[RC5] Losing lot of time with a computer crash

Andreas Beckmann andreas.beckmann at student.uni-halle.de
Wed Jan 10 23:04:24 EST 2001


-----Original Message-----
From: Ben Brockert <benb at iastate.edu>
To: rc5 at lists.distributed.net <rc5 at lists.distributed.net>
Date: Wednesday, January 10, 2001 10:20 PM
Subject: Re: [RC5] Losing lot of time with a computer crash


>Aaron Nielsen wrote:
>Anybody know how much this hurts the performance? It must to some
>degree. I started using it on my old slow computer that crashed a lot
>(crashing at 75% of an OGR block sucks) and I've kept doing it, even
>though now the RC5 blocks always take less than the equivalent of three
>minutes. Would it be more efficient to turn this off when I'm roasting
>through RC5, but on when I'm trying to stay #1 on my team in OGR?

Checkpoint writing is done if either
- 10% of the packet has been done (if percentage is measurable, e.g. RC5) or
- 10 minutes since last checkpoint have elapsed
but not earlier than 30 seconds after last checkpointing.

A <3 minutes RC5 packet costs you:
load_packet
7 checkpoint writes (create, 5* refresh, clear)
save_packet

Is this too much?

But if your RC5 packets last less than 3 minutes you probably should check your preferred block size setting. Using a small blocksize causes more overhead (network traffic, load/save often overhead). All current clients support a preferred blocksize up to 33 (32*2^28 keys).


Andreas
andreasb at distributed.net

-- 
CPU time is a terrible thing to waste!
Join the world's fastest computer - http://www.distributed.net - Team #11292


--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list