OV - exec - Loop 8

0x203c10 VMOVUPD	-0x60(%R11,%R12,8),%YMM0    [2]

0x203c17 VMOVUPD	-0x40(%R11,%R12,8),%YMM1    [2]

0x203c1e VMOVUPD	-0x20(%R11,%R12,8),%YMM2    [2]

0x203c25 VMOVUPD	(%R11,%R12,8),%YMM3    [2]

0x203c2b VADDPD	-0x60(%R10,%R12,8),%YMM0,%YMM0    [1]

0x203c32 VADDPD	-0x40(%R10,%R12,8),%YMM1,%YMM1    [1]

0x203c39 VADDPD	-0x20(%R10,%R12,8),%YMM2,%YMM2    [1]

0x203c40 VADDPD	(%R10,%R12,8),%YMM3,%YMM3    [1]

0x203c46 VMOVUPD	%YMM0,-0x60(%R15,%R12,8)    [3]

0x203c4d VMOVUPD	%YMM1,-0x40(%R15,%R12,8)    [3]

0x203c54 VMOVUPD	%YMM2,-0x20(%R15,%R12,8)    [3]

0x203c5b VMOVUPD	%YMM3,(%R15,%R12,8)    [3]

0x203c61 ADD	$0x10,%R12

0x203c65 CMP	%R12,%R9

0x203c68 JNE	203c10

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-365-2841/stream/build/stream/src/stream.c: 344 - 344

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

344: 	    c[j] = a[j]+b[j];

Coverage (%)	Name	Source Location	Module
►99.81+	__kmp_invoke_microtask		libomp.so
○	__kmp_invoke_task_func		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	1.00
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.50
Bottlenecks	P6, P7, P8, P9,
Function	main.omp_outlined.31
Source	stream.c:344-344
Source loop unroll info	unrolled by 16
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	16
CQA cycles	3.00
CQA cycles if no scalar integer	3.00
CQA cycles if FP arith vectorized	3.00
CQA cycles if fully vectorized	1.50
Front-end cycles	1.75
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	3.00
P7 cycles	3.00
P8 cycles	3.00
P9 cycles	3.00
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	2.00
P13 cycles	2.00
P14 cycles	2.00
P15 cycles	2.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	15.00
Nb uops	14.00
Nb loads	8.00
Nb stores	4.00
Nb stack references	0.00
FLOP/cycle	5.33
Nb FLOP add-sub	16.00
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	128.00
Stride 0	0.00
Stride 1	3.00
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	100.00
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	50.00
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	1.50
Bottlenecks	P6, P7, P8, P9,
Function	main.omp_outlined.31
Source	stream.c:344-344
Source loop unroll info	unrolled by 16
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	16
CQA cycles	3.00
CQA cycles if no scalar integer	3.00
CQA cycles if FP arith vectorized	3.00
CQA cycles if fully vectorized	1.50
Front-end cycles	1.75
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	3.00
P7 cycles	3.00
P8 cycles	3.00
P9 cycles	3.00
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	2.00
P13 cycles	2.00
P14 cycles	2.00
P15 cycles	2.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	15.00
Nb uops	14.00
Nb loads	8.00
Nb stores	4.00
Nb stack references	0.00
FLOP/cycle	5.33
Nb FLOP add-sub	16.00
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	128.00
Stride 0	0.00
Stride 1	3.00
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	100.00
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	50.00
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	main.omp_outlined.31
Source file and lines	stream.c:344-344
Module	exec

The loop is defined in /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-365-2841/stream/build/stream/src/stream.c:344.

It is main loop of related source loop which is unrolled by 16 (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)

By removing all these bottlenecks, you can lower the cost of an iteration from 3.00 to 2.00 cycles (1.50x speedup).

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

4 AVX instructions are processing arithmetic or math operations on double precision FP elements in vector mode (four at a time).

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

The binary loop is composed of 16 FP arithmetical operations:

16: addition or subtraction

The binary loop is loading 256 bytes (32 double precision FP elements). The binary loop is storing 128 bytes (16 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.04 FP operations per loaded or stored byte.

General properties

nb instructions	15
nb uops	14
loop length	90
used x86 registers	5
used mmx registers	0
used xmm registers	0
used ymm registers	4
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	1.75 cycles
front end	1.75 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	3.00	3.00	3.00	3.00	0.00	0.00	2.00	2.00	2.00	2.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	3.00	3.00	3.00	3.00	0.00	0.00	2.00	2.00	2.00	2.00

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	1.75
Dispatch	3.00
Data deps.	1.00
Overall L1	3.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
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	50%
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 3.00 cycles. At this rate:

33% of peak load performance is reached (85.33 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
33% of peak store performance is reached (42.67 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: 203c10

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VMOVUPD -0x60(%R11,%R12,8),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD -0x40(%R11,%R12,8),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD -0x20(%R11,%R12,8),%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD (%R11,%R12,8),%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VADDPD -0x60(%R10,%R12,8),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD -0x40(%R10,%R12,8),%YMM1,%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD -0x20(%R10,%R12,8),%YMM2,%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD (%R10,%R12,8),%YMM3,%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVUPD %YMM0,-0x60(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM1,-0x40(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM2,-0x20(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM3,(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
ADD $0x10,%R12	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %R12,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
JNE 203c10 <main.omp_outlined.31+0x110>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0.33-1	N/A

Function	main.omp_outlined.31
Source file and lines	stream.c:344-344
Module	exec

The loop is defined in /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-365-2841/stream/build/stream/src/stream.c:344.

It is main loop of related source loop which is unrolled by 16 (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)

By removing all these bottlenecks, you can lower the cost of an iteration from 3.00 to 2.00 cycles (1.50x speedup).

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

4 AVX instructions are processing arithmetic or math operations on double precision FP elements in vector mode (four at a time).

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

The binary loop is composed of 16 FP arithmetical operations:

16: addition or subtraction

The binary loop is loading 256 bytes (32 double precision FP elements). The binary loop is storing 128 bytes (16 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.04 FP operations per loaded or stored byte.

General properties

nb instructions	15
nb uops	14
loop length	90
used x86 registers	5
used mmx registers	0
used xmm registers	0
used ymm registers	4
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	1.75 cycles
front end	1.75 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	3.00	3.00	3.00	3.00	0.00	0.00	2.00	2.00	2.00	2.00
cycles	0.33	0.33	0.33	0.33	0.33	0.33	3.00	3.00	3.00	3.00	0.00	0.00	2.00	2.00	2.00	2.00

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	1.75
Dispatch	3.00
Data deps.	1.00
Overall L1	3.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
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	50%
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 3.00 cycles. At this rate:

33% of peak load performance is reached (85.33 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
33% of peak store performance is reached (42.67 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: 203c10

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VMOVUPD -0x60(%R11,%R12,8),%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD -0x40(%R11,%R12,8),%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD -0x20(%R11,%R12,8),%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VMOVUPD (%R11,%R12,8),%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	3	0.50	vect (50.0%)
VADDPD -0x60(%R10,%R12,8),%YMM0,%YMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD -0x40(%R10,%R12,8),%YMM1,%YMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD -0x20(%R10,%R12,8),%YMM2,%YMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VADDPD (%R10,%R12,8),%YMM3,%YMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	2	0.50	vect (50.0%)
VMOVUPD %YMM0,-0x60(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM1,-0x40(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM2,-0x20(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
VMOVUPD %YMM3,(%R15,%R12,8)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	4	0.50	vect (50.0%)
ADD $0x10,%R12	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %R12,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
JNE 203c10 <main.omp_outlined.31+0x110>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0.33-1	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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code

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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code