Path: ...!weretis.net!feeder8.news.weretis.net!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Vir Campestris <vir.campestris@invalid.invalid>
Newsgroups: comp.lang.c++
Subject: Re: OT: Re: Sieve of Erastosthenes optimized to the max
Date: Mon, 12 Aug 2024 15:32:33 +0100
Organization: A noiseless patient Spider
Lines: 28
Message-ID: <v9d6e1$3avh0$2@dont-email.me>
References: <ul41d4$2koct$1@raubtier-asyl.eternal-september.org>
 <utn1fp$3oeks$2@raubtier-asyl.eternal-september.org>
 <utng4n$3rn1f$2@dont-email.me>
 <utoh9d$6lrr$1@raubtier-asyl.eternal-september.org>
 <utq0ag$hvrl$3@dont-email.me>
 <utq0os$ibqn$1@raubtier-asyl.eternal-september.org>
 <utq11p$icmm$1@dont-email.me> <v25c87$1ld9m$1@dont-email.me>
 <86r0duwqgg.fsf@linuxsc.com> <v39o3l$1lvju$1@dont-email.me>
 <86o78mpnlf.fsf@linuxsc.com> <v3fv2u$2ursd$1@dont-email.me>
 <86ttibod7n.fsf@linuxsc.com> <86h6eaoi2r.fsf@linuxsc.com>
 <v4sool$1grge$1@dont-email.me> <867celixcw.fsf@linuxsc.com>
 <v5sg9s$mat4$1@dont-email.me> <86zfr0b9hw.fsf@linuxsc.com>
 <v638ud$25623$1@dont-email.me> <861q3o5do1.fsf@linuxsc.com>
 <v7tdts$29195$1@dont-email.me> <868qx45v5g.fsf@linuxsc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Mon, 12 Aug 2024 16:32:33 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="ba62fd7f861f01396a4b4b18eeb110c0";
	logging-data="3505696"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX19tCZ8mosOnuEcAHFdQciw30zWa0+Rw+wg="
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:DMkn8+lvl1NfmL5Kd+pVB7M30z8=
Content-Language: en-GB
In-Reply-To: <868qx45v5g.fsf@linuxsc.com>
Bytes: 3026

(replying to two messages at once)

On 10/08/2024 15:07, Tim Rentsch wrote:
>> I suspect the cost of extracting the divided value from the 7
>> bytes will be prohibitive.  You may find it's best to have one
>> table for the N/30 parts (big entries, word aligned) and another
>> for the N%30 part (small entries, byte aligned).  BICBW.
> Two tables, absolutely.  What does BICBW stand for?

But I Could Be Wrong. I thought that one was fairly well known.

On 11/08/2024 08:00, Tim Rentsch wrote:
 > Note that the %30 tables can be shared.  There are 8 of them, one
 > for each of the eight residues mod 30 of the original prime.

I started coding that up. Then it occurred to me that there wasn't going 
to be much saving by storing a pointer (8 bytes on my machine) to the 
correct 8-byte long table. I suppose I could store a byte array index 
instead, but I've gone off in other directions.

You've picked up my cache management tweak.

I'll read through your code. But it might not be today. We've got the 
hottest day of the year here, and I don't have aircon in my garden 
office. In winter the computer and me are enough to keep it warm, but 
right now it's getting warmer by the minute.

Andy