🚀 go-pugleaf

In article <1993Jan22.172623.15644@oracle.pnl.gov>, jordan@warped.pnl.gov (Randy
Jordan) writes:
|>In article <C19M1B.KHM@dscomsa.desy.de>, hallam@zeus02.desy.de (Phill
|>Hallam-Baker) writes:
|>|>In article <1993Jan20.001847.1807@eco.twg.com>, reece@eco.twg.com (Reece R.
|>|>Pollack) writes:
|>|>
|>|>
|>|>Each of our events is approx 100Kb
|>|>
|>|>We take 10 million of them per second.
|>|>
|>	1MB per second?

No, !00Kb times 10 MHz equals 1 Terabyte /sec

|>|>Even with filtering we take Terabytes of data per month.
|>|>
|>	Wrong...you take in 3.6TB per hour at the rate stated!

No, with filtering, we reject down to a rate of 5 Hz. Also thats 3.6 Eta-bytes
per hour, not Terrabytes.

If you are interested I can email one of our papers. But basicaly its a three
stage trigger running at 1KHz, 100Hz and 5Hz triggers, data transport volume is
minimized by data locality, geometric and pipelined parallelism are employed and
the whole shooting match cost over $50 Million just for the electronics.

This is why we need VMS and can't use UNIX!

|>|>2Gb is small beans, we will be getting machines with that sort of memory
|>within
|>|>the next few years. The successor machine to ZEUS will be producing 2 GB
|>|>events.
|>|>
|>	2GB is image size NOT dbms size.

Would it give the game away unduly if I said that I fail to make the
distinction? We already have code systems in excess of 1 million lines of
FORTRAN - and those were written by hand. The next generation of systems will be
much much larger and involve much larger amounts of code.

We already have one system that uses a five level compiler scheme, ten pages of
source code balloon to tens of thousands of lines of compiler code. The data is
the specification and the specification is the program.


--

Phill Hallam-Baker