trafficserver-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil Sorber <p...@sorber.net>
Subject Re: RamCacheCLFUS
Date Mon, 18 Feb 2013 23:07:32 GMT
On Mon, Feb 18, 2013 at 4:59 PM, Phil Sorber <phil@sorber.net> wrote:

>
> On Mon, Feb 18, 2013 at 4:41 PM, Tomasz Kuzemko <tomasz@kuzemko.net>wrote:
>
>> Did not look at the code, but seems like it is part of some general hash
>> table implementation.
>>
>>
> It is for a hash table AFAICT.
>
>
>> Good explanation why these are primes:
>>
>> http://stackoverflow.com/**questions/1145217/why-should-**
>> hash-functions-use-a-prime-**number-modulus<http://stackoverflow.com/questions/1145217/why-should-hash-functions-use-a-prime-number-modulus>
>>
>>
> Ah, that is very interesting and makes good sense. I will add a comment.
>
>
Upon further inspection it seems like this is duplicated code in both
RamCache implementations. Perhaps this should be abstracted out and added
to libts. Or replaced with a libts implementation if it already exists.


> Thanks.
>
>
>> Best regards,
>> Tomasz Kuzemko
>> tomasz@kuzemko.net
>>
>> W dniu 18.02.2013 20:50, Phil Sorber pisze:
>>
>>  In looking through RamCacheCLFUS I came upon this:
>>>
>>> static const int bucket_sizes[] = {
>>>    127, 251, 509, 1021, 2039, 4093, 8191, 16381, 32749, 65521, 131071,
>>> 262139,
>>>    524287, 1048573, 2097143, 4194301, 8388593, 16777213, 33554393,
>>> 67108859,
>>>    134217689, 268435399, 536870909, 1073741789, 2147483647
>>> };
>>>
>>> Looking through git blame and JIRA it looks like it was part of the
>>> initial
>>> patch for CLFUS. In talking with James in IRC he pointed out that it
>>> looks
>>> like the largest prime less than a power of 2. This seems like a very
>>> deliberate choice, but I can't figure out why and there are no docs
>>> explaining. Can someone, perhaps John, shed some light on it?
>>>
>>> Thanks.
>>>
>>>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message