Stress Test on Large Data Bases

CASA Modification Request 20C108, November 2007

---

---

1. Introduction

Casa should run efficiently, especially for the very large data bases that EVLA and ALMA will produce.

[This is what I [BG] wrote in an email to Joe McMullin previously]

1. Take a bread and butter processing mode (say spectral line small mosaic) and time all its steps (visibility filling to conversion of output images to FITS) for ALMA size data (both input and output).

2. Compare timing w/ AIPS. Does not have to be exact, purpose is to look for any remaining gross serial performance problems (eat your vegetables before dessert), not to understand small (compared to total run time) differences here and there.

(It could well be one of our regression test already fulfills 1&2).

3. Analyze each step, indicating performance increases possibly available through a. embarassingly parallel; and b. parallel algorithm; developments.

That is, I see this large data investigation as being the precursor to defining our parallel development program, i.e. with the results in hand we can define where we get the most "bang per buck" for the next-stage parallel development.

2. Background

3. Requirements

A general requirement is that casa must handle large data bases with a speed that is similar to other packages.

4. Design

5. Deployment Checklist

6. Test Plan

Large simulated or real data bases should be read into casa and the passed through nominal editing, calibration and imaging tasks. The speed compared with AIPS is a useful guideline. .

6.1 Internal Testing

6.2 Sponsor Testing

Sponsor should be involved with the developer is choosing the appropriate data and running the tasks.

6.3 Integration/Regression Tests

At least one regression test should deal with a large data base.

6.4 Testing for Scientific Validity

---

Signatures

APPROVED: I acknowledge that my request is fully contained in this MR, and if the CASA development group delivers exactly what I specified, I will be happy.

ACCEPTED: I acknowledge that I have validated the completed code according to the acceptance tests, and I am happy with the results.

Written	- - - - -
Checked	- - - - -
Approved by Scientific Sponsor	- - - - -
Accepted/Delivered by Sponsor	- - - - -

Symbols:

• Use %X% if MR is not complete (will display )
• Use %Y% if MR iscomplete (will display )

---

Discussion Area

-- NicoleRadziwill - 30 Oct 2007

-- EdFomalont - 06 Nov 2007

-- EdFomalont - 07 Nov 2007