OV - exec - Function hypre_BoomerAMGCoarsenPMIS.extracted.15

	Function: hypre_BoomerAMGCoarsenPMIS.extracted.15	Module: exec	Source: par_coarsen.c:2354-2381	Coverage (incl. loops): 0.92% \| (excl. loops): 0.00%

/home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/par_coarsen.c: 2354 - 2381

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

2354: #pragma omp parallel for private(ig, i, jS, j, jj) HYPRE_SMP_SCHEDULE

2355: #endif

2356:         for (ig = 0; ig < graph_size; ig++)

2357:         {

2358:            i = graph_array[ig];

2359:            if (measure_array[i] > 1)

2360:            {

2361:               for (jS = S_diag_i[i]; jS < S_diag_i[i+1]; jS++)

2362:               {

2363:                  j = S_diag_j[jS];

2364:                  if (measure_array[j] > 1)

2365:                  {

2366:                     if (measure_array[i] > measure_array[j])

2367:                        CF_marker[j] = 0;

2368:                     else if (measure_array[j] > measure_array[i])

2369:                        CF_marker[i] = 0;

2370:                  }

2371:               } /* for each local neighbor j of i */

2372:               for (jS = S_offd_i[i]; jS < S_offd_i[i+1]; jS++)

2373:               {

2374:                  jj = S_offd_j[jS];

2375:                  j = num_variables+jj;

2376:                  if (measure_array[j] > 1)

2377:                  {

2378:                     if (measure_array[i] > measure_array[j])

2379:                        CF_marker_offd[jj] = 0;

2380:                     else if (measure_array[j] > measure_array[i])

2381:                        CF_marker[i] = 0;

0x43d790 PUSH	%RBP

0x43d791 MOV	%RSP,%RBP

0x43d794 PUSH	%R15

0x43d796 PUSH	%R14

0x43d798 PUSH	%R13

0x43d79a PUSH	%R12

0x43d79c PUSH	%RBX

0x43d79d SUB	$0x38,%RSP

0x43d7a1 MOV	%R9,-0x50(%RBP)

0x43d7a5 MOV	%R8,%R14

0x43d7a8 MOV	%RCX,%R15

0x43d7ab MOV	%RDX,%R12

0x43d7ae MOV	0x30(%RBP),%RAX

0x43d7b2 MOV	%RAX,-0x48(%RBP)

0x43d7b6 MOV	0x28(%RBP),%R13

0x43d7ba MOV	0x20(%RBP),%RAX

0x43d7be MOV	%RAX,-0x58(%RBP)

0x43d7c2 MOV	0x18(%RBP),%RAX

0x43d7c6 MOV	%RAX,-0x38(%RBP)

0x43d7ca MOV	0x10(%RBP),%RBX

0x43d7ce MOVL	$0,-0x30(%RBP)

0x43d7d5 MOV	(%RDI),%ESI

0x43d7d7 MOVQ	$0,-0x40(%RBP)

0x43d7df MOVQ	$0x1,-0x60(%RBP)

0x43d7e7 SUB	$0x8,%RSP

0x43d7eb LEA	-0x60(%RBP),%RAX

0x43d7ef LEA	-0x30(%RBP),%RCX

0x43d7f3 LEA	-0x40(%RBP),%R8

0x43d7f7 LEA	0x40(%RBP),%R9

0x43d7fb MOV	$0x605b10,%EDI

0x43d800 MOV	%ESI,-0x2c(%RBP)

0x43d803 MOV	$0x22,%EDX

0x43d808 PUSH	$0x1

0x43d80a PUSH	$0x1

0x43d80c PUSH	%RAX

0x43d80d CALL	41dc20 <__kmpc_for_static_init_8@plt>

0x43d812 MOV	-0x48(%RBP),%R11

0x43d816 MOV	-0x50(%RBP),%R10

0x43d81a ADD	$0x20,%RSP

0x43d81e MOV	-0x40(%RBP),%RCX

0x43d822 MOV	0x40(%RBP),%RAX

0x43d826 CMP	%RAX,%RCX

0x43d829 JBE	43d846

0x43d82b MOV	$0x605b30,%EDI

0x43d830 MOV	-0x2c(%RBP),%ESI

0x43d833 ADD	$0x38,%RSP

0x43d837 POP	%RBX

0x43d838 POP	%R12

0x43d83a POP	%R13

0x43d83c POP	%R14

0x43d83e POP	%R15

0x43d840 POP	%RBP

0x43d841 JMP	41d3d0

0x43d846 MOVSD	0x1867e2(%RIP),%XMM0

0x43d84e JMP	43d85c

(300) 0x43d850 LEA	0x1(%RCX),%RDX

(300) 0x43d854 CMP	%RAX,%RCX

(300) 0x43d857 MOV	%RDX,%RCX

(300) 0x43d85a JE	43d82b

(300) 0x43d85c MOV	(%R11,%RCX,8),%RDX

(300) 0x43d860 MOVSD	(%R13,%RDX,8),%XMM1

(300) 0x43d867 UCOMISD	%XMM1,%XMM0

(300) 0x43d86b JAE	43d850

(300) 0x43d86d MOV	(%R12,%RDX,8),%RSI

(300) 0x43d871 CMP	0x8(%R12,%RDX,8),%RSI

(300) 0x43d876 MOV	-0x38(%RBP),%R8

(300) 0x43d87a JL	43d8b2

(300) 0x43d87c MOV	(%R14,%RDX,8),%RSI

(300) 0x43d880 CMP	0x8(%R14,%RDX,8),%RSI

(300) 0x43d885 JGE	43d850

(300) 0x43d887 MOV	-0x38(%RBP),%RDI

(300) 0x43d88b LEA	(%RDI,%RDX,8),%RDI

(300) 0x43d88f JMP	43d8fd

0x43d891 NOPW	%CS:(%RAX,%RAX,1)

  (302) 0x43d8a0 MOVQ	$0,(%R8,%RDI,8)

  (302) 0x43d8a8 INC	%RSI

  (302) 0x43d8ab CMP	0x8(%R12,%RDX,8),%RSI

  (302) 0x43d8b0 JGE	43d87c

  (302) 0x43d8b2 MOV	(%R15,%RSI,8),%RDI

  (302) 0x43d8b6 MOVSD	(%R13,%RDI,8),%XMM2

  (302) 0x43d8bd UCOMISD	%XMM2,%XMM0

  (302) 0x43d8c1 JAE	43d8a8

  (302) 0x43d8c3 UCOMISD	%XMM2,%XMM1

  (302) 0x43d8c7 JA	43d8a0

  (302) 0x43d8c9 UCOMISD	%XMM1,%XMM2

  (302) 0x43d8cd MOV	%RDX,%RDI

  (302) 0x43d8d0 JA	43d8a0

  (302) 0x43d8d2 JMP	43d8a8

0x43d8d4 NOPW	%CS:(%RAX,%RAX,1)

  (301) 0x43d8e0 MOV	-0x58(%RBP),%R9

  (301) 0x43d8e4 LEA	(%R9,%R8,8),%R8

  (301) 0x43d8e8 MOVQ	$0,(%R8)

  (301) 0x43d8ef INC	%RSI

  (301) 0x43d8f2 CMP	0x8(%R14,%RDX,8),%RSI

  (301) 0x43d8f7 JGE	43d850

  (301) 0x43d8fd MOV	(%R10,%RSI,8),%R8

  (301) 0x43d901 LEA	(%R8,%RBX,1),%R9

  (301) 0x43d905 MOVSD	(%R13,%R9,8),%XMM2

  (301) 0x43d90c UCOMISD	%XMM2,%XMM0

  (301) 0x43d910 JAE	43d8ef

  (301) 0x43d912 UCOMISD	%XMM2,%XMM1

  (301) 0x43d916 JA	43d8e0

  (301) 0x43d918 UCOMISD	%XMM1,%XMM2

  (301) 0x43d91c MOV	%RDI,%R8

  (301) 0x43d91f JA	43d8e8

  (301) 0x43d921 JMP	43d8ef

0x43d923 NOPW	%CS:(%RAX,%RAX,1)

Coverage (%)	Name	Source Location	Module
►96.53+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►2.67+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	hypre_BoomerAMGCoarsenPMIS	par_coarsen.c:2354	exec
○	hypre_BoomerAMGSetup	par_amg_setup.c:612	exec
○	hypre_PCGSetup	pcg.c:234	exec
○	main	amg.c:398	exec
○	__libc_start_call_main		libc.so.6
○	__libc_start_main		libc.so.6
○	_start		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 /

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 icx_5

Source file and lines	par_coarsen.c:2354-2381
Module	exec

The function is defined in /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/par_coarsen.c:2354-2381.

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 7.00 to 0.83 cycles (8.40x 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 10% 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 7.00 to 0.52 cycles (13.45x 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.

No data for this section

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.

__kmpc_for_static_init_8@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/par_coarsen.c:2354

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This function is probably writing/copying data or processing integer elements.

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

The binary function does not contain any FP arithmetical operations. The binary function is loading 88 bytes. The binary function is storing 56 bytes.

General properties

nb instructions	58
nb uops	56
loop length	232
used x86 registers	16
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	0
nb stack references	14

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	7.00 cycles
front end	7.00 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	2.67	2.67	2.67	2.67	2.67	2.67	5.25	5.25	5.25	5.25	0.00	0.00	0.00	0.00	0.00	0.00
cycles	2.67	2.67	2.67	2.67	2.67	2.67	5.25	5.25	5.25	5.25	0.00	0.00	0.00	0.00	0.00	0.00

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

Cycles summary

Front-end	7.00
Dispatch	5.25
Overall L1	7.00

Vectorization ratios

INT

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

all	0%
load	0%
store	NA (no store vectorizable/vectorized instructions)
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)

INT+FP

all	0%
load	0%
store	0%
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	0%

Vector efficiency ratios

INT

all	10%
load	10%
store	9%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	11%

all	12%
load	12%
store	NA (no store vectorizable/vectorized instructions)
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)

INT+FP

all	10%
load	10%
store	9%
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	11%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 7.00 cycles. At this rate:

4% of peak load performance is reached (12.57 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (8.00 out of 128.00 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). By removing all these bottlenecks, you can lower the cost of an iteration from 7.00 to 5.25 cycles (1.33x speedup).

ASM code

In the binary file, the address of the function is: 43d790

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
PUSH %R15	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R14	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R13	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R12	1	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	1	0.50	N/A
SUB $0x38,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV %R9,-0x50(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV %R8,%R14	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV %RCX,%R15	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV %RDX,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV 0x30(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x48(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x28(%RBP),%R13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV 0x20(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x58(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x18(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x38(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x10(%RBP),%RBX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOVL $0,-0x30(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOV (%RDI),%ESI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (6.3%)
MOVQ $0,-0x40(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOVQ $0x1,-0x60(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
SUB $0x8,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
LEA -0x60(%RBP),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA -0x30(%RBP),%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA -0x40(%RBP),%R8	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA 0x40(%RBP),%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
MOV $0x605b10,%EDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV %ESI,-0x2c(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOV $0x22,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (6.3%)
PUSH $0x1	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH $0x1	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RAX	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
CALL 41dc20 <__kmpc_for_static_init_8@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0.50	N/A
MOV -0x48(%RBP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV -0x50(%RBP),%R10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
ADD $0x20,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV -0x40(%RBP),%RCX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV 0x40(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
CMP %RAX,%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (12.5%)
JBE 43d846 <hypre_BoomerAMGCoarsenPMIS.extracted.15+0xb6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	1	0.50-1	N/A
MOV $0x605b30,%EDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV -0x2c(%RBP),%ESI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (6.3%)
ADD $0x38,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
POP %RBX	1	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	1	0.33	N/A
POP %R13	1	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R14	1	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R15	1	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	1	0.33	N/A
JMP 41d3d0 <__kmpc_for_static_fini@plt>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	1	N/A
MOVSD 0x1867e2(%RIP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	1	0.50	scal (12.5%)
JMP 43d85c <hypre_BoomerAMGCoarsenPMIS.extracted.15+0xcc>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	1	N/A
NOPW %CS:(%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
NOPW %CS:(%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
NOPW %CS:(%RAX,%RAX,1)	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 icx_5

Source file and lines	par_coarsen.c:2354-2381
Module	exec

The function is defined in /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/par_coarsen.c:2354-2381.

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 7.00 to 0.83 cycles (8.40x 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.

No data for this section

CALL instructions

Detected function call instructions.

Details

__kmpc_for_static_init_8@plt: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/173-989-6379/intel/AMG/build/AMG/AMG/parcsr_ls/par_coarsen.c:2354

Type of elements and instruction set

No instructions are processing arithmetic or math operations on FP elements. This function is probably writing/copying data or processing integer elements.

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

The binary function does not contain any FP arithmetical operations. The binary function is loading 88 bytes. The binary function is storing 56 bytes.

General properties

nb instructions	58
nb uops	56
loop length	232
used x86 registers	16
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	0
nb stack references	14

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	7.00 cycles
front end	7.00 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	2.67	2.67	2.67	2.67	2.67	2.67	5.25	5.25	5.25	5.25	0.00	0.00	0.00	0.00	0.00	0.00
cycles	2.67	2.67	2.67	2.67	2.67	2.67	5.25	5.25	5.25	5.25	0.00	0.00	0.00	0.00	0.00	0.00

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

Cycles summary

Front-end	7.00
Dispatch	5.25
Overall L1	7.00

Vectorization ratios

INT

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

all	0%
load	0%
store	NA (no store vectorizable/vectorized instructions)
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)

INT+FP

all	0%
load	0%
store	0%
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	0%

Vector efficiency ratios

INT

all	10%
load	10%
store	9%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	11%

all	12%
load	12%
store	NA (no store vectorizable/vectorized instructions)
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)

INT+FP

all	10%
load	10%
store	9%
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	11%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 7.00 cycles. At this rate:

4% of peak load performance is reached (12.57 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
6% of peak store performance is reached (8.00 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

ASM code

In the binary file, the address of the function is: 43d790

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	Latency	Recip. throughput	Vectorization
PUSH %RBP	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0	0	0.17	N/A
PUSH %R15	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R14	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R13	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %R12	1	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	1	0.50	N/A
SUB $0x38,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV %R9,-0x50(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV %R8,%R14	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV %RCX,%R15	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV %RDX,%R12	1	0	0	0	0	0	0	0	0	0	0	0	0.17	scal (12.5%)
MOV 0x30(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x48(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x28(%RBP),%R13	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV 0x20(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x58(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x18(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
MOV %RAX,-0x38(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (12.5%)
MOV 0x10(%RBP),%RBX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOVL $0,-0x30(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOV (%RDI),%ESI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (6.3%)
MOVQ $0,-0x40(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOVQ $0x1,-0x60(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
SUB $0x8,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
LEA -0x60(%RBP),%RAX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA -0x30(%RBP),%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA -0x40(%RBP),%R8	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
LEA 0x40(%RBP),%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.25	N/A
MOV $0x605b10,%EDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV %ESI,-0x2c(%RBP)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	4	0.50	scal (6.3%)
MOV $0x22,%EDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (6.3%)
PUSH $0x1	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH $0x1	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
PUSH %RAX	1	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A
CALL 41dc20 <__kmpc_for_static_init_8@plt>	2	0	0	0	0.33	0.33	0.33	0.25	0.25	0.25	0.25	0	0.50	N/A
MOV -0x48(%RBP),%R11	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV -0x50(%RBP),%R10	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
ADD $0x20,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV -0x40(%RBP),%RCX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (12.5%)
MOV 0x40(%RBP),%RAX	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	N/A
CMP %RAX,%RCX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	scal (12.5%)
JBE 43d846 <hypre_BoomerAMGCoarsenPMIS.extracted.15+0xb6>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	1	0.50-1	N/A
MOV $0x605b30,%EDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
MOV -0x2c(%RBP),%ESI	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	3	0.25	scal (6.3%)
ADD $0x38,%RSP	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	1	0.17	N/A
POP %RBX	1	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	1	0.33	N/A
POP %R13	1	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R14	1	0	0	0	0	0	0	0	0	0	0	1	0.33	N/A
POP %R15	1	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	1	0.33	N/A
JMP 41d3d0 <__kmpc_for_static_fini@plt>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	1	N/A
MOVSD 0x1867e2(%RIP),%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	1	0.50	scal (12.5%)
JMP 43d85c <hypre_BoomerAMGCoarsenPMIS.extracted.15+0xcc>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	1	N/A
NOPW %CS:(%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
NOPW %CS:(%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A
NOPW %CS:(%RAX,%RAX,1)	0	0	0	0	0	0	0	0	0	0	0	0	0.09	N/A

Name	Coverage (%)	Time (s)
▼hypre_BoomerAMGCoarsenPMIS.extracted.15–	0.92	0.48
▼Loop 300 - par_coarsen.c:2354-2381 - exec–	0.06	0.03
○Loop 302 - par_coarsen.c:2361-2381 - exec	0.86	0.42
○Loop 301 - par_coarsen.c:2368-2380 - exec	0.00	0.01

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

CALL instructions

Details

Type of elements and instruction set

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

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

CALL instructions

Details

Type of elements and instruction set

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code