[issue39143] Implementing sub-generation steps in the gc

[Joannah Nanjekye]

Joannah Nanjekye <nanjekyejoannah at gmail.com> added the comment:

> The most common reason is that when promotions of the youngest generations happen, 
>some very young objects that just arrived in the generation are >promoted because we have >reached a threshold, and its death will be >delayed.

What threshold is this? Is it based on the number of objects in the young generation in that if the number of objects in the young generation reaches some value, then promotion occurs. If this is the case, we can use another scheme for the start before exploring steps IMHO.

Using Shaw’s Bucket Brigade Scheme  of semi-spaces or generational steps is good but this means a generation with n steps guarantees n scavenges before an object is promoted to the next generation. The surviving objects are copied between these pairs of semispaces b times before they are promoted to the next step. An n-bucket scheme guarantees that objects will be promoted to the old generation If they have survived between nb and nb-1 scavenges. For example, a 2 bucket scheme, objects are copied up to three times before being promoted to the next generation. This method is good but we will still incur the overhead of copying objects between the semi-spaces.

I have looked at a similar problem before for a different runtime and instead think to consider basing promotion on the number of collections survived by objects in the young generation. This requires using five bits from one of two header words of each object to record its age. The objects that reach the collection threshold are promoted to the old generation. This will give the young objects enough time to die thereby reducing how often objects are promoted to the old generation. This in turn reduces the frequency of major collections hence reduced pause times.

I would only consider the generational semi-spaces after we have explored basing promotion on the age of an object in a generation. If this doesn't work, then we can use the bucket brigade scheme.

----------
nosy: +nanjekyejoannah

_______________________________________
Python tracker <report at bugs.python.org>
<https://bugs.python.org/issue39143>
_______________________________________