Unexpected overhead in test execution times when the number of test cases increases

[kaadhina]

Samples.zip

Description

Unexpected overhead in test execution times when the number of test cases increases. The increase is non-linear with the increase in number of tests.

Steps to reproduce

1. Execute a test run with one assembly containing 10k test cases => Takes ~ 10.65s
2. Execute a test run with one assembly containing 50k test cases => Takes ~ 165.87s

[Execute a test run with five assemblies each containing 10k test cases => Takes ~ 50.1s]

Expected behavior

It should take around 50s for the second case

[pvlakshm]

Is this represented in the perf workloads we use? If so, is this a regression?

[Evangelink]

There are still multiple performance issues with MSTest but this issue as-is is not really actionable so I will close it.