Updated my post above.
But the problem might remain: On what time scale does it spike up that much? the GC won't do "every 10ms" cleanups and the like, which is the reason this kind of tests normally totally break as they don't test the GC in a real circumstance for which a GC is actually geared.
But the problem might remain: On what time scale does it spike up that much? the GC won't do "every 10ms" cleanups and the like, which is the reason this kind of tests normally totally break as they don't test the GC in a real circumstance for which a GC is actually geared.