

Vectorization

Details

Workaround

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code