Subject: Re: Compiler timings on varous MVII NetBSDs etc.
To: Lord Isildur <mrfusion@umbar.vaxpower.org>
From: Matt London <matt@knm.yi.org>
List: port-vax
Date: 01/26/2001 18:37:09
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

On Fri, 26 Jan 2001, Lord Isildur wrote:

> On Fri, 26 Jan 2001, David Brownlee wrote:
> 
> > 	Could you send the program to the list (or make it available),
> > 	so people with identical hardware and different OS versions can
> > 	get exact numbers?
> > 
> > 		David/absolute		-- www.netbsd.org: No hype required --
> > 
>
> sure. the source is at the bottom here. it's pretty decent at beating
> a machine's memory to benchmark memory speed. run it with a large
> size to effectively fault every cached access and test the speed of
> core.. and dont laugh at my code too much :-)
> 
> the argument tells it how many words of memory to devote to the sieve,
> and it puts a bunch of not veryuseful dots on stderr, and optionally prints
> the primes on stdout (a few extra numbers sneak in, i never bothered to 
> fix  it ;) , to shut off the printing, compile with -DSILENT

> unsigned int* field,foo,bar,blarf,*mark,*space,m,x,*b,c,snum,bnum;
> /* bar will mark the current factor. blarf is the marker for
>  * the current bit. mark will mark the current word.
>  * m speeds up finding blarf's bit, while blarf is used for masking
>  * the actual bit. 
>  */
> if(argc<2){puts("Are you asking me how many bananas per square furlong there\nare in Costa Rica?\n");exit(0);}

You would be shot at my uni for writing code like that! *grin*

Not the most readable code in the world, but I guess it works (I've not
tried it yet). I'll run it up on my 8M MV3100 m10e, GENERIC kernel (I need
to do a new kernel too), and grab some times :&)

*grins* Today's been kinda neat, got my last exam out the way (Formal
Specification - hence the comment on the code ;&) and it went well, got to
talk to my g/f, which was nice, cause I didn't get to speak to her last
night (long story - she lives in canada, I'm in the UK), then isildur's
code gave me something else to grin about. All in all, not a bad day at
all :&)

- -- Matt

- ---
E-mail:
  matt@pkl.net, matt@knm.yi.org, matt@printf.net
  matt@m-techdiagnostics.ltd.uk, matthew.london@stud.umist.ac.uk
  mattl@vcd.student.utwente.nl, mattl@fnmail.com, mlondon@mail.talk-101.com
Web Page:
  http://knm.yi.org/
  http://pkl.net/~matt/

PGP Key fingerprint = 00BF 19FE D5F5 8EAD 2FD5  D102 260E 8BA7 EEE4 8D7F
PGP Key http://knm.yi.org/matt-pgp.html

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.0 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE6ccPjJg6Lp+7kjX8RAqrYAJ9+1mCxRaRt2i7hRttYrF8MXhBCQQCeNlAE
NHK/Suz3oTNXK0BP6BjOuLc=
=EB6k
-----END PGP SIGNATURE-----