OV - exec - Function main._omp

	Function: main._omp_fn.7	Module: exec	Source: stream.c:354-356	Coverage (incl. loops): 27.61% \| (excl. loops): 0.00%

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c: 354 - 356

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

354: #pragma omp parallel for

355: 	for (j=0; j<STREAM_ARRAY_SIZE; j++)

356: 	    a[j] = b[j]+scalar*c[j];

0x401a90 PUSH	%RBP

0x401a91 MOV	$0x401060,%RAX

0x401a9b MOV	%RSP,%RBP

0x401a9e PUSH	%R12

0x401aa0 PUSH	%RBX

0x401aa1 MOV	%RDI,%RBX

0x401aa4 CALL	%RAX

0x401aa6 MOVSXD	%EAX,%R12

0x401aa9 MOV	$0x401030,%RAX

0x401ab3 CALL	%RAX

0x401ab5 MOVSXD	%EAX,%RCX

0x401ab8 MOV	$0x33450000,%EAX

0x401abd CQTO

0x401abf IDIV	%R12

0x401ac2 CMP	%RDX,%RCX

0x401ac5 JL	401c80

0x401acb IMUL	%RAX,%RCX

0x401acf ADD	%RCX,%RDX

0x401ad2 ADD	%RDX,%RAX

0x401ad5 CMP	%RAX,%RDX

0x401ad8 JGE	401c77

0x401ade MOV	%RAX,%R8

0x401ae1 MOV	$0x406180,%RDI

0x401aeb MOV	$0x19a686180,%RSI

0x401af5 MOV	$0x334906180,%R9

0x401aff SUB	%RDX,%R8

0x401b02 VMOVSD	(%RBX),%XMM0

0x401b06 AND	$0x7,%R8D

0x401b0a JE	401bcd

0x401b10 CMP	$0x1,%R8

0x401b14 JE	401bb0

0x401b1a CMP	$0x2,%R8

0x401b1e JE	401b9c

0x401b20 CMP	$0x3,%R8

0x401b24 JE	401b88

0x401b26 CMP	$0x4,%R8

0x401b2a JE	401b74

0x401b2c CMP	$0x5,%R8

0x401b30 JE	401b60

0x401b32 CMP	$0x6,%R8

0x401b36 JE	401b4c

0x401b38 VMOVSD	(%RDI,%RDX,8),%XMM1

0x401b3d VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM1

0x401b43 VMOVSD	%XMM1,(%R9,%RDX,8)

0x401b49 INC	%RDX

0x401b4c VMOVSD	(%RDI,%RDX,8),%XMM2

0x401b51 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM2

0x401b57 VMOVSD	%XMM2,(%R9,%RDX,8)

0x401b5d INC	%RDX

0x401b60 VMOVSD	(%RDI,%RDX,8),%XMM3

0x401b65 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM3

0x401b6b VMOVSD	%XMM3,(%R9,%RDX,8)

0x401b71 INC	%RDX

0x401b74 VMOVSD	(%RDI,%RDX,8),%XMM4

0x401b79 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM4

0x401b7f VMOVSD	%XMM4,(%R9,%RDX,8)

0x401b85 INC	%RDX

0x401b88 VMOVSD	(%RDI,%RDX,8),%XMM5

0x401b8d VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM5

0x401b93 VMOVSD	%XMM5,(%R9,%RDX,8)

0x401b99 INC	%RDX

0x401b9c VMOVSD	(%RDI,%RDX,8),%XMM6

0x401ba1 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM6

0x401ba7 VMOVSD	%XMM6,(%R9,%RDX,8)

0x401bad INC	%RDX

0x401bb0 VMOVSD	(%RDI,%RDX,8),%XMM7

0x401bb5 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM7

0x401bbb VMOVSD	%XMM7,(%R9,%RDX,8)

0x401bc1 INC	%RDX

0x401bc4 CMP	%RDX,%RAX

0x401bc7 JE	401c77

(3) 0x401bcd VMOVSD	(%RDI,%RDX,8),%XMM8

(3) 0x401bd2 VMOVSD	0x8(%RDI,%RDX,8),%XMM9

(3) 0x401bd8 VFMADD213SD	(%RSI,%RDX,8),%XMM0,%XMM8

(3) 0x401bde VFMADD213SD	0x8(%RSI,%RDX,8),%XMM0,%XMM9

(3) 0x401be5 VMOVSD	0x10(%RDI,%RDX,8),%XMM10

(3) 0x401beb VMOVSD	0x18(%RDI,%RDX,8),%XMM11

(3) 0x401bf1 VFMADD213SD	0x10(%RSI,%RDX,8),%XMM0,%XMM10

(3) 0x401bf8 VFMADD213SD	0x18(%RSI,%RDX,8),%XMM0,%XMM11

(3) 0x401bff VMOVSD	0x20(%RDI,%RDX,8),%XMM12

(3) 0x401c05 VMOVSD	0x28(%RDI,%RDX,8),%XMM13

(3) 0x401c0b VFMADD213SD	0x20(%RSI,%RDX,8),%XMM0,%XMM12

(3) 0x401c12 VFMADD213SD	0x28(%RSI,%RDX,8),%XMM0,%XMM13

(3) 0x401c19 VMOVSD	0x30(%RDI,%RDX,8),%XMM14

(3) 0x401c1f VMOVSD	0x38(%RDI,%RDX,8),%XMM15

(3) 0x401c25 VFMADD213SD	0x30(%RSI,%RDX,8),%XMM0,%XMM14

(3) 0x401c2c VFMADD213SD	0x38(%RSI,%RDX,8),%XMM0,%XMM15

(3) 0x401c33 VMOVSD	%XMM8,(%R9,%RDX,8)

(3) 0x401c39 VMOVSD	%XMM9,0x8(%R9,%RDX,8)

(3) 0x401c40 VMOVSD	%XMM10,0x10(%R9,%RDX,8)

(3) 0x401c47 VMOVSD	%XMM11,0x18(%R9,%RDX,8)

(3) 0x401c4e VMOVSD	%XMM12,0x20(%R9,%RDX,8)

(3) 0x401c55 VMOVSD	%XMM13,0x28(%R9,%RDX,8)

(3) 0x401c5c VMOVSD	%XMM14,0x30(%R9,%RDX,8)

(3) 0x401c63 VMOVSD	%XMM15,0x38(%R9,%RDX,8)

(3) 0x401c6a ADD	$0x8,%RDX

(3) 0x401c6e CMP	%RDX,%RAX

(3) 0x401c71 JNE	401bcd

0x401c77 POP	%RBX

0x401c78 POP	%R12

0x401c7a POP	%RBP

0x401c7b RET

0x401c7c NOPL	(%RAX)

0x401c80 INC	%RAX

0x401c83 XOR	%EDX,%EDX

0x401c85 JMP	401acb

0x401c8a NOPW	(%RAX,%RAX,1)

Coverage (%)	Name	Source Location	Module
○98.95	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○1.05	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0

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 /

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

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run gcc_6

Source file and lines	stream.c:354-356
Module	exec

The function is defined in /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354-356.

Warnings:
Ignoring paths for analysis

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 13.33 to 7.33 cycles (1.82x speedup).

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Vectorization

Your function is not vectorized. Only 12% of vector register length is used (average across all SSE/AVX instructions). By vectorizing your function, you can lower the cost of an iteration from 13.33 to 3.00 cycles (4.44x 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 function can use their full power.

Workaround

Try another compiler or update/tune your current one
Make array accesses unit-stride:
- If your function 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.

FMA

Detected 7 FMA (fused multiply-add) operations.

CALL instructions

Detected function call instructions.

Details

Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.

unknown: 2 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

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.

IDIV: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

Conversion instructions

Detected expensive conversion instructions.

Details

CQTO: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

Workaround

Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.

Type of elements and instruction set

7 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 function (in the source code) and the binary function

The binary function is composed of 14 FP arithmetical operations:

7: addition or subtraction (all inside FMA instructions)
7: multiply (all inside FMA instructions)

The binary function is loading 120 bytes (15 double precision FP elements). The binary function is storing 56 bytes (7 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.08 FP operations per loaded or stored byte.

General properties

nb instructions	80
nb uops	80
loop length	342
used x86 registers	11
used mmx registers	0
used xmm registers	8
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	13.33 cycles
front end	13.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	9.50	9.50	9.50	9.50	8.00	8.00	8.00	8.00	3.50	3.50	0.00	0.00	3.50	3.50
cycles	9.50	9.50	9.50	9.50	8.00	8.00	8.00	8.00	3.50	3.50	0.00	0.00	3.50	3.50

Execution ports to units layout:

ALU0/BRU0: ALU, BRU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1: BRU
AGU0 (256 bits): store address, load
AGU1 (256 bits): store address, load
AGU2 (256 bits): store address, load
FP0 (256 bits): VPU, DIV/SQRT
FP1 (256 bits): VPU, DIV/SQRT
FP2 (256 bits): VPU
FP3 (256 bits): VPU
FP4 (256 bits): FP store data
FP5 (256 bits): FP store data

Cycles executing div or sqrt instructions

7.00-12.00

Cycles summary

Front-end	13.33
Dispatch	9.50
DIV/SQRT	7.00-12.00
Overall L1	13.33

Vectorization ratios

INT

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

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

INT+FP

all	0%
load	0%
store	0%
mul	0%
add-sub	0%
fma	0%
div/sqrt	0%
other	0%

Vector efficiency ratios

INT

all	11%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	12%
add-sub	12%
fma	NA (no fma vectorizable/vectorized instructions)
other	11%

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

INT+FP

all	12%
load	12%
store	12%
mul	12%
add-sub	12%
fma	12%
div/sqrt	12%
other	11%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 13.33 cycles. At this rate:

9% of peak load performance is reached (9.00 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (4.20 out of 64.00 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 function is: 401a90

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV $0x401060,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
PUSH %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RDI,%RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
CALL %RAX	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0	0	0	0	0	0	2	N/A
MOVSXD %EAX,%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
MOV $0x401030,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
CALL %RAX	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0	0	0	0	0	0	2	N/A
MOVSXD %EAX,%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
MOV $0x33450000,%EAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CQTO																		scal (12.5%)
IDIV %R12	2	2	0	0	0	0	0	0	0	0	0	0	0	0	0	9-19	7-12	scal (12.5%)
CMP %RDX,%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JL 401c80 <main._omp_fn.7+0x1f0>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
IMUL %RAX,%RCX	1	0	1	0	0	0	0	0	0	0	0	0	0	0	0	3	1	scal (12.5%)
ADD %RCX,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD %RDX,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %RAX,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JGE 401c77 <main._omp_fn.7+0x1e7>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
MOV %RAX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV $0x406180,%RDI	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV $0x19a686180,%RSI	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV $0x334906180,%R9	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
SUB %RDX,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
VMOVSD (%RBX),%XMM0	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
AND $0x7,%R8D	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
JE 401bcd <main._omp_fn.7+0x13d>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x1,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401bb0 <main._omp_fn.7+0x120>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x2,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b9c <main._omp_fn.7+0x10c>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x3,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b88 <main._omp_fn.7+0xf8>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x4,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b74 <main._omp_fn.7+0xe4>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x5,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b60 <main._omp_fn.7+0xd0>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x6,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b4c <main._omp_fn.7+0xbc>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%RDI,%RDX,8),%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM1,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM2	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM2	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM2,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM3,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM4,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM5	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM5	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM5,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM6,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM7	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM7	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM7,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %RDX,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401c77 <main._omp_fn.7+0x1e7>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
POP %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
RET	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50	N/A
NOPL (%RAX)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
INC %RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.25	N/A
JMP 401acb <main._omp_fn.7+0x3b>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	0	1	N/A
NOPW (%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A

The code analyzed by CQA in that panel excludes loops and represents 0.00% of application time for run gcc_6

Source file and lines	stream.c:354-356
Module	exec

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 13.33 to 7.33 cycles (1.82x speedup).

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Vectorization

Details

Workaround

Try another compiler or update/tune your current one
Make array accesses unit-stride:
- If your function 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.

FMA

Detected 7 FMA (fused multiply-add) operations.

CALL instructions

Detected function call instructions.

Details

unknown: 2 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

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.

IDIV: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

Conversion instructions

Detected expensive conversion instructions.

Details

CQTO: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz16.benchmarkcenter.megware.com/177-374-0958/stream/build/stream/src/stream.c:354

Workaround

Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.

Type of elements and instruction set

7 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 function (in the source code) and the binary function

The binary function is composed of 14 FP arithmetical operations:

7: addition or subtraction (all inside FMA instructions)
7: multiply (all inside FMA instructions)

The binary function is loading 120 bytes (15 double precision FP elements). The binary function is storing 56 bytes (7 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.08 FP operations per loaded or stored byte.

General properties

nb instructions	80
nb uops	80
loop length	342
used x86 registers	11
used mmx registers	0
used xmm registers	8
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	13.33 cycles
front end	13.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	9.50	9.50	9.50	9.50	8.00	8.00	8.00	8.00	3.50	3.50	0.00	0.00	3.50	3.50
cycles	9.50	9.50	9.50	9.50	8.00	8.00	8.00	8.00	3.50	3.50	0.00	0.00	3.50	3.50

Execution ports to units layout:

ALU0/BRU0: ALU, BRU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1: BRU
AGU0 (256 bits): store address, load
AGU1 (256 bits): store address, load
AGU2 (256 bits): store address, load
FP0 (256 bits): VPU, DIV/SQRT
FP1 (256 bits): VPU, DIV/SQRT
FP2 (256 bits): VPU
FP3 (256 bits): VPU
FP4 (256 bits): FP store data
FP5 (256 bits): FP store data

Cycles executing div or sqrt instructions

7.00-12.00

Cycles summary

Front-end	13.33
Dispatch	9.50
DIV/SQRT	7.00-12.00
Overall L1	13.33

Vectorization ratios

INT

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

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

INT+FP

all	0%
load	0%
store	0%
mul	0%
add-sub	0%
fma	0%
div/sqrt	0%
other	0%

Vector efficiency ratios

INT

all	11%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	12%
add-sub	12%
fma	NA (no fma vectorizable/vectorized instructions)
other	11%

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

INT+FP

all	12%
load	12%
store	12%
mul	12%
add-sub	12%
fma	12%
div/sqrt	12%
other	11%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 13.33 cycles. At this rate:

9% of peak load performance is reached (9.00 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (4.20 out of 64.00 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 function is: 401a90

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV $0x401060,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
PUSH %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RDI,%RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
CALL %RAX	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0	0	0	0	0	0	2	N/A
MOVSXD %EAX,%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
MOV $0x401030,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
CALL %RAX	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0	0	0	0	0	0	2	N/A
MOVSXD %EAX,%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
MOV $0x33450000,%EAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CQTO																		scal (12.5%)
IDIV %R12	2	2	0	0	0	0	0	0	0	0	0	0	0	0	0	9-19	7-12	scal (12.5%)
CMP %RDX,%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JL 401c80 <main._omp_fn.7+0x1f0>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
IMUL %RAX,%RCX	1	0	1	0	0	0	0	0	0	0	0	0	0	0	0	3	1	scal (12.5%)
ADD %RCX,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
ADD %RDX,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %RAX,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JGE 401c77 <main._omp_fn.7+0x1e7>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
MOV %RAX,%R8	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV $0x406180,%RDI	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV $0x19a686180,%RSI	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
MOV $0x334906180,%R9	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
SUB %RDX,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
VMOVSD (%RBX),%XMM0	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
AND $0x7,%R8D	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (6.3%)
JE 401bcd <main._omp_fn.7+0x13d>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x1,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401bb0 <main._omp_fn.7+0x120>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x2,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b9c <main._omp_fn.7+0x10c>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x3,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b88 <main._omp_fn.7+0xf8>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x4,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b74 <main._omp_fn.7+0xe4>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x5,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b60 <main._omp_fn.7+0xd0>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
CMP $0x6,%R8	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401b4c <main._omp_fn.7+0xbc>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
VMOVSD (%RDI,%RDX,8),%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM1,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM2	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM2	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM2,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM3,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM4,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM5	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM5	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM5,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM6,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
VMOVSD (%RDI,%RDX,8),%XMM7	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50	scal (12.5%)
VFMADD213SD (%RSI,%RDX,8),%XMM0,%XMM7	1	0	0	0	0	0	0.33	0.33	0.33	0.50	0.50	0	0	0	0	4	0.50	scal (12.5%)
VMOVSD %XMM7,(%R9,%RDX,8)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1	scal (12.5%)
INC %RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
CMP %RDX,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	scal (12.5%)
JE 401c77 <main._omp_fn.7+0x1e7>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A
POP %RBX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R12	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %RBP	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
RET	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50	N/A
NOPL (%RAX)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
INC %RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25	N/A
XOR %EDX,%EDX	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.25	N/A
JMP 401acb <main._omp_fn.7+0x3b>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	0	1	N/A
NOPW (%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A

Name	Coverage (%)	Time (s)
▼main._omp_fn.7–	27.61	7.42
○Loop 3 - stream.c:356-356 - exec	27.61	7.24

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

FMA

CALL instructions

Details

Complex instructions

Details

Conversion instructions

Details

Workaround

Type of elements and instruction set

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

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

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

FMA

CALL instructions

Details

Complex instructions

Details

Conversion instructions

Details

Workaround

Type of elements and instruction set

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

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