OV - exec - Loop 2133

	Loop Id: 2133	Module: exec	Source: ams.c:3659-3659	Coverage: 2.42%

0x5274e1 VMOVUPD	(%R12,%R11,1),%YMM7    [1]

0x5274e7 VMOVUPD	%YMM7,(%RBX,%R11,1)    [2]

0x5274ed VMOVUPD	0x20(%R12,%R11,1),%YMM8    [1]

0x5274f4 VMOVUPD	%YMM8,0x20(%RBX,%R11,1)    [2]

0x5274fb VMOVUPD	0x40(%R12,%R11,1),%YMM9    [1]

0x527502 VMOVUPD	%YMM9,0x40(%RBX,%R11,1)    [2]

0x527509 VMOVUPD	0x60(%R12,%R11,1),%YMM10    [1]

0x527510 VMOVUPD	%YMM10,0x60(%RBX,%R11,1)    [2]

0x527517 VMOVUPD	0x80(%R12,%R11,1),%YMM11    [1]

0x527521 VMOVUPD	%YMM11,0x80(%RBX,%R11,1)    [2]

0x52752b VMOVUPD	0xa0(%R12,%R11,1),%YMM12    [1]

0x527535 VMOVUPD	%YMM12,0xa0(%RBX,%R11,1)    [2]

0x52753f VMOVUPD	0xc0(%R12,%R11,1),%YMM13    [1]

0x527549 VMOVUPD	%YMM13,0xc0(%RBX,%R11,1)    [2]

0x527553 VMOVUPD	0xe0(%R12,%R11,1),%YMM14    [1]

0x52755d VMOVUPD	%YMM14,0xe0(%RBX,%R11,1)    [2]

0x527567 ADD	$0x100,%R11

0x52756e CMP	%RSI,%R11

0x527571 JNE	5274e1

/home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c: 3659 - 3659

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

3659:          Vtemp_data[i] = u_data[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
►3.00+	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0
○	hypre_ParCSRRelaxThreads	ams.c:3662	exec
○	hypre_BoomerAMGCycle	par_cycle.c:334	exec
○	hypre_BoomerAMGSolve	par_amg_solve.c:235	exec
○	hypre_PCGSolve	pcg.c:545	exec
○	main	amg.c:419	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	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.78
Bottlenecks	P6, P7, P8, P9, P14, P15,
Function	hypre_ParCSRRelaxThreads._omp_fn.0
Source	ams.c:3659-3659
Source loop unroll info	unrolled by 4
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	4
CQA cycles	4.00
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	4.00
CQA cycles if fully vectorized	2.00
Front-end cycles	2.25
P0 cycles	0.33
P1 cycles	0.33
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.33
P5 cycles	0.33
P6 cycles	4.00
P7 cycles	4.00
P8 cycles	4.00
P9 cycles	4.00
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	4.00
P15 cycles	4.00
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	18.00
Nb loads	8.00
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	128.00
Bytes prefetched	0.00
Bytes loaded	256.00
Bytes stored	256.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	50.00
Vector-efficiency ratio store	50.00
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	NA

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.78
Bottlenecks	P6, P7, P8, P9, P14, P15,
Function	hypre_ParCSRRelaxThreads._omp_fn.0
Source	ams.c:3659-3659
Source loop unroll info	unrolled by 4
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	4
CQA cycles	4.00
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	4.00
CQA cycles if fully vectorized	2.00
Front-end cycles	2.25
P0 cycles	0.33
P1 cycles	0.33
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.33
P5 cycles	0.33
P6 cycles	4.00
P7 cycles	4.00
P8 cycles	4.00
P9 cycles	4.00
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	4.00
P15 cycles	4.00
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	18.00
Nb loads	8.00
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	128.00
Bytes prefetched	0.00
Bytes loaded	256.00
Bytes stored	256.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	50.00
Vector-efficiency ratio store	50.00
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	NA

Path /

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

Function	hypre_ParCSRRelaxThreads._omp_fn.0
Source file and lines	ams.c:3659-3659
Module	exec

The loop is defined in /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/ams.c:3659.

It is main loop of related source loop which is unrolled by 4 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is vectorized, but using only 256 out of 512 bits (AVX/AVX2 instructions on AVX-512 processors).

Details

All SSE/AVX instructions are used in vector version (process two or more data elements in vector registers).

Workaround

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

Execution units bottlenecks

Performance is limited by:

reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

Workaround

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

512-bits vectorization

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

Workaround

Recompile with -mprefer-vector-width=512

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 256 bytes. The binary loop is storing 256 bytes.

General properties

nb instructions	19
nb uops	18
loop length	150
used x86 registers	4
used mmx registers	0
used xmm registers	0
used ymm registers	8
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	2.25 cycles
front end	2.25 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.33	0.33	0.33	0.33	0.33	0.33	4.00	4.00	4.00	4.00	0.00	0.00	0.00	0.00	4.00	4.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	4.00	4.00	4.00	4.00	0.00	0.00	0.00	0.00	4.00	4.00

Execution ports to units layout:

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

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

Cycles summary

Front-end	2.25
Dispatch	4.00
Data deps.	1.00
Overall L1	4.00

Vectorization ratios

all	100%
load	100%
store	100%
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	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	50%
load	50%
store	50%
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	NA (no other vectorizable/vectorized instructions)

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 4.00 cycles. At this rate:

25% of peak load performance is reached (64.00 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
50% of peak store performance is reached (64.00 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 5274e1

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P14	P15	Latency	Recip. throughput	Vectorization
VMOVUPD (%R12,%R11,1),%YMM7	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM7,(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x20(%R12,%R11,1),%YMM8	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM8,0x20(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x40(%R12,%R11,1),%YMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM9,0x40(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x60(%R12,%R11,1),%YMM10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM10,0x60(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x80(%R12,%R11,1),%YMM11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM11,0x80(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xa0(%R12,%R11,1),%YMM12	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM12,0xa0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xc0(%R12,%R11,1),%YMM13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM13,0xc0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xe0(%R12,%R11,1),%YMM14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM14,0xe0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
ADD $0x100,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
CMP %RSI,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
JNE 5274e1 <hypre_ParCSRRelaxThreads._omp_fn.0+0x141>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50-1	N/A

Function	hypre_ParCSRRelaxThreads._omp_fn.0
Source file and lines	ams.c:3659-3659
Module	exec

gain
potential
hint
expert

Vectorization

Your loop is vectorized, but using only 256 out of 512 bits (AVX/AVX2 instructions on AVX-512 processors).

Details

All SSE/AVX instructions are used in vector version (process two or more data elements in vector registers).

Workaround

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

Execution units bottlenecks

Performance is limited by:

reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

Workaround

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

512-bits vectorization

Workaround

Recompile with -mprefer-vector-width=512

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 256 bytes. The binary loop is storing 256 bytes.

General properties

nb instructions	19
nb uops	18
loop length	150
used x86 registers	4
used mmx registers	0
used xmm registers	0
used ymm registers	8
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	2.25 cycles
front end	2.25 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.33	0.33	0.33	0.33	0.33	0.33	4.00	4.00	4.00	4.00	0.00	0.00	0.00	0.00	4.00	4.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	4.00	4.00	4.00	4.00	0.00	0.00	0.00	0.00	4.00	4.00

Execution ports to units layout:

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

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

Cycles summary

Front-end	2.25
Dispatch	4.00
Data deps.	1.00
Overall L1	4.00

Vectorization ratios

all	100%
load	100%
store	100%
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	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	50%
load	50%
store	50%
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	NA (no other vectorizable/vectorized instructions)

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 4.00 cycles. At this rate:

25% of peak load performance is reached (64.00 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
50% of peak store performance is reached (64.00 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 5274e1

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P14	P15	Latency	Recip. throughput	Vectorization
VMOVUPD (%R12,%R11,1),%YMM7	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM7,(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x20(%R12,%R11,1),%YMM8	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM8,0x20(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x40(%R12,%R11,1),%YMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM9,0x40(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x60(%R12,%R11,1),%YMM10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM10,0x60(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0x80(%R12,%R11,1),%YMM11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM11,0x80(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xa0(%R12,%R11,1),%YMM12	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM12,0xa0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xc0(%R12,%R11,1),%YMM13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM13,0xc0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD 0xe0(%R12,%R11,1),%YMM14	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	3	0.50	vect (50.0%)
VMOVUPD %YMM14,0xe0(%RBX,%R11,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (50.0%)
ADD $0x100,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
CMP %RSI,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
JNE 5274e1 <hypre_ParCSRRelaxThreads._omp_fn.0+0x141>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50-1	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

512-bits vectorization

Workaround

Type of elements and instruction set

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

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

Workaround

512-bits vectorization

Workaround

Type of elements and instruction set

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code