OV - exec - Loop 2547

0x4ad780 INC	%R11

0x4ad783 CMP	%RBX,%R11

0x4ad786 JGE	4ad7c0

0x4ad788 MOV	(%R12,%R11,8),%R15    [1]

0x4ad78c MOV	-0x60(%RBP),%RSI    [2]

0x4ad790 CMPQ	$0,(%RSI,%R15,8)    [6]

0x4ad795 JLE	4ad780

0x4ad797 MOV	0x78(%RBP),%RSI    [2]

0x4ad79b MOV	(%RSI,%R15,8),%R15    [5]

0x4ad79f CMP	%RAX,%R15

0x4ad7a2 JE	4ad780

0x4ad7a4 MOV	-0x40(%RBP),%RSI    [2]

0x4ad7a8 CMP	%RDX,(%RSI,%R15,8)    [4]

0x4ad7ac JGE	4ad780

0x4ad7ae MOV	%R8,(%RSI,%R15,8)    [4]

0x4ad7b2 INC	%R8

0x4ad7b5 MOV	%R8,-0x30(%RBP)    [2]

0x4ad7b9 MOV	0x8(%R13,%R10,8),%RBX    [3]

0x4ad7be JMP	4ad780

/home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c: 1743 - 1752

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

1743:                  for (jj2 = S_diag_i[i2]; jj2 < S_diag_i[i2+1]; jj2++)

1744:                  {

1745:                     i3 = S_diag_j[jj2];

1746:                     if (CF_marker[i3] > 0)

1747:                     {

1748:                        index = fine_to_coarse[i3];

1749:                        if (index != ic && S_marker[index] < jj_row_begin_diag)

1750:                        {

1751:                           S_marker[index] = num_nonzeros_diag;

1752:                           num_nonzeros_diag++;

Coverage (%)	Name	Source Location	Module
►100.00+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.19
Bottlenecks
Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source	par_strength.c:1743-1752
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	2.13
CQA cycles if no scalar integer	2.13
CQA cycles if FP arith vectorized	2.13
CQA cycles if fully vectorized	0.27
Front-end cycles	2.13
P0 cycles	1.63
P1 cycles	1.00
P2 cycles	1.92
P3 cycles	1.92
P4 cycles	0.25
P5 cycles	0.90
P6 cycles	1.63
P7 cycles	0.25
P8 cycles	0.25
P9 cycles	0.25
P10 cycles	0.85
P11 cycles	1.92
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	12.75
Nb uops	12.25
Nb loads	5.75
Nb stores	0.50
Nb stack references	2.50
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
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	23.14
Bytes prefetched	0.00
Bytes loaded	46.00
Bytes stored	4.00
Stride 0	1.25
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	2.25
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	12.50

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.17
Bottlenecks	micro-operation queue,
Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source	par_strength.c:1743-1752
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	1.17
CQA cycles if no scalar integer	1.17
CQA cycles if FP arith vectorized	1.17
CQA cycles if fully vectorized	0.15
Front-end cycles	1.17
P0 cycles	1.00
P1 cycles	0.80
P2 cycles	1.00
P3 cycles	1.00
P4 cycles	0.00
P5 cycles	0.60
P6 cycles	1.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	0.60
P11 cycles	1.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	7.00
Nb uops	7.00
Nb loads	3.00
Nb stores	0.00
Nb stack references	1.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
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	20.57
Bytes prefetched	0.00
Bytes loaded	24.00
Bytes stored	0.00
Stride 0	1.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	1.00
Vectorization ratio all	0.00
Vectorization ratio load	NA
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	NA
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	12.50

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.11
Bottlenecks	micro-operation queue,
Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source	par_strength.c:1743-1752
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	1.67
CQA cycles if no scalar integer	1.67
CQA cycles if FP arith vectorized	1.67
CQA cycles if fully vectorized	0.21
Front-end cycles	1.67
P0 cycles	1.50
P1 cycles	0.80
P2 cycles	1.67
P3 cycles	1.67
P4 cycles	0.00
P5 cycles	0.60
P6 cycles	1.50
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	0.60
P11 cycles	1.67
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	11.00
Nb uops	10.00
Nb loads	5.00
Nb stores	0.00
Nb stack references	2.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
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	24.00
Bytes prefetched	0.00
Bytes loaded	40.00
Bytes stored	0.00
Stride 0	1.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	2.00
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	12.50

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.25
Bottlenecks	micro-operation queue,
Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source	par_strength.c:1743-1752
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	3.33
CQA cycles if no scalar integer	3.33
CQA cycles if FP arith vectorized	3.33
CQA cycles if fully vectorized	0.42
Front-end cycles	3.33
P0 cycles	2.00
P1 cycles	1.40
P2 cycles	2.67
P3 cycles	2.67
P4 cycles	1.00
P5 cycles	1.40
P6 cycles	2.00
P7 cycles	1.00
P8 cycles	1.00
P9 cycles	1.00
P10 cycles	1.20
P11 cycles	2.67
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	19.00
Nb uops	19.00
Nb loads	8.00
Nb stores	2.00
Nb stack references	4.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
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	24.00
Bytes prefetched	0.00
Bytes loaded	64.00
Bytes stored	16.00
Stride 0	2.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	3.00
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	12.50

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.17
Bottlenecks	micro-operation queue, P2, P3, P11,
Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source	par_strength.c:1743-1752
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	2.33
CQA cycles if no scalar integer	2.33
CQA cycles if FP arith vectorized	2.33
CQA cycles if fully vectorized	0.29
Front-end cycles	2.33
P0 cycles	2.00
P1 cycles	1.00
P2 cycles	2.33
P3 cycles	2.33
P4 cycles	0.00
P5 cycles	1.00
P6 cycles	2.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	1.00
P11 cycles	2.33
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	14.00
Nb uops	13.00
Nb loads	7.00
Nb stores	0.00
Nb stack references	3.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
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	24.00
Bytes prefetched	0.00
Bytes loaded	56.00
Bytes stored	0.00
Stride 0	1.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	3.00
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	12.50

Path /

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

Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source file and lines	par_strength.c:1743-1752
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c:1743-1752.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 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

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 2.13 to 0.27 cycles (8.00x speedup).

Details

All SSE/AVX 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:
- 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)

No data for this section

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Irregular (variable stride) or indirect: 2 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 46 bytes. The binary loop is storing 4 bytes.

Unroll opportunity

Loop is data access bound.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor and if some data references are common to consecutive iterations. This can be done manually. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma unroll_and_jam N, unroll_and_jam(N), unroll N or unroll(N)

General properties

nb instructions	12.75
nb uops	12.25
loop length	42.25
used x86 registers	8
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	2.50

Front-end

MACRO FUSION NOT POSSIBLE

micro-operation queue	2.13 cycles
front end	2.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	1.63	1.00	1.92	1.92	0.25	0.90	1.63	0.25	0.25	0.25	0.85	1.92
cycles	1.63	1.00	1.92	1.92	0.25	0.90	1.63	0.25	0.25	0.25	0.85	1.92

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	2.13
Dispatch	1.92
Data deps.	1.00
Overall L1	2.13

Vectorization ratios

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

Vector efficiency ratios

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 2.13 cycles. At this rate:

11% of peak load performance is reached (21.94 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
0% of peak store performance is reached (1.20 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source file and lines	par_strength.c:1743-1752
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c:1743-1752.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 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

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 1.17 to 0.15 cycles (8.00x speedup).

Details

All SSE/AVX 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:
- 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

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Irregular (variable stride) or indirect: 1 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 24 bytes.

Unroll opportunity

Loop body is too small to efficiently use resources.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor. This can be done manually. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma unroll_and_jam N, unroll_and_jam(N), unroll N or unroll(N)

General properties

nb instructions	7
nb uops	7
loop length	23
used x86 registers	6
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	1

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	1.17 cycles
front end	1.17 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	1.00	0.80	1.00	1.00	0.00	0.60	1.00	0.00	0.00	0.00	0.60	1.00
cycles	1.00	0.80	1.00	1.00	0.00	0.60	1.00	0.00	0.00	0.00	0.60	1.00

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	1.17
Dispatch	1.00
Data deps.	1.00
Overall L1	1.17

Vectorization ratios

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

Vector efficiency ratios

all	12%
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)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	12%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.17 cycles. At this rate:

10% of peak load performance is reached (20.57 out of 192.00 bytes loaded 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 1.17 to 1.00 cycles (1.17x speedup).

ASM code

In the binary file, the address of the loop is: 4ad783

Instruction	Nb FU	P0	P1	P2	P3	P5	P6	P10	P11	Latency	Recip. throughput	Vectorization
INC %R11	1	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%R11	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	scal (12.5%)
JGE 4ad7c0 <hypre_BoomerAMGCreate2ndS.extracted.17+0x300>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV (%R12,%R11,8),%R15	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
MOV -0x60(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMPQ $0,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0.20	0.20	0.20	0.33	1	0.33	N/A
JLE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A

Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source file and lines	par_strength.c:1743-1752
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c:1743-1752.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 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

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 1.67 to 0.21 cycles (8.00x speedup).

Details

All SSE/AVX 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:
- 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

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Irregular (variable stride) or indirect: 2 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 40 bytes.

Unroll opportunity

Loop is data access bound.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor and if some data references are common to consecutive iterations. This can be done manually. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma unroll_and_jam N, unroll_and_jam(N), unroll N or unroll(N)

General properties

nb instructions	11
nb uops	10
loop length	36
used x86 registers	7
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	2

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	1.67 cycles
front end	1.67 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	1.50	0.80	1.67	1.67	0.00	0.60	1.50	0.00	0.00	0.00	0.60	1.67
cycles	1.50	0.80	1.67	1.67	0.00	0.60	1.50	0.00	0.00	0.00	0.60	1.67

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	1.67
Dispatch	1.67
Data deps.	1.00
Overall L1	1.67

Vectorization ratios

all	0%
load	0%
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)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Vector efficiency ratios

all	12%
load	12%
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)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	12%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.67 cycles. At this rate:

12% of peak load performance is reached (24.00 out of 192.00 bytes loaded 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 1.67 to 1.50 cycles (1.11x speedup).

ASM code

In the binary file, the address of the loop is: 4ad783

Instruction	Nb FU	P0	P1	P2	P3	P5	P6	P10	P11	Latency	Recip. throughput	Vectorization
INC %R11	1	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%R11	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	scal (12.5%)
JGE 4ad7c0 <hypre_BoomerAMGCreate2ndS.extracted.17+0x300>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV (%R12,%R11,8),%R15	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
MOV -0x60(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMPQ $0,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0.20	0.20	0.20	0.33	1	0.33	scal (12.5%)
JLE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV 0x78(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
MOV (%RSI,%R15,8),%R15	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMP %RAX,%R15	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	N/A
JE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A

Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source file and lines	par_strength.c:1743-1752
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c:1743-1752.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 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

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 3.33 to 0.42 cycles (8.00x speedup).

Details

All SSE/AVX 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:
- 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

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Irregular (variable stride) or indirect: 3 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 64 bytes. The binary loop is storing 16 bytes.

Unroll opportunity

Loop is data access bound.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor and if some data references are common to consecutive iterations. This can be done manually. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma unroll_and_jam N, unroll_and_jam(N), unroll N or unroll(N)

General properties

nb instructions	19
nb uops	19
loop length	64
used x86 registers	11
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	4

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	3.33 cycles
front end	3.33 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	2.00	1.40	2.67	2.67	1.00	1.40	2.00	1.00	1.00	1.00	1.20	2.67
cycles	2.00	1.40	2.67	2.67	1.00	1.40	2.00	1.00	1.00	1.00	1.20	2.67

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	3.33
Dispatch	2.67
Data deps.	1.00
Overall L1	3.33

Vectorization ratios

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

Vector efficiency ratios

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

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 3.33 cycles. At this rate:

10% of peak load performance is reached (19.20 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
3% of peak store performance is reached (4.80 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 3.33 to 2.67 cycles (1.25x speedup).

ASM code

In the binary file, the address of the loop is: 4ad783

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput	Vectorization
INC %R11	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%R11	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	scal (12.5%)
JGE 4ad7c0 <hypre_BoomerAMGCreate2ndS.extracted.17+0x300>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A
MOV (%R12,%R11,8),%R15	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
MOV -0x60(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
CMPQ $0,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0	0.20	0.20	0	0	0	0.20	0.33	1	0.33	scal (12.5%)
JLE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A
MOV 0x78(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
MOV (%RSI,%R15,8),%R15	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
CMP %RAX,%R15	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	scal (12.5%)
JE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A
MOV -0x40(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
CMP %RDX,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0	0.20	0.20	0	0	0	0.20	0.33	1	0.33	scal (12.5%)
JGE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A
MOV %R8,(%RSI,%R15,8)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	1	0.50	scal (12.5%)
INC %R8	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
MOV %R8,-0x30(%RBP)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	1	0.50	scal (12.5%)
MOV 0x8(%R13,%R10,8),%RBX	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	scal (12.5%)
JMP 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0	0	0	0	0	0	0	0	0	0	0	0	0	5.84	N/A

Function	hypre_BoomerAMGCreate2ndS.extracted.17
Source file and lines	par_strength.c:1743-1752
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-927-0874/intel/AMG/build/AMG/AMG/parcsr_ls/par_strength.c:1743-1752.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 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

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 2.33 to 0.29 cycles (8.00x speedup).

Details

All SSE/AVX 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:
- 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

Performance is limited by reading data from caches/RAM (load units are a bottleneck). By removing all these bottlenecks, you can lower the cost of an iteration from 2.33 to 2.00 cycles (1.17x speedup).

Workaround

Read less array elements
Provide more information to your compiler:
- hardcode the bounds of the corresponding 'for' loop
- use the 'restrict' C99 keyword

No data for this section

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Irregular (variable stride) or indirect: 3 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 56 bytes.

Unroll opportunity

Loop is data access bound.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor and if some data references are common to consecutive iterations. This can be done manually. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma unroll_and_jam N, unroll_and_jam(N), unroll N or unroll(N)

General properties

nb instructions	14
nb uops	13
loop length	46
used x86 registers	8
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	0
nb stack references	3

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	2.33 cycles
front end	2.33 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	2.00	1.00	2.33	2.33	0.00	1.00	2.00	0.00	0.00	0.00	1.00	2.33
cycles	2.00	1.00	2.33	2.33	0.00	1.00	2.00	0.00	0.00	0.00	1.00	2.33

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	2.33
Dispatch	2.33
Data deps.	1.00
Overall L1	2.33

Vectorization ratios

all	0%
load	0%
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)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Vector efficiency ratios

all	12%
load	12%
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)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	12%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 2.33 cycles. At this rate:

12% of peak load performance is reached (24.00 out of 192.00 bytes loaded 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 2.33 to 2.00 cycles (1.17x speedup).

ASM code

In the binary file, the address of the loop is: 4ad783

Instruction	Nb FU	P0	P1	P2	P3	P5	P6	P10	P11	Latency	Recip. throughput	Vectorization
INC %R11	1	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%R11	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	scal (12.5%)
JGE 4ad7c0 <hypre_BoomerAMGCreate2ndS.extracted.17+0x300>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV (%R12,%R11,8),%R15	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
MOV -0x60(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMPQ $0,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0.20	0.20	0.20	0.33	1	0.33	scal (12.5%)
JLE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV 0x78(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
MOV (%RSI,%R15,8),%R15	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMP %RAX,%R15	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	scal (12.5%)
JE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A
MOV -0x40(%RBP),%RSI	1	0	0	0.33	0.33	0	0	0	0.33	1	0.33	N/A
CMP %RDX,(%RSI,%R15,8)	1	0.20	0.20	0.33	0.33	0.20	0.20	0.20	0.33	1	0.33	N/A
JGE 4ad780 <hypre_BoomerAMGCreate2ndS.extracted.17+0x2c0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A

Report Configuration

Vectorization

Details

Workaround

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Unroll opportunity

Workaround

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Vectorization

Details

Workaround

Execution units bottlenecks

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Unroll opportunity

Workaround

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

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Unroll opportunity

Workaround

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

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Unroll opportunity

Workaround

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