Re: gencgc performance and soft-real time

Daniel Barlow <dan@xxxxxxxxxx> writes:

> The only conservative collection going on is when it scavenges the
> call stack: pages pointed to by potential pointers (potentially
> pointed to?) are advanced to the next generation and marked
> dont_move so that they won't be relocated.

Before someone misinterprets that, let me add that such pages are
promoted to new-space, which in gencgc means their generation number
in the page table is set to the new-space generation number, which is
sort of an implementation trick.  It's not that objects on such pages
are tenured to the next older generation.

[...]

> The question mark is that - at least in SBCL, I haven't checked in
> CMUCL - purify can actually free more memory than a full gc does, so
> whatever extra it does needs to be looked at.

My guess would be that purify is more effective because it actually
does less than GC; it examines a lot less roots.