Re: Performance of string deserialisation
- From: "drewnoakes" <drewnoakes@xxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: Tue, 6 Sep 2005 08:57:13 -0700
Hi Helge,
Thanks for your posts.
I like this idea of calculating a hash value from the byte sequence and
using that as a key in some sort of dictionary. I can see a way in which
that would work where I wouldn't need to allocate any objects unless the
string doesn't yet exist. I'll think about that for a while...
I'd previously been stuck on a different idea. I guess it's the same
concept as the prefix tree you mention -- a structure where common subsets
form nodes in a tree which can be traversed top-down when deserialising to
reach the string value at a leaf.
Anyway, I've a few things to try now and it seems to be quite challenging.
Thanks again.
.
- References:
- Performance of string deserialisation
- From: drewnoakes
- Re: Performance of string deserialisation
- From: Helge Jensen
- Re: Performance of string deserialisation
- From: Helge Jensen
- Performance of string deserialisation
- Prev by Date: Massive Concurrency / Multiple ThreadPools
- Next by Date: Re: Performance of string deserialisation
- Previous by thread: Re: Performance of string deserialisation
- Next by thread: .NET performance counters showing more than 100% network utilization
- Index(es):
Relevant Pages
|