Re: [boost] [hash] regular behaviour of hash function for double values

Subject: Re: [boost] [hash] regular behaviour of hash function for double values
From: Kai Schroeder (kaischroeder3_at_[hidden])
Date: 2012-01-31 11:19:28

I have found that this behaviour of boost hash is described in the
documentation:
http://www.boost.org/doc/libs/1_48_0/doc/html/hash/rationale.html

So, this is by design. This rationale is understandable, however it
can lead to difficult to find performance problems as in our case.
Btw.: std::hash in MSVC10 does not show this behaviour. Well, I tend
to agree that this is a bug in TBB, still I would personally prefer a
better hash function (even if it is a bit slower) by default. Maybe a
fast_hash option could be provided as an alternative.

Best regards,
Kai

On 1/31/12, Olaf van der Spek <ml_at_[hidden]> wrote:
> On Tue, Jan 31, 2012 at 4:10 PM, Kai Schroeder
> <kaischroeder3_at_[hidden]> wrote:
>> Especially since some hash table implementations (e.g. intel's
>> threading building blocks) only use the last few bits to determine the
>> bucket, this can lead to very bad caching behaviour for the rather
>> common case of regularly sampled points (equidistant samplings of
>> power of 2 are e.g. common in computer graphics and numerical
>> algorithms).
>>
>> Are you aware of this? Would you consider it a bug?
>
> Yes, sounds like a bug in TBB. I don't think it's a good idea to discard
> bits.
>
> Olaf
>
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>