OV - exec - Function _ZN16miniqmcreference19DiracDeterminantRefIN11qmcplusplus13DelayedUpdateIddEEE14evaluateRatiosERNS1

	Function: _ZN16miniqmcreference19DiracDeterminantRefIN11qmcplusplus13DelayedUpdateIddEEE14evaluateRa ...	Module: exec	Source: DiracDeterminantRef.cpp:211-217 [...]	Coverage: 0.01%

/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/unique_ptr.h: 173 - 173

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

173:       pointer    _M_ptr() const { return std::get<0>(_M_t); }

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/QMCWaveFunctions/DiracDeterminantRef.cpp: 211 - 217

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

211: {

212:   SPOVTimer->start();

213:   const int WorkingIndex = VP.refPtcl - FirstIndex;

214:   invRow_id              = WorkingIndex;

215:   updateEng.getInvRow(psiM, WorkingIndex, invRow);

216:   Phi->evaluateDetRatios(VP, psiV, invRow, ratios);

217:   SPOVTimer->stop();

0x44ef70 PUSH	%RBP

0x44ef71 MOV	%RSP,%RBP

0x44ef74 PUSH	%R15

0x44ef76 PUSH	%R14

0x44ef78 PUSH	%R12

0x44ef7a PUSH	%RBX

0x44ef7b MOV	%RDX,%R14

0x44ef7e MOV	%RSI,%R15

0x44ef81 MOV	%RDI,%RBX

0x44ef84 MOV	0x460(%RDI),%RDI

0x44ef8b CALL	497e00 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>

0x44ef90 MOV	0x2a0(%R15),%EDX

0x44ef97 SUB	0x478(%RBX),%EDX

0x44ef9d MOV	%EDX,0x430(%RBX)

0x44efa3 LEA	0x208(%RBX),%RDI

0x44efaa LEA	0xd0(%RBX),%RSI

0x44efb1 LEA	0x408(%RBX),%R12

0x44efb8 MOV	%R12,%RCX

0x44efbb CALL	44bc20 <_ZN11qmcplusplus13DelayedUpdateIddE9getInvRowINS_6VectorIdSaIdEEEEEvRKNS_6MatrixIdS4_EEiRT_>

0x44efc0 MOV	0x470(%RBX),%RDI

0x44efc7 LEA	0x190(%RBX),%RDX

0x44efce MOV	(%RDI),%RAX

0x44efd1 MOV	%R15,%RSI

0x44efd4 MOV	%R12,%RCX

0x44efd7 MOV	%R14,%R8

0x44efda CALLQ	0x20(%RAX)

0x44efdd MOV	0x460(%RBX),%RDI

0x44efe4 POP	%RBX

0x44efe5 POP	%R12

0x44efe7 POP	%R14

0x44efe9 POP	%R15

0x44efeb POP	%RBP

0x44efec JMP	497fd0

0x44eff1 NOPW	%CS:(%RAX,%RAX,1)

0x44effb NOPL	(%RAX,%RAX,1)

Path /

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

Source file and lines	DiracDeterminantRef.cpp:211-217
Module	exec

The function is defined in:

/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/unique_ptr.h:173
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/QMCWaveFunctions/DiracDeterminantRef.cpp:211-217

gain
potential
hint
expert

Vectorization

Your function is not vectorized. Only 11% 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 6.33 to 0.65 cycles (9.81x 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.

_ZN11qmcplusplus13DelayedUpdateIddE9getInvRowINS_6VectorIdSaIdEEEEEvRKNS_6MatrixIdS4_EEiRT_: 1 occurrences
_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv: 1 occurrences
unknown: 1 occurrences

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 48 bytes. The binary function is storing 4 bytes.

General properties

nb instructions	33
nb uops	38
loop length	129
used x86 registers	12
used mmx registers	0
used xmm registers	0
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	6.33 cycles
front end	6.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	2.00	1.75	1.75	1.50	2.00	3.67	3.67	3.67	0.00	0.00	0.00	0.00	0.00	0.00
cycles	2.00	1.75	1.75	1.50	2.00	3.67	3.67	3.67	0.00	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

ALU0/BRU0: ALU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1:
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

Cycles summary

Front-end	6.33
Dispatch	3.67
Overall L1	6.33

Vectorization ratios

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

all	11%
load	NA (no load vectorizable/vectorized instructions)
store	6%
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	12%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 6.33 cycles. At this rate:

7% of peak load performance is reached (7.58 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
0% of peak store performance is reached (0.63 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). By removing all these bottlenecks, you can lower the cost of an iteration from 6.33 to 3.67 cycles (1.73x speedup).

ASM code

In the binary file, the address of the function is: 44ef70

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	Latency	Recip. throughput
PUSH %RBP	1	0	0	0	0	0	0	0	0	1	0.50
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0.17
PUSH %R15	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %R14	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %R12	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %RBX	1	0	0	0	0	0	0	0	0	1	0.50
MOV %RDX,%R14	1	0	0	0	0	0	0	0	0	0	0.17
MOV %RSI,%R15	1	0	0	0	0	0	0	0	0	0	0.17
MOV %RDI,%RBX	1	0	0	0	0	0	0	0	0	0	0.17
MOV 0x460(%RDI),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
CALL 497e00 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0.50
MOV 0x2a0(%R15),%EDX	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
SUB 0x478(%RBX),%EDX	1	0.25	0.25	0.25	0.25	0	0.33	0.33	0.33	1	0.33
MOV %EDX,0x430(%RBX)	1	0	0	0	0	0	0.33	0.33	0.33	4	0.50
LEA 0x208(%RBX),%RDI	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
LEA 0xd0(%RBX),%RSI	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
LEA 0x408(%RBX),%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
MOV %R12,%RCX	1	0	0	0	0	0	0	0	0	0	0.17
CALL 44bc20 <_ZN11qmcplusplus13DelayedUpdateIddE9getInvRowINS_6VectorIdSaIdEEEEEvRKNS_6MatrixIdS4_EEiRT_>	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0.50
MOV 0x470(%RBX),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
LEA 0x190(%RBX),%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
MOV (%RDI),%RAX	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
MOV %R15,%RSI	1	0	0	0	0	0	0	0	0	0	0.17
MOV %R12,%RCX	1	0	0	0	0	0	0	0	0	0	0.17
MOV %R14,%R8	1	0	0	0	0	0	0	0	0	0	0.17
CALLQ 0x20(%RAX)	4	0.50	0	0	0	0.50	0.67	0.67	0.67	0	2.20
MOV 0x460(%RBX),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
POP %RBX	1	0	0	0	0	0	0	0	0	1	0.33
POP %R12	1	0	0	0	0	0	0	0	0	1	0.33
POP %R14	1	0	0	0	0	0	0	0	0	1	0.33
POP %R15	1	0	0	0	0	0	0	0	0	1	0.33
POP %RBP	1	0	0	0	0	0	0	0	0	1	0.33
JMP 497fd0 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE4stopEv>	1	0.50	0	0	0	0.50	0	0	0	0	1

Source file and lines	DiracDeterminantRef.cpp:211-217
Module	exec

The function is defined in:

/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/unique_ptr.h:173
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/QMCWaveFunctions/DiracDeterminantRef.cpp:211-217

gain
potential
hint
expert

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

_ZN11qmcplusplus13DelayedUpdateIddE9getInvRowINS_6VectorIdSaIdEEEEEvRKNS_6MatrixIdS4_EEiRT_: 1 occurrences
_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv: 1 occurrences
unknown: 1 occurrences

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 48 bytes. The binary function is storing 4 bytes.

General properties

nb instructions	33
nb uops	38
loop length	129
used x86 registers	12
used mmx registers	0
used xmm registers	0
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	6.33 cycles
front end	6.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	2.00	1.75	1.75	1.50	2.00	3.67	3.67	3.67	0.00	0.00	0.00	0.00	0.00	0.00
cycles	2.00	1.75	1.75	1.50	2.00	3.67	3.67	3.67	0.00	0.00	0.00	0.00	0.00	0.00

Execution ports to units layout:

ALU0/BRU0: ALU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1:
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

Cycles summary

Front-end	6.33
Dispatch	3.67
Overall L1	6.33

Vectorization ratios

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

all	11%
load	NA (no load vectorizable/vectorized instructions)
store	6%
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	12%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each call to the function takes 6.33 cycles. At this rate:

7% of peak load performance is reached (7.58 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
0% of peak store performance is reached (0.63 out of 64.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

ASM code

In the binary file, the address of the function is: 44ef70

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	Latency	Recip. throughput
PUSH %RBP	1	0	0	0	0	0	0	0	0	1	0.50
MOV %RSP,%RBP	1	0	0	0	0	0	0	0	0	0	0.17
PUSH %R15	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %R14	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %R12	1	0	0	0	0	0	0	0	0	1	0.50
PUSH %RBX	1	0	0	0	0	0	0	0	0	1	0.50
MOV %RDX,%R14	1	0	0	0	0	0	0	0	0	0	0.17
MOV %RSI,%R15	1	0	0	0	0	0	0	0	0	0	0.17
MOV %RDI,%RBX	1	0	0	0	0	0	0	0	0	0	0.17
MOV 0x460(%RDI),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
CALL 497e00 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE5startEv>	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0.50
MOV 0x2a0(%R15),%EDX	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
SUB 0x478(%RBX),%EDX	1	0.25	0.25	0.25	0.25	0	0.33	0.33	0.33	1	0.33
MOV %EDX,0x430(%RBX)	1	0	0	0	0	0	0.33	0.33	0.33	4	0.50
LEA 0x208(%RBX),%RDI	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
LEA 0xd0(%RBX),%RSI	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
LEA 0x408(%RBX),%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
MOV %R12,%RCX	1	0	0	0	0	0	0	0	0	0	0.17
CALL 44bc20 <_ZN11qmcplusplus13DelayedUpdateIddE9getInvRowINS_6VectorIdSaIdEEEEEvRKNS_6MatrixIdS4_EEiRT_>	2	0.50	0	0	0	0.50	0.33	0.33	0.33	0	0.50
MOV 0x470(%RBX),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
LEA 0x190(%RBX),%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	1	0.25
MOV (%RDI),%RAX	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
MOV %R15,%RSI	1	0	0	0	0	0	0	0	0	0	0.17
MOV %R12,%RCX	1	0	0	0	0	0	0	0	0	0	0.17
MOV %R14,%R8	1	0	0	0	0	0	0	0	0	0	0.17
CALLQ 0x20(%RAX)	4	0.50	0	0	0	0.50	0.67	0.67	0.67	0	2.20
MOV 0x460(%RBX),%RDI	1	0	0	0	0	0	0.33	0.33	0.33	3	0.33
POP %RBX	1	0	0	0	0	0	0	0	0	1	0.33
POP %R12	1	0	0	0	0	0	0	0	0	1	0.33
POP %R14	1	0	0	0	0	0	0	0	0	1	0.33
POP %R15	1	0	0	0	0	0	0	0	0	1	0.33
POP %RBP	1	0	0	0	0	0	0	0	0	1	0.33
JMP 497fd0 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE4stopEv>	1	0.50	0	0	0	0.50	0	0	0	0	1

Name	Coverage (%)	Time (s)
○_ZN16miniqmcreference19DiracDeterminantRefIN11qmcplusplus13DelayedUpdateIddEEE14evaluateRatiosERNS1_18VirtualParticleSetERSt6vectorIdSaIdEE	0.01	0.01

Report Configuration

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

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