Re: hash table sizes

[Jes Sorensen <jes@wildopensource.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=ascii, Size: 825 bytes --]

>>>>> "Jörn" == Jörn Engel <joern@wohnheim.fh-wedel.de> writes:

Jörn> [pruned CC: list]
Jörn> On Fri, 28 November 2003 13:35:36 -0600, Jack Steiner wrote:
>> You proposed above to limit the allocation to the amount of memory
>> on a node.

Jörn> Jes didn't _limit_ the allocation to the memory on a node, he
Jörn> _based_ it on it, instead of total memory for all nodes.
Jörn> Therefore a 1024 node NUMA machine with 2GB per node has no
Jörn> bigger hash tables, than a single CPU machine with 2GB total
Jörn> memory, however big that may be.

Jörn> Unless I didn't understand his patch, that is. :)

Yep, thats exactly what my patch did. There's a gotcha if node 0 has a
lot less memory than the remaining nodes, but I also suspect the size
of those hash tables was already out of whack as memory has gone up.

Thanks,
Jes