[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Chicken-hackers] Experiments with heap shrinkage, -:hs
From: |
Kooda |
Subject: |
Re: [Chicken-hackers] Experiments with heap shrinkage, -:hs |
Date: |
Sat, 20 May 2017 13:11:50 +0200 |
On Fri, 19 May 2017 15:29:56 +0200 (CEST) Sven Hartrumpf
<address@hidden> wrote:
> Hi all.
>
> I would like to experiment with heap shrinking (-:hs) for long
> running, memory/GC intensive processes.
>
> I checked runtime.c and I read the following shrink condition:
>
> count < percentage(percentage(heap_size, C_heap_shrinkage),
> DEFAULT_HEAP_SHRINKAGE_USED)
>
> This would imply that heap usage must be < 12.5 % to lead to a heap
> shrink because C_heap_shrinkage (via DEFAULT_HEAP_SHRINKAGE) is 50 %
> and DEFAULT_HEAP_SHRINKAGE_USED is 25 %.
> This (< 12.5 %) is very unlikely for many applications :-)
> How about:
>
> count < percentage(heap_size, DEFAULT_HEAP_SHRINKAGE_USED) ?
>
> Ciao
> Sven
>
My 2 cents, I don’t know much about CHICKEN’s GC yet:
This sounds good to me, the current behaviour seems a little bit
strange.
Maybe we could also introduce a new runtime flag that would indicate at
which point we trigger heap shrinkage (that would replace
DEFAULT_HEAP_SHRINKAGE_USED)?