pyecsca.sca.stacked_traces.combine module

class pyecsca.sca.stacked_traces.combine.BaseTraceManager(traces)

Bases: object

Base class for trace managers

average()

Average traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

conditional_average(cond)

Average traces for which the condition is True, sample-wise.

Parameters:

• traces –

• condition –

Return type:

CombinedTrace

Returns:

standard_deviation()

Compute the sample standard-deviation of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

variance()

Compute the sample variance of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

average_and_variance()

Compute the sample average and variance of the traces, sample-wise.

Parameters:

traces –

Return type:

List[CombinedTrace]

Returns:

add()

Add traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

class pyecsca.sca.stacked_traces.combine.GPUTraceManager(traces, tpb=128, chunk=False, chunk_size=None, chunk_memory_ratio=None, stream_count=None)

Bases: BaseTraceManager

Manager for operations with stacked traces on GPU

Parameters:

• traces (StackedTraces) – Stacked traces on which to operate.

• tpb (Union[int, Tuple[int, ...]]) – Threads per block to use for GPU operations.

• chunk (bool) – Whether to chunk the traces.

• chunk_size (Optional[int]) – Number of samples to use for chunking. Chunks will be chunk_size x trace_count.

• chunk_memory_ratio (Optional[float]) – Part of available memory to use for chunking.

• stream_count (Optional[int]) – Number of streams to use for chunking.

static chunk_size_from_ratio(chunk_memory_ratio, element_size=None, item_size=None, chunk_item_count=None)

Return type:

int

property traces_shape: Tuple[int, ...]

average()

Average traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

conditional_average(cond)

Average traces for which the condition is True, sample-wise.

Parameters:

• traces –

• condition –

Return type:

CombinedTrace

Returns:

standard_deviation()

Compute the sample standard-deviation of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

variance()

Compute the sample variance of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

average_and_variance()

Compute the sample average and variance of the traces, sample-wise.

Parameters:

traces –

Return type:

List[CombinedTrace]

Returns:

add()

Add traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

pearson_corr(intermediate_values)

Return type:

CombinedTrace

run(func, inputs=None, output_count=1)

Return type:

Union[CombinedTrace, List[CombinedTrace]]

class pyecsca.sca.stacked_traces.combine.CPUTraceManager(traces)

Bases: object

Manager for operations on stacked traces on CPU.

traces: StackedTraces

average()

Compute the average of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

conditional_average(condition)

Compute the conditional average of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

standard_deviation()

Compute the sample standard-deviation of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

variance()

Compute the sample variance of the traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

average_and_variance()

Compute the average and sample variance of the traces, sample-wise.

Parameters:

traces –

Return type:

List[CombinedTrace]

Returns:

add()

Add traces, sample-wise.

Parameters:

traces –

Return type:

CombinedTrace

Returns:

pearson_corr(intermediate_values)

Calculates the Pearson correlation coefficient between the given samples and intermediate values sample-wise.

The result is equivalent to:

>>> np.corrcoef(self.traces.samples,
                intermediate_values,
                rowvar=False)[-1, :-1]

but a different implementation is used for better time-efficiency, which doesn’t compute the whole correlation matrix.

Parameters:

intermediate_values (npt.NDArray[np.number]) – A 1D array of shape (n,) containing the intermediate values.

Return type:

CombinedTrace