Skip to content

Conversation

@jschmieg
Copy link
Contributor

…zation of Node child

Copy link

@gjerecze gjerecze left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The performance drop is significant: from 264k to 132k on a single-threaded put. Simple test without drain gives 171k.

// Temporarily disable the node
nodeLeafCompressed_new->key = -1;
pmemobj_flush(_pm_pool.get_handle(),
&nodeLeafCompressed_new->depth, 2 * sizeof(int));

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why 2*sizeof(int)?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

To flush depth and type. Now I changed to separate "flushes".

const int PREALLOC_LEVELS = 1;
// size of table for actions for each Node
#define ACTION_NUMBER_NODE256 (1 + 256)
#define ACTION_NUMBER_NODE256 (1 + 256 + 2 * 256)

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we add defines for the magic numbers? It's getting harder to understand which one is used for what.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done

@jschmieg jschmieg requested a review from gjerecze January 18, 2019 07:54
Copy link

@gjerecze gjerecze left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We have further drop in performance to 120K.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants