Here we'd like to collect the metrics for the platform scalability. Don't limit your thoughts on the already created entries, maybe we will work smth. absolutely different out!
A possible template might be:
Variable: what is a changing parameter by measuring
Fixed: what remains fixed
Metric: measurable value (mainly execution time, but maybe smth. else)
Boundary Limits: approximate boundary limits, if any can be foreseen.
Ways to achieves: how we can achieve the best Limits values
The platform scalability definition |
|||||||
No. |
Scalability |
Variable |
Fixed |
Metric |
Boundaries |
Ways to achieve |
Came from (not obligatory) |
1 |
Strong scaling by growing number of computing cores |
No. of cores |
Total problem size |
Workflow execution time |
- |
data streaming, distribution, parallelization |
Alexey |
2 |
Weak scaling by growing number of computing cores |
No. of cores |
Problem size per core |
Workflow execution time |
- |
data streaming, distribution, parallelization |
Alexey |
3 |
Scaling to the size of the problem |
Problem size |
Number of Processors |
Workflow execution time |
- |
Lower algorithm complexity |
|
4 |
Scaling to the complexity of the problem |
Problem complexity |
Number of Processors, Problem size |
Workflow execution time |
- |
Lower algorithm complexity |
|