Module adapton_lab::labdef
[−]
[src]
Defines lab parameters LabParams and LabDef, the parameters
for running the test diagram from the Adapton Lab
README.
Structs
| EngineMetrics |
For each engine, for each sampled subcomputation, we record the real time (in nanoseconds) and engine-based counters for DCG costs. |
| EngineSample |
To sample a single engine, we record metrics for processing the
input (left vertical edge in |
| GenerateParams |
Parameters for generating and editing input; See |
| LabDef |
lab definition: generic notion of an incremental computation
that can be evaluated and tested. We instantiate this structure
once for each test in our test suite. We implement the |
| LabParams |
Parameters to running a single lab experiment. |
| LabResults |
The result of a lab is a sequence of samples. |
| Sample |
The experiment consists of a loop over samples. For each sample, we switch back and forth between using the Naive engine, and using the DCG engine. We want to interleave this way for each sample in order to compare outputs and metrics (counts and timings) on a fine-grained scale. |
| SampleParams |
Parameters for collecting a single sample. In addition to these parameters, the experiment maintains a Rng based on the input_seeds, below; this Rng is given to Edit::edit to generate psuedo-random edits, in batches. For each engine, this Rng is sequenced across successive samples. Given an input_seeds vector, there is one unique Rng sequence for each engine's sequence of samples. |
Enums
| NominalStrategy |
A bit that controls how names are placed in the input; See |
Traits
| Compute |
Generic notion of a computation to run naively and incrementally.
It has specific |
| ComputeDemand |
Like Compute, but also provides a |
| Edit |
Generic process for editing an input randomly, in a stateful sequence of edits.
See |
| Generate |
Generic method for generating a random input.
See |
| Lab |
lab: Abstracts over parts of a lab definition of type |