OV - exec - Loop 2075

	Loop Id: 2075	Module: exec	Source: ams.c:3532-3534	Coverage: 0.07%

0x521c9b MOV	(%R9,%RDX,8),%RCX    [1]

0x521c9f VCOMISD	(%R12,%RCX,8),%XMM11    [2]

0x521ca5 JBE	521cb7

0x521ca7 VMOVSD	(%R11,%RDX,8),%XMM8    [3]

0x521cad VXORPD	%XMM3,%XMM8,%XMM9

0x521cb1 VMOVSD	%XMM9,(%R11,%RDX,8)    [3]

0x521cb7 INC	%RDX

0x521cba MOV	(%R9,%RDX,8),%RSI    [1]

0x521cbe VCOMISD	(%R12,%RSI,8),%XMM11    [2]

0x521cc4 JBE	521cd6

0x521cc6 VMOVSD	(%R11,%RDX,8),%XMM10    [3]

0x521ccc VXORPD	%XMM3,%XMM10,%XMM0

0x521cd0 VMOVSD	%XMM0,(%R11,%RDX,8)    [3]

0x521cd6 LEA	0x1(%RDX),%RAX

0x521cda MOV	(%R9,%RAX,8),%RDI    [1]

0x521cde VCOMISD	(%R12,%RDI,8),%XMM11    [2]

0x521ce4 JBE	521cf6

0x521ce6 VMOVSD	(%R11,%RAX,8),%XMM13    [3]

0x521cec VXORPD	%XMM3,%XMM13,%XMM6

0x521cf0 VMOVSD	%XMM6,(%R11,%RAX,8)    [3]

0x521cf6 LEA	0x2(%RDX),%RCX

0x521cfa MOV	(%R9,%RCX,8),%RSI    [1]

0x521cfe VCOMISD	(%R12,%RSI,8),%XMM11    [2]

0x521d04 JBE	521d16

0x521d06 VMOVSD	(%R11,%RCX,8),%XMM4    [3]

0x521d0c VXORPD	%XMM3,%XMM4,%XMM5

0x521d10 VMOVSD	%XMM5,(%R11,%RCX,8)    [3]

0x521d16 LEA	0x3(%RDX),%RAX

0x521d1a MOV	(%R9,%RAX,8),%RDI    [1]

0x521d1e VCOMISD	(%R12,%RDI,8),%XMM11    [2]

0x521d24 JBE	521d36

0x521d26 VMOVSD	(%R11,%RAX,8),%XMM2    [3]

0x521d2c VXORPD	%XMM3,%XMM2,%XMM15

0x521d30 VMOVSD	%XMM15,(%R11,%RAX,8)    [3]

0x521d36 LEA	0x4(%RDX),%RCX

0x521d3a MOV	(%R9,%RCX,8),%RSI    [1]

0x521d3e VCOMISD	(%R12,%RSI,8),%XMM11    [2]

0x521d44 JBE	521d56

0x521d46 VMOVSD	(%R11,%RCX,8),%XMM14    [3]

0x521d4c VXORPD	%XMM3,%XMM14,%XMM1

0x521d50 VMOVSD	%XMM1,(%R11,%RCX,8)    [3]

0x521d56 LEA	0x5(%RDX),%RAX

0x521d5a MOV	(%R9,%RAX,8),%RDI    [1]

0x521d5e VCOMISD	(%R12,%RDI,8),%XMM11    [2]

0x521d64 JBE	521d76

0x521d66 VMOVSD	(%R11,%RAX,8),%XMM7    [3]

0x521d6c VXORPD	%XMM3,%XMM7,%XMM8

0x521d70 VMOVSD	%XMM8,(%R11,%RAX,8)    [3]

0x521d76 LEA	0x6(%RDX),%RCX

0x521d7a MOV	(%R9,%RCX,8),%RSI    [1]

0x521d7e VCOMISD	(%R12,%RSI,8),%XMM11    [2]

0x521d84 JBE	521d96

0x521d86 VMOVSD	(%R11,%RCX,8),%XMM9    [3]

0x521d8c VXORPD	%XMM3,%XMM9,%XMM10

0x521d90 VMOVSD	%XMM10,(%R11,%RCX,8)    [3]

0x521d96 ADD	$0x7,%RDX

0x521d9a CMP	%RDX,%R13

0x521d9d JNE	521c9b

/home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c: 3532 - 3534

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

3532:       for (i = ns; i < ne; i++)

3533:          if (A_diag_data[A_diag_I[i]] < 0)

3534:             l1_norm[i] = -l1_norm[i];

Coverage (%)	Name	Source Location	Module
►96.82+	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►3.18+	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0
○	hypre_ParCSRComputeL1NormsThre[...]	ams.c:3546	exec
○	hypre_BoomerAMGSetup	par_amg_setup.c:1409	exec
○	hypre_PCGSetup	pcg.c:234	exec
○	main	amg.c:398	exec
○	__libc_start_call_main		libc.so.6
○	__libc_start_main		libc.so.6
○	_start	amg.c:253	exec

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	1.02
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	1.02
Bottlenecks	micro-operation queue,
Function	hypre_ParCSRComputeL1NormsThreads._omp_fn.0
Source	ams.c:3532-3534
Source loop unroll info	multi-versionned
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	8.13
CQA cycles if no scalar integer	8.00
CQA cycles if FP arith vectorized	8.13
CQA cycles if fully vectorized	1.02
Front-end cycles	8.13
P0 cycles	2.67
P1 cycles	2.67
P2 cycles	2.67
P3 cycles	3.00
P4 cycles	3.00
P5 cycles	3.00
P6 cycles	8.00
P7 cycles	8.00
P8 cycles	8.00
P9 cycles	8.00
P10 cycles	4.00
P11 cycles	4.00
P12 cycles	4.00
P13 cycles	4.00
P14 cycles	8.00
P15 cycles	8.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	58.00
Nb uops	65.00
Nb loads	24.00
Nb stores	8.00
Nb stack references	0.00
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	31.51
Bytes prefetched	0.00
Bytes loaded	192.00
Bytes stored	64.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	25.00
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	50.00
Vector-efficiency ratio all	15.63
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	18.75

Metric	Value
CQA speedup if no scalar integer	1.02
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	1.02
Bottlenecks	micro-operation queue,
Function	hypre_ParCSRComputeL1NormsThreads._omp_fn.0
Source	ams.c:3532-3534
Source loop unroll info	multi-versionned
Source loop unroll confidence level	NA
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	8.13
CQA cycles if no scalar integer	8.00
CQA cycles if FP arith vectorized	8.13
CQA cycles if fully vectorized	1.02
Front-end cycles	8.13
P0 cycles	2.67
P1 cycles	2.67
P2 cycles	2.67
P3 cycles	3.00
P4 cycles	3.00
P5 cycles	3.00
P6 cycles	8.00
P7 cycles	8.00
P8 cycles	8.00
P9 cycles	8.00
P10 cycles	4.00
P11 cycles	4.00
P12 cycles	4.00
P13 cycles	4.00
P14 cycles	8.00
P15 cycles	8.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	58.00
Nb uops	65.00
Nb loads	24.00
Nb stores	8.00
Nb stack references	0.00
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	31.51
Bytes prefetched	0.00
Bytes loaded	192.00
Bytes stored	64.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	25.00
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	50.00
Vector-efficiency ratio all	15.63
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	18.75

Path /

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

Function	hypre_ParCSRComputeL1NormsThreads._omp_fn.0
Source file and lines	ams.c:3532-3534
Module	exec

The loop is defined in /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3532-3534.

The related source loop is multi-versionned.
Warnings:

Ignoring paths for analysis
Too many paths. If you really need to analyze all of the 256 paths individually, rerun with max-paths=256
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 256 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 probably not vectorized. Only 15% of vector register length is used (average across all SSE/AVX instructions). By vectorizing your loop, you can lower the cost of an iteration from 8.13 to 1.02 cycles (8.00x speedup).

Details

Store and arithmetical 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
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

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

VCOMISD: 8 occurrences►
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533

Type of elements and instruction set

8 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in scalar mode (one at a time).

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 192 bytes (24 double precision FP elements). The binary loop is storing 64 bytes (8 double precision FP elements).

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 by recompiling with -funroll-loops and/or -floop-unroll-and-jam. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma GCC unroll N

General properties

nb instructions	58
nb uops	65
loop length	264
used x86 registers	9
used mmx registers	0
used xmm registers	15
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	8.13 cycles
front end	8.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	2.67	2.67	2.67	3.00	3.00	3.00	8.00	8.00	8.00	8.00	4.00	4.00	4.00	4.00	8.00	8.00
cycles	2.67	2.67	2.67	3.00	3.00	3.00	8.00	8.00	8.00	8.00	4.00	4.00	4.00	4.00	8.00	8.00

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	8.13
Dispatch	8.00
Overall L1	8.13

Vectorization ratios

all	25%
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	50%

Vector efficiency ratios

all	15%
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	18%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 8.13 cycles. At this rate:

9% of peak load performance is reached (23.63 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (7.88 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the loop is: 521c9b

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
MOV (%R9,%RDX,8),%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
VCOMISD (%R12,%RCX,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cb7 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x737>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RDX,8),%XMM8	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM8,%XMM9	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM9,(%R11,%RDX,8)	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 (12.5%)
INC %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
MOV (%R9,%RDX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cd6 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x756>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RDX,8),%XMM10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM10,%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	vect (25.0%)
VMOVSD %XMM0,(%R11,%RDX,8)	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 (12.5%)
LEA 0x1(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cf6 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x776>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM13,%XMM6	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM6,(%R11,%RAX,8)	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 (12.5%)
LEA 0x2(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d16 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x796>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM4,%XMM5	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM5,(%R11,%RCX,8)	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 (12.5%)
LEA 0x3(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d36 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7b6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM2,%XMM15	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM15,(%R11,%RAX,8)	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 (12.5%)
LEA 0x4(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d56 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7d6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM14,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM1,(%R11,%RCX,8)	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 (12.5%)
LEA 0x5(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d76 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7f6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM7	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM7,%XMM8	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM8,(%R11,%RAX,8)	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 (12.5%)
LEA 0x6(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d96 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x816>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM9,%XMM10	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM10,(%R11,%RCX,8)	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 (12.5%)
ADD $0x7,%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
CMP %RDX,%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
JNE 521c9b <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x71b>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A

Function	hypre_ParCSRComputeL1NormsThreads._omp_fn.0
Source file and lines	ams.c:3532-3534
Module	exec

The loop is defined in /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3532-3534.

The related source loop is multi-versionned.
Warnings:

Ignoring paths for analysis
Too many paths. If you really need to analyze all of the 256 paths individually, rerun with max-paths=256
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

Vectorization

Details

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)

Execution units bottlenecks

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

No data for this section

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

VCOMISD: 8 occurrences►
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3533

Type of elements and instruction set

8 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in scalar mode (one at a time).

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

Unroll opportunity

Loop is data access bound.

Workaround

General properties

nb instructions	58
nb uops	65
loop length	264
used x86 registers	9
used mmx registers	0
used xmm registers	15
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	8.13 cycles
front end	8.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	2.67	2.67	2.67	3.00	3.00	3.00	8.00	8.00	8.00	8.00	4.00	4.00	4.00	4.00	8.00	8.00
cycles	2.67	2.67	2.67	3.00	3.00	3.00	8.00	8.00	8.00	8.00	4.00	4.00	4.00	4.00	8.00	8.00

Execution ports to units layout:

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

Cycles executing div or sqrt instructions

Cycles summary

Front-end	8.13
Dispatch	8.00
Overall L1	8.13

Vectorization ratios

all	25%
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	50%

Vector efficiency ratios

all	15%
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	18%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 8.13 cycles. At this rate:

9% of peak load performance is reached (23.63 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (7.88 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the loop is: 521c9b

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
MOV (%R9,%RDX,8),%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
VCOMISD (%R12,%RCX,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cb7 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x737>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RDX,8),%XMM8	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM8,%XMM9	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM9,(%R11,%RDX,8)	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 (12.5%)
INC %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
MOV (%R9,%RDX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cd6 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x756>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RDX,8),%XMM10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM10,%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	vect (25.0%)
VMOVSD %XMM0,(%R11,%RDX,8)	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 (12.5%)
LEA 0x1(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521cf6 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x776>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM13,%XMM6	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM6,(%R11,%RAX,8)	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 (12.5%)
LEA 0x2(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d16 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x796>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM4,%XMM5	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM5,(%R11,%RCX,8)	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 (12.5%)
LEA 0x3(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d36 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7b6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM2,%XMM15	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM15,(%R11,%RAX,8)	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 (12.5%)
LEA 0x4(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d56 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7d6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM14,%XMM1	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM1,(%R11,%RCX,8)	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 (12.5%)
LEA 0x5(%RDX),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
MOV (%R9,%RAX,8),%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	N/A
VCOMISD (%R12,%RDI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d76 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x7f6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RAX,8),%XMM7	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM7,%XMM8	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM8,(%R11,%RAX,8)	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 (12.5%)
LEA 0x6(%RDX),%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	0.25	N/A
MOV (%R9,%RCX,8),%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
VCOMISD (%R12,%RSI,8),%XMM11	2	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	0.50	0.50	6	0.50	scal (12.5%)
JBE 521d96 <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x816>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%R11,%RCX,8),%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.50	scal (12.5%)
VXORPD %XMM3,%XMM9,%XMM10	1	0	0	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25	vect (25.0%)
VMOVSD %XMM10,(%R11,%RCX,8)	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 (12.5%)
ADD $0x7,%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
CMP %RDX,%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
JNE 521c9b <hypre_ParCSRComputeL1NormsThreads._omp_fn.0+0x71b>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

Complex instructions

Details

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

Complex instructions

Details

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