Mips vs specint95 for Hall D CDR and white paper

[Elliott Wolin <wolin@jlab.org>]

Hi Dave,

Ian tells me that the 750 MHz PIII's give 35 specint95.  Mark and Stepan tell me
it takes about 100 msec to fully reconstruct CLAS events on a 300 MHz farm node
(12 specint95, according to Ian).

Thus CLAS full reconstruction takes 1.2 specint95-sec;  divide by 10 for Hall D
online filtering, then x2 for safety (includes i/o overhead) (as in chapter 7),
gives about 0.25 specint95-sec to filter.  (do we still want to use the factor
of 10 for filtering?).

If we assume a factor of 10 improvement over 750 MHz for Hall D then a single
processor will deliver 350 specint95, or a single processor can filter 1400 Hz
of events.  Duals will process 2800 events/sec. 

Thus the 180 kHz event rate means we'll need about 64 dual processors for the
level 3 farm.  Alternatively 180 kHz x 0.25 specint95-sec means we'll need 45000
specint95 equivalents.

Chapter 7 needs to be reworded in these terms...I'll rewrite that section if you
want.

				Sincerely,
					Elliott
 

================================================================================

   Help feed the hungry, one click at a time:  http://www.thehungersite.com

    "Science plays a major role in satisfying the minor needs of mankind"

================================================================================
begin:vcard 
n:Wolin;Elliott 
tel;fax:757-269-5800
tel;home:757-229-2724
tel;work:757-269-7365
x-mozilla-html:FALSE
org:Jefferson Lab;Physics Department
version:2.1
email;internet:wolin@jlab.org
title:Physicist
note:A218 CEBAF Center
adr;quoted-printable:;;12000 Jefferson Ave MS12H=0D=0A;Newport News;VA;23606;US
x-mozilla-cpt:;21408
fn:Elliott Wolin
end:vcard