OV - exec - Loop 956

	Loop Id: 956	Module: exec	Source: MultiBsplineRef.hpp:68-71	Coverage: 19.27%

0x47d160 VMULPD	(%R9,%RCX,1),%YMM4,%YMM1    [8]

0x47d166 VMULPD	(%R11,%RCX,1),%YMM2,%YMM23    [2]

0x47d16d VMULPD	(%RDI,%RCX,1),%YMM4,%YMM0    [4]

0x47d172 VMULPD	(%R10,%RCX,1),%YMM2,%YMM24    [1]

0x47d179 VMULPD	0x20(%R11,%RCX,1),%YMM2,%YMM25    [2]

0x47d181 VFMADD231PD	(%RDX,%RCX,1),%YMM5,%YMM1    [11]

0x47d187 VFMADD231PD	(%R8,%RCX,1),%YMM3,%YMM23    [7]

0x47d18e VFMADD231PD	(%RAX,%RCX,1),%YMM5,%YMM0    [3]

0x47d194 VFMADD231PD	(%RSI,%RCX,1),%YMM3,%YMM24    [10]

0x47d19b VFMADD231PD	0x20(%RCX,%R8,1),%YMM3,%YMM25    [6]

0x47d1a3 VMULPD	0x20(%R10,%RCX,1),%YMM2,%YMM26    [1]

0x47d1ab VADDPD	%YMM23,%YMM1,%YMM1

0x47d1b1 MOV	0xd8(%RSP),%RBX    [9]

0x47d1b9 VADDPD	%YMM24,%YMM0,%YMM0

0x47d1bf VFMADD213PD	(%R15,%RCX,1),%YMM7,%YMM1    [5]

0x47d1c5 VFMADD231PD	0x20(%RSI,%RCX,1),%YMM3,%YMM26    [10]

0x47d1cd VFMADD132PD	%YMM6,%YMM1,%YMM0

0x47d1d2 VMULPD	0x20(%R9,%RCX,1),%YMM4,%YMM1    [8]

0x47d1d9 VMOVUPD	%YMM0,(%R15,%RCX,1)    [5]

0x47d1df VFMADD231PD	0x20(%RCX,%RDX,1),%YMM5,%YMM1    [6]

0x47d1e6 VMULPD	0x20(%RCX,%RDI,1),%YMM4,%YMM0    [6]

0x47d1ec VADDPD	%YMM25,%YMM1,%YMM1

0x47d1f2 VFMADD231PD	0x20(%RCX,%RAX,1),%YMM5,%YMM0    [6]

0x47d1f9 VFMADD213PD	0x20(%R15,%RCX,1),%YMM7,%YMM1    [5]

0x47d200 VADDPD	%YMM26,%YMM0,%YMM0

0x47d206 VFMADD132PD	%YMM6,%YMM1,%YMM0

0x47d20b VMOVUPD	%YMM0,0x20(%R15,%RCX,1)    [5]

0x47d212 ADD	$0x40,%RCX

0x47d216 CMP	%RBX,%RCX

0x47d219 JNE	47d160

/home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp: 68 - 71

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

68:       for (size_t n = 0; n < num_splines; n++)

69:         vals[n] += pre00 *

70:             (c[0] * coefs[n] + c[1] * coefs[n + zs] + c[2] * coefs[n + 2 * zs] +

71:              c[3] * coefs[n + 3 * zs]);

Coverage (%)	Name	Source Location	Module
►49.95+	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:178	exec
○	qmcplusplus::WaveFunction::eva[...]	stl_vector.h:993	exec
○	main._omp_fn.1	NonLocalPP.hpp:126	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
►47.76+	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:178	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:266	exec
○	main._omp_fn.1	NonLocalPP.hpp:126	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
►1.18+	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:178	exec
○	qmcplusplus::WaveFunction::eva[...]	stl_vector.h:993	exec
○	main._omp_fn.1	NonLocalPP.hpp:126	exec
○	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0
►1.12+	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:178	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:266	exec
○	main._omp_fn.1	NonLocalPP.hpp:126	exec
○	GOMP_parallel	libgomp.h:982	libgomp.so.1.0.0

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.95
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.20
Bottlenecks	P1,
Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source	MultiBsplineRef.hpp:68-71
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	12.00
CQA cycles if no scalar integer	12.00
CQA cycles if FP arith vectorized	6.17
CQA cycles if fully vectorized	6.00
Front-end cycles	7.83
P0 cycles	10.00
P1 cycles	12.00
P2 cycles	6.33
P3 cycles	6.33
P4 cycles	1.00
P5 cycles	4.00
P6 cycles	1.00
P7 cycles	1.00
P8 cycles	1.00
P9 cycles	1.00
P10 cycles	0.00
P11 cycles	6.33
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	30.00
Nb uops	29.00
Nb loads	19.00
Nb stores	2.00
Nb stack references	1.00
FLOP/cycle	12.00
Nb FLOP add-sub	16.00
Nb FLOP mul	32.00
Nb FLOP fma	48.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	54.00
Bytes prefetched	0.00
Bytes loaded	584.00
Bytes stored	64.00
Stride 0	1.00
Stride 1	5.00
Stride n	5.00
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	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
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	50.00
Vector-efficiency ratio add_sub	50.00
Vector-efficiency ratio fma	50.00
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	1.95
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.20
Bottlenecks	P1,
Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source	MultiBsplineRef.hpp:68-71
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	12.00
CQA cycles if no scalar integer	12.00
CQA cycles if FP arith vectorized	6.17
CQA cycles if fully vectorized	6.00
Front-end cycles	7.83
P0 cycles	10.00
P1 cycles	12.00
P2 cycles	6.33
P3 cycles	6.33
P4 cycles	1.00
P5 cycles	4.00
P6 cycles	1.00
P7 cycles	1.00
P8 cycles	1.00
P9 cycles	1.00
P10 cycles	0.00
P11 cycles	6.33
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	30.00
Nb uops	29.00
Nb loads	19.00
Nb stores	2.00
Nb stack references	1.00
FLOP/cycle	12.00
Nb FLOP add-sub	16.00
Nb FLOP mul	32.00
Nb FLOP fma	48.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	54.00
Bytes prefetched	0.00
Bytes loaded	584.00
Bytes stored	64.00
Stride 0	1.00
Stride 1	5.00
Stride n	5.00
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	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
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	50.00
Vector-efficiency ratio add_sub	50.00
Vector-efficiency ratio fma	50.00
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	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source file and lines	MultiBsplineRef.hpp:68-71
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:68-71.

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

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 execution of FP multiply or FMA (fused multiply-add) operations (the FP multiply/FMA unit is a bottleneck). By removing all these bottlenecks, you can lower the cost of an iteration from 12.00 to 10.00 cycles (1.20x speedup).

Workaround

Reduce the number of FP multiply/FMA instructions

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

FMA

Detected 48 FMA (fused multiply-add) operations. Presence of both ADD/SUB and MUL operations.

Workaround

Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible. For instance a + b*c is a valid FMA (MUL then ADD). However (a+b)* c cannot be translated into an FMA (ADD then MUL).

Slow data structures access

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

Details

Constant non-unit stride: 5 occurrence(s)

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

Workaround

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

Vector unaligned load/store instructions

Detected 2 optimal vector unaligned load/store instructions.

Details

VMOVUPD: 2 occurrences►
- /home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:69
- /home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:69

Workaround

Use vector aligned instructions:

align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Type of elements and instruction set

24 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 144 FP arithmetical operations:

64: addition or subtraction (48 inside FMA instructions)
80: multiply (48 inside FMA instructions)

The binary loop is loading 584 bytes (73 double precision FP elements). The binary loop is storing 64 bytes (8 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.22 FP operations per loaded or stored byte.

Unroll opportunity

Loop is potentially 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	30
nb uops	29
loop length	191
used x86 registers	12
used mmx registers	0
used xmm registers	0
used ymm registers	12
used zmm registers	0
nb stack references	1
ADD-SUB / MUL ratio	0.50

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	7.83 cycles
front end	7.83 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	10.00	10.00	6.33	6.33	1.00	4.00	1.00	1.00	1.00	1.00	0.00	6.33
cycles	10.00	12.00	6.33	6.33	1.00	4.00	1.00	1.00	1.00	1.00	0.00	6.33

Execution ports to units layout:

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

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

Cycles summary

Front-end	7.83
Dispatch	12.00
Data deps.	1.00
Overall L1	12.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
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	50%
add-sub	50%
fma	50%
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 12.00 cycles. At this rate:

25% of peak load performance is reached (48.67 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
4% of peak store performance is reached (5.33 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: 47d160

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput	Vectorization
VMULPD (%R9,%RCX,1),%YMM4,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%R11,%RCX,1),%YMM2,%YMM23	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%RDI,%RCX,1),%YMM4,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%R10,%RCX,1),%YMM2,%YMM24	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%R11,%RCX,1),%YMM2,%YMM25	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RDX,%RCX,1),%YMM5,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%R8,%RCX,1),%YMM3,%YMM23	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RAX,%RCX,1),%YMM5,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RSI,%RCX,1),%YMM3,%YMM24	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%R8,1),%YMM3,%YMM25	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%R10,%RCX,1),%YMM2,%YMM26	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM23,%YMM1,%YMM1	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
MOV 0xd8(%RSP),%RBX	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
VADDPD %YMM24,%YMM0,%YMM0	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD213PD (%R15,%RCX,1),%YMM7,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD 0x20(%RSI,%RCX,1),%YMM3,%YMM26	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD132PD %YMM6,%YMM1,%YMM0	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	4	0.50	vect (50.0%)
VMULPD 0x20(%R9,%RCX,1),%YMM4,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMOVUPD %YMM0,(%R15,%RCX,1)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%RDX,1),%YMM5,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%RCX,%RDI,1),%YMM4,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM25,%YMM1,%YMM1	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%RAX,1),%YMM5,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD213PD 0x20(%R15,%RCX,1),%YMM7,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM26,%YMM0,%YMM0	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD132PD %YMM6,%YMM1,%YMM0	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	4	0.50	vect (50.0%)
VMOVUPD %YMM0,0x20(%R15,%RCX,1)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	0.50	vect (50.0%)
ADD $0x40,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%RCX	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	N/A
JNE 47d160 <_ZN16miniqmcreference17einspline_spo_refIdE8evaluateERKN11qmcplusplus11ParticleSetEiRNS2_6VectorIdSaIdEEE+0x5b0>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A

Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source file and lines	MultiBsplineRef.hpp:68-71
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

Workaround

Reduce the number of FP multiply/FMA instructions

512-bits vectorization

Workaround

Recompile with -mprefer-vector-width=512

FMA

Detected 48 FMA (fused multiply-add) operations. Presence of both ADD/SUB and MUL operations.

Workaround

Slow data structures access

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

Details

Constant non-unit stride: 5 occurrence(s)

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

Workaround

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

Vector unaligned load/store instructions

Detected 2 optimal vector unaligned load/store instructions.

Details

VMOVUPD: 2 occurrences►
- /home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:69
- /home/eoseret/qaas_runs_GNR/173-773-5521/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:69

Workaround

Use vector aligned instructions:

align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Type of elements and instruction set

24 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 144 FP arithmetical operations:

64: addition or subtraction (48 inside FMA instructions)
80: multiply (48 inside FMA instructions)

The binary loop is loading 584 bytes (73 double precision FP elements). The binary loop is storing 64 bytes (8 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.22 FP operations per loaded or stored byte.

Unroll opportunity

Loop is potentially data access bound.

Workaround

General properties

nb instructions	30
nb uops	29
loop length	191
used x86 registers	12
used mmx registers	0
used xmm registers	0
used ymm registers	12
used zmm registers	0
nb stack references	1
ADD-SUB / MUL ratio	0.50

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	7.83 cycles
front end	7.83 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	10.00	10.00	6.33	6.33	1.00	4.00	1.00	1.00	1.00	1.00	0.00	6.33
cycles	10.00	12.00	6.33	6.33	1.00	4.00	1.00	1.00	1.00	1.00	0.00	6.33

Execution ports to units layout:

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

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

Cycles summary

Front-end	7.83
Dispatch	12.00
Data deps.	1.00
Overall L1	12.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
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	50%
add-sub	50%
fma	50%
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 12.00 cycles. At this rate:

25% of peak load performance is reached (48.67 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
4% of peak store performance is reached (5.33 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: 47d160

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput	Vectorization
VMULPD (%R9,%RCX,1),%YMM4,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%R11,%RCX,1),%YMM2,%YMM23	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%RDI,%RCX,1),%YMM4,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD (%R10,%RCX,1),%YMM2,%YMM24	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%R11,%RCX,1),%YMM2,%YMM25	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RDX,%RCX,1),%YMM5,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%R8,%RCX,1),%YMM3,%YMM23	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RAX,%RCX,1),%YMM5,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD (%RSI,%RCX,1),%YMM3,%YMM24	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%R8,1),%YMM3,%YMM25	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%R10,%RCX,1),%YMM2,%YMM26	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM23,%YMM1,%YMM1	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
MOV 0xd8(%RSP),%RBX	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	1	0.33	N/A
VADDPD %YMM24,%YMM0,%YMM0	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD213PD (%R15,%RCX,1),%YMM7,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD231PD 0x20(%RSI,%RCX,1),%YMM3,%YMM26	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD132PD %YMM6,%YMM1,%YMM0	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	4	0.50	vect (50.0%)
VMULPD 0x20(%R9,%RCX,1),%YMM4,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMOVUPD %YMM0,(%R15,%RCX,1)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%RDX,1),%YMM5,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VMULPD 0x20(%RCX,%RDI,1),%YMM4,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM25,%YMM1,%YMM1	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD231PD 0x20(%RCX,%RAX,1),%YMM5,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VFMADD213PD 0x20(%R15,%RCX,1),%YMM7,%YMM1	1	0.50	0.50	0.33	0.33	0	0	0	0	0	0	0	0.33	4	0.50	vect (50.0%)
VADDPD %YMM26,%YMM0,%YMM0	1	0	0.50	0	0	0	0.50	0	0	0	0	0	0	3	0.50	vect (50.0%)
VFMADD132PD %YMM6,%YMM1,%YMM0	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	4	0.50	vect (50.0%)
VMOVUPD %YMM0,0x20(%R15,%RCX,1)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	0.50	vect (50.0%)
ADD $0x40,%RCX	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RBX,%RCX	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	N/A
JNE 47d160 <_ZN16miniqmcreference17einspline_spo_refIdE8evaluateERKN11qmcplusplus11ParticleSetEiRNS2_6VectorIdSaIdEEE+0x5b0>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

512-bits vectorization

Workaround

FMA

Workaround

Slow data structures access

Details

Workaround

Vector unaligned load/store instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

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

Workaround

512-bits vectorization

Workaround

FMA

Workaround

Slow data structures access

Details

Workaround

Vector unaligned load/store instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

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