One layer's accumulated time-per-phase, in seconds – see the file-level note for the reconciliation invariants (PhaseSum() vs. the top-level totals vs. external wall-clock).
More...
#include <Profile.h>
|
| double | PhaseSum () const noexcept |
| | Sum of every sub-phase timer (excludes the top-level calculateStateTotal/updateStateTotal, which measure the same work and would double-count if included – see the inline note).
|
| void | Print (const std::string &label) const |
| | Prints this accumulator's per-phase breakdown (with percentages of PhaseSum()) plus the top-level/sub-phase reconciliation check, to stdout.
|
One layer's accumulated time-per-phase, in seconds – see the file-level note for the reconciliation invariants (PhaseSum() vs. the top-level totals vs. external wall-clock).
◆ PhaseSum()
| double Deep::ProfileAccumulator::PhaseSum |
( |
| ) |
const |
|
inlinenoexcept |
Sum of every sub-phase timer (excludes the top-level calculateStateTotal/updateStateTotal, which measure the same work and would double-count if included – see the inline note).
◆ Print()
| void Deep::ProfileAccumulator::Print |
( |
const std::string & | label | ) |
const |
|
inline |
Prints this accumulator's per-phase breakdown (with percentages of PhaseSum()) plus the top-level/sub-phase reconciliation check, to stdout.
- Parameters
-
| label | Heading printed above this accumulator's rows, e.g. a layer's shape and identity (see LayerProfile()). |
◆ activationDerivative
| double Deep::ProfileAccumulator::activationDerivative = 0.0 |
UpdateState: activation derivative.
◆ activationForward
| double Deep::ProfileAccumulator::activationForward = 0.0 |
CalculateState: forward activation function.
◆ biasAdd
| double Deep::ProfileAccumulator::biasAdd = 0.0 |
CalculateState: bias-add after the forward GEMM.
◆ bottomUpConstruction
| double Deep::ProfileAccumulator::bottomUpConstruction = 0.0 |
UpdateState: assembling the bottom-up term.
◆ calculateStateCalls
| long long Deep::ProfileAccumulator::calculateStateCalls = 0 |
Number of CalculateState() calls recorded.
◆ calculateStateTotal
| double Deep::ProfileAccumulator::calculateStateTotal = 0.0 |
CalculateState(), whole-call total.
◆ energyCalculation
| double Deep::ProfileAccumulator::energyCalculation = 0.0 |
CalculateState: the energy reduction itself.
◆ errorConstruction
| double Deep::ProfileAccumulator::errorConstruction = 0.0 |
CalculateState: building the error term.
◆ feedbackGemm
| double Deep::ProfileAccumulator::feedbackGemm = 0.0 |
UpdateState: the feedback GEMM.
◆ forwardGemm
| double Deep::ProfileAccumulator::forwardGemm = 0.0 |
CalculateState: the forward prediction GEMM.
◆ localErrorTerm
| double Deep::ProfileAccumulator::localErrorTerm = 0.0 |
UpdateState: dz_dt = -p*e (was ownGradientTerm – not a backprop gradient).
◆ saxpyUpdate
| double Deep::ProfileAccumulator::saxpyUpdate = 0.0 |
UpdateState: the final z += ir*dz SAXPY.
◆ updatePrecisionTotal
| double Deep::ProfileAccumulator::updatePrecisionTotal = 0.0 |
UpdatePrecision(), whole-call total.
◆ updateStateCalls
| long long Deep::ProfileAccumulator::updateStateCalls = 0 |
Number of UpdateState() calls recorded.
◆ updateStateTotal
| double Deep::ProfileAccumulator::updateStateTotal = 0.0 |
UpdateState(), whole-call total.
◆ updateWeightsTotal
| double Deep::ProfileAccumulator::updateWeightsTotal = 0.0 |
UpdateWeights(), whole-call total.
The documentation for this struct was generated from the following file: