Unoptimized builds are more likely to introduce high variations on runtime. We still want to run them to catch problems with the microbenchmark code itself, but we don't want to create BFs based on performance regressions on these variants.
- is related to
-
SERVER-62180 Enable signal processing (or red/green) checks for Google Microbenchmark
-
- Closed
-
- related to
-
SERVER-65359 Don't run benchmarks on DEBUG/*SAN builds
-
- Closed
-
-
SERVER-87396 Add new resmoke hook to check result of a benchmark
-
- Closed
-
-
SERVER-65986 Disable threshold checks for highly variable benchmarks
-
- Closed
-