OV - exec - Loop 80

0x416e50 VMOVUPD	(%R10,%R12,8),%ZMM2    [1]

0x416e57 VMULPD	(%RDI,%R12,8),%ZMM2,%ZMM3    [8]

0x416e5e VADDPD	(%RBX,%R12,8),%ZMM3,%ZMM4    [9]

0x416e65 VSUBPD	(%R15,%R12,8),%ZMM4,%ZMM4    [6]

0x416e6c VMOVUPD	(%R9,%R12,8),%ZMM5    [7]

0x416e73 VFMADD213PD	(%RCX,%R12,8),%ZMM3,%ZMM5    [2]

0x416e7a VSUBPD	(%R11,%R12,8),%ZMM5,%ZMM3    [3]

0x416e81 VDIVPD	%ZMM4,%ZMM3,%ZMM3

0x416e87 VADDPD	(%R13,%R12,8),%ZMM2,%ZMM2    [4]

0x416e8f VSUBPD	(%R14,%R12,8),%ZMM2,%ZMM2    [5]

0x416e96 VDIVPD	%ZMM2,%ZMM4,%ZMM2

0x416e9c VMOVUPD	%ZMM2,(%RDI,%R12,8)    [8]

0x416ea3 VMOVUPD	%ZMM3,(%R9,%R12,8)    [7]

0x416eaa ADD	$0x8,%R12

0x416eae CMP	%R8,%R12

0x416eb1 JB	416e50

/home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90: 255 - 261

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

255:         DO j=x_min,x_max

256:           pre_mass_s=density1(j,k)*pre_vol(j,k)

257:           post_mass_s=pre_mass_s+mass_flux_y(j,k)-mass_flux_y(j,k+1)

258:           post_ener_s=(energy1(j,k)*pre_mass_s+ener_flux(j,k)-ener_flux(j,k+1))/post_mass_s

259:           advec_vol_s=pre_vol(j,k)+vol_flux_y(j,k)-vol_flux_y(j,k+1)

260:           density1(j,k)=post_mass_s/advec_vol_s

261:           energy1(j,k)=post_ener_s

Coverage (%)	Name	Source Location	Module
►100.00+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	4.92
Bottlenecks	P0,
Function	advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2
Source	advec_cell_kernel.f90:255-261
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	32.00
CQA cycles if no scalar integer	32.00
CQA cycles if FP arith vectorized	32.00
CQA cycles if fully vectorized	32.00
Front-end cycles	4.33
P0 cycles	6.50
P1 cycles	3.50
P2 cycles	3.00
P3 cycles	3.00
P4 cycles	1.00
P5 cycles	6.50
P6 cycles	1.00
P7 cycles	1.00
P8 cycles	1.00
P9 cycles	1.00
P10 cycles	0.00
P11 cycles	3.00
DIV/SQRT cycles	32.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	16.00
Nb uops	19.00
Nb loads	9.00
Nb stores	2.00
Nb stack references	0.00
FLOP/cycle	2.50
Nb FLOP add-sub	40.00
Nb FLOP mul	8.00
Nb FLOP fma	8.00
Nb FLOP div	16.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	22.00
Bytes prefetched	0.00
Bytes loaded	576.00
Bytes stored	128.00
Stride 0	0.00
Stride 1	9.00
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	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	NA
Vector-efficiency ratio all	100.00
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	100.00
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
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	1.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	4.92
Bottlenecks	P0,
Function	advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2
Source	advec_cell_kernel.f90:255-261
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	32.00
CQA cycles if no scalar integer	32.00
CQA cycles if FP arith vectorized	32.00
CQA cycles if fully vectorized	32.00
Front-end cycles	4.33
P0 cycles	6.50
P1 cycles	3.50
P2 cycles	3.00
P3 cycles	3.00
P4 cycles	1.00
P5 cycles	6.50
P6 cycles	1.00
P7 cycles	1.00
P8 cycles	1.00
P9 cycles	1.00
P10 cycles	0.00
P11 cycles	3.00
DIV/SQRT cycles	32.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	16.00
Nb uops	19.00
Nb loads	9.00
Nb stores	2.00
Nb stack references	0.00
FLOP/cycle	2.50
Nb FLOP add-sub	40.00
Nb FLOP mul	8.00
Nb FLOP fma	8.00
Nb FLOP div	16.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	22.00
Bytes prefetched	0.00
Bytes loaded	576.00
Bytes stored	128.00
Stride 0	0.00
Stride 1	9.00
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	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	NA
Vector-efficiency ratio all	100.00
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	100.00
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
Vector-efficiency ratio other	NA

Path /

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

Function	advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2
Source file and lines	advec_cell_kernel.f90:255-261
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:255-261.

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

gain
potential
hint
expert

Vectorization

Your loop is fully vectorized, using full register length.

Details

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

Execution units bottlenecks

Performance is limited by execution of divide and square root operations (the divide/square root unit is a bottleneck). By removing all these bottlenecks, you can lower the cost of an iteration from 32.00 to 6.50 cycles (4.92x speedup).

Workaround

Reduce the number of division or square root instructions:
- If denominator is constant over iterations, use reciprocal (replace x/y with x*(1/y)). Check precision impact. This will be done by your compiler with no-prec-div or Ofast
Check whether you really need double precision. If not, switch to single precision to speedup execution

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 32.00 to 4.50 cycles (7.11x speedup).

FMA

Detected 8 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).

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.

VDIVPD: 2 occurrences►
- /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:258
- /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:260

Type of elements and instruction set

9 AVX-512 instructions are processing arithmetic or math operations on double precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 80 FP arithmetical operations:

48: addition or subtraction (8 inside FMA instructions)
16: multiply (8 inside FMA instructions)
16: divide

The binary loop is loading 576 bytes (72 double precision FP elements). The binary loop is storing 128 bytes (16 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.11 FP operations per loaded or stored byte.

General properties

nb instructions	16
nb uops	19
loop length	99
used x86 registers	11
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	4
nb stack references	0
ADD-SUB / MUL ratio	5.00

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	4.33 cycles
front end	4.33 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	6.50	0.00	3.00	3.00	1.00	6.50	1.00	1.00	1.00	1.00	0.00	3.00
cycles	6.50	3.50	3.00	3.00	1.00	6.50	1.00	1.00	1.00	1.00	0.00	3.00

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	32.00
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	4.33
Dispatch	6.50
DIV/SQRT	32.00
Data deps.	1.00
Overall L1	32.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
div/sqrt	100%
other	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
div/sqrt	100%
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 32.00 cycles. At this rate:

9% of peak load performance is reached (18.00 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
3% of peak store performance is reached (4.00 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: 416e50

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput	Vectorization
VMOVUPD (%R10,%R12,8),%ZMM2	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	0-1	0.50	vect (100.0%)
VMULPD (%RDI,%R12,8),%ZMM2,%ZMM3	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	4	0.50	vect (100.0%)
VADDPD (%RBX,%R12,8),%ZMM3,%ZMM4	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VSUBPD (%R15,%R12,8),%ZMM4,%ZMM4	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VMOVUPD (%R9,%R12,8),%ZMM5	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	0-1	0.50	vect (100.0%)
VFMADD213PD (%RCX,%R12,8),%ZMM3,%ZMM5	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	4	0.50	vect (100.0%)
VSUBPD (%R11,%R12,8),%ZMM5,%ZMM3	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VDIVPD %ZMM4,%ZMM3,%ZMM3	3	2.50	0	0	0	0	0.50	0	0	0	0	0	0	22-24	16	vect (100.0%)
VADDPD (%R13,%R12,8),%ZMM2,%ZMM2	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VSUBPD (%R14,%R12,8),%ZMM2,%ZMM2	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VDIVPD %ZMM2,%ZMM4,%ZMM2	3	2.50	0	0	0	0	0.50	0	0	0	0	0	0	22-24	16	vect (100.0%)
VMOVUPD %ZMM2,(%RDI,%R12,8)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	1	vect (100.0%)
VMOVUPD %ZMM3,(%R9,%R12,8)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	1	vect (100.0%)
ADD $0x8,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %R8,%R12	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	N/A
JB 416e50 <advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2+0xb90>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A

Function	advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2
Source file and lines	advec_cell_kernel.f90:255-261
Module	exec

The loop is defined in /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:255-261.

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

gain
potential
hint
expert

Vectorization

Your loop is fully vectorized, using full register length.

Details

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

Execution units bottlenecks

Performance is limited by execution of divide and square root operations (the divide/square root unit is a bottleneck). By removing all these bottlenecks, you can lower the cost of an iteration from 32.00 to 6.50 cycles (4.92x speedup).

Workaround

Reduce the number of division or square root instructions:
- If denominator is constant over iterations, use reciprocal (replace x/y with x*(1/y)). Check precision impact. This will be done by your compiler with no-prec-div or Ofast
Check whether you really need double precision. If not, switch to single precision to speedup execution

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 32.00 to 4.50 cycles (7.11x speedup).

FMA

Detected 8 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).

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.

VDIVPD: 2 occurrences►
- /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:258
- /home/eoseret/qaas_runs_GNR/173-814-3702/intel/CloverLeaf1.3-FC/build/CloverLeaf1.3-FC/CloverLeaf_ref/kernels/advec_cell_kernel.f90:260

Type of elements and instruction set

9 AVX-512 instructions are processing arithmetic or math operations on double precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 80 FP arithmetical operations:

48: addition or subtraction (8 inside FMA instructions)
16: multiply (8 inside FMA instructions)
16: divide

The binary loop is loading 576 bytes (72 double precision FP elements). The binary loop is storing 128 bytes (16 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.11 FP operations per loaded or stored byte.

General properties

nb instructions	16
nb uops	19
loop length	99
used x86 registers	11
used mmx registers	0
used xmm registers	0
used ymm registers	0
used zmm registers	4
nb stack references	0
ADD-SUB / MUL ratio	5.00

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	4.33 cycles
front end	4.33 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	6.50	0.00	3.00	3.00	1.00	6.50	1.00	1.00	1.00	1.00	0.00	3.00
cycles	6.50	3.50	3.00	3.00	1.00	6.50	1.00	1.00	1.00	1.00	0.00	3.00

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	32.00
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	4.33
Dispatch	6.50
DIV/SQRT	32.00
Data deps.	1.00
Overall L1	32.00

Vectorization ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
div/sqrt	100%
other	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
div/sqrt	100%
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 32.00 cycles. At this rate:

9% of peak load performance is reached (18.00 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))
3% of peak store performance is reached (4.00 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: 416e50

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput	Vectorization
VMOVUPD (%R10,%R12,8),%ZMM2	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	0-1	0.50	vect (100.0%)
VMULPD (%RDI,%R12,8),%ZMM2,%ZMM3	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	4	0.50	vect (100.0%)
VADDPD (%RBX,%R12,8),%ZMM3,%ZMM4	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VSUBPD (%R15,%R12,8),%ZMM4,%ZMM4	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VMOVUPD (%R9,%R12,8),%ZMM5	1	0	0	0.33	0.33	0	0	0	0	0	0	0	0.33	0-1	0.50	vect (100.0%)
VFMADD213PD (%RCX,%R12,8),%ZMM3,%ZMM5	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	4	0.50	vect (100.0%)
VSUBPD (%R11,%R12,8),%ZMM5,%ZMM3	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VDIVPD %ZMM4,%ZMM3,%ZMM3	3	2.50	0	0	0	0	0.50	0	0	0	0	0	0	22-24	16	vect (100.0%)
VADDPD (%R13,%R12,8),%ZMM2,%ZMM2	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VSUBPD (%R14,%R12,8),%ZMM2,%ZMM2	1	0.50	0	0.33	0.33	0	0.50	0	0	0	0	0	0.33	3	0.65	vect (100.0%)
VDIVPD %ZMM2,%ZMM4,%ZMM2	3	2.50	0	0	0	0	0.50	0	0	0	0	0	0	22-24	16	vect (100.0%)
VMOVUPD %ZMM2,(%RDI,%R12,8)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	1	vect (100.0%)
VMOVUPD %ZMM3,(%R9,%R12,8)	1	0	0	0	0	0.50	0	0	0.50	0.50	0.50	0	0	0-1	1	vect (100.0%)
ADD $0x8,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %R8,%R12	1	0.20	0.20	0	0	0	0.20	0.20	0	0	0	0.20	0	1	0.20	N/A
JB 416e50 <advec_cell_kernel_module_mp_advec_cell_kernel_.DIR.OMP.PARALLEL.2+0xb90>	1	0.50	0	0	0	0	0	0.50	0	0	0	0	0	0	0.50	N/A

Report Configuration

Vectorization

Details

Execution units bottlenecks

Workaround

Expensive FP math instructions/calls

FMA

Workaround

Complex instructions

Details

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

Execution units bottlenecks

Workaround

Expensive FP math instructions/calls

FMA

Workaround

Complex instructions

Details

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