Level 1 Calo Software Meeting at RAL on Tuesday 9 January 2001.
---------------------------------------------------------------

Present: Murrough, Steve, Bill, Norman, Gilles.

The informal agenda was:

- Status of ROD work
- Status of test vector work
- TTC document
- Local controllers document
- Hardware purchases
- Plans for the year
- Report of a meeting at UCL
- Software organisation


ROD status
----------

Norman reported for Bruce (who was ill) on the ROD status.
Bruce had found some firmware problems with the readout code. These
are being fixed. There are also problems with the TTC: apart from
an embarassing error in the connector layout which can be bricolaged
there is another firmware problem in TTC commands to the ROD (or DSS?).
The ROD Bruce had is now back with Viraj and co.

At the recent Level1-Level2-Dataflow phone conference we stated that
the ROD-ROIB test wouldnt be before March...

The event fragment format for CPM ROI data is defined - it should be
circulated. Viraj and co are too busy to produce more documentation.

ACTION: Norman to assemble all eight different Glink and Slink data
formats going into and out of the RODs and produce a little document.

Norman had asked Viraj about mixing amd matching ROD formatting
algorithms on one ROD module. This is possible in a limited way:
the ROD has two formatting FPGAs each handling two channels. They
are both loaded from the same memory, so must run the same code,
but the formatting of each channel in the pair could be different.

Later in the meeting we looked again at ROD requirements for the
slice tests (and also for separate CP and JEP system tests) given
these firmware permutations:

                             N. RODs required for
Function          N.Chans    CP / JEP / Slice test
CPM slice data       4       1  /  0  /   1
CPM ROI data         4       1  /  0  /   1
JEM slice data       4       0  /  1  /   1
JEM ROI data         4       0  /  1  /   1
CMM CP+Jet slice    2+1      1  /  1  /   1
CMM Et slice+ROI    1+1      0  /  1  /   1

ie six RODs are needed for the slice test, each with different
firmware. For simultaneous CP and JEP system tests in different
places seven would be preferable and would allow a spare for the
slice tests.


Test vectors
------------

Bill has revamped his test vector generation program to allow greater
flexibility in generating different kinds of test vector, eg CPM slice
and ROI data. He needs the ROI event fragment format before he can
generate the expected ROD output data!

ACTION: Bill get this from Norman/Viraj

Bruce has been told where to get Bills software, not sure if he has
actually used it:

ACTION: Bill talk to Bruce about this...

Bill has been thinking about future extension/expansion. We should
probably discuss this in more detail at the next meeting. For some
purposes his present approach of generating all stages of data from
an input specification is OK. However an alternative of simulating
the output data given that modules input data may be preferable.
However this also requires knowing what thresholds etc will be loaded
in the CPM etc.

Who will do the work of generating test vectors for the JEM and other
modules?


TTC document
------------

Murrough has produced a few draft documents recently, among them one
on TTC addresses and broadcasts.
http://www.hep.ph.qmw.ac.uk/~landon/l1soft/docs/

There have been some comments from hardware people, none from software
community. Norman had independently come up with a TTCrx addressing scheme...

ACTION: Murrough & Norman agree and circulate a single proposal
        (an agreement was reached after the meeting)
        
On a related topic, we need to define exactly how each module will
derive its VME address from the geographic addressing scheme, ie how
large each modules address space will be.

ACTION: Norman to produce a short document on this


Run Control
-----------

Murrough circulated a draft document on our Local Controller(s) to the
software community. No feedback yet...
http://www.hep.ph.qmw.ac.uk/~landon/l1soft/docs/


The document is a first attempt at listing the actions needed for each
module in the various state transition of the run control system. So
far only for normal physics runs. Calibration runs need more thought.

One issue is whether any extra synchronisation between crates will be
required. The ideal case is no, each crate with its own local controller
can execute its state transitions independently (even though we need to
establish communication between modules in different crates). This will
be done via TTC broadcast commands (eg to get LVDS to lock). We need to
discuss with module designers whether TTC commands may interfere with
simultaneous VME access to different module functions.


Hardware purchases
------------------

As reported at the end of last year, Murrough has tested the new VME
board at QMW, integrated the VMELinux driver into HDMC and tested both
that and the Hannappel driver. All seems to work (bus errors are reported
only by the VMELinux driver; memory mapped access is only available via
the Hannappel driver).

Murrough has temporarily lost his VME crate to the marauding QMW SCT
group so wasnt able to test remote booting via ethernet.
Maybe David Botterill can be asked to get this going...?

We agreed that Norman will buy another three of these boards which should
be adequate for the slice tests while leaving one somewhere in the UK.

ACTION: Murrough to provide Norman with a specification.

We need to check what crate and disks we need

ACTION: Steve to make an audit at both RAL and Bham


Plans for the year
------------------

The ROD-ROIB test has priority. For this we need test vectors (and the
ROI formatting firmware).

CPMs are expected maybe by April/May. Some run control, database stuff
and more test vectors will be needed. Also monitoring programs...

Simulation of modules will require a description of their interconnections
(which is more than the minimal hardware configuration in the database
foreseen for the run controllers)

For the short term, Murrough will continue to work on run control and
databases, Bill on test vectors, Steve and maybe Gilles on monitoring
and histogramming etc. The Bham contingent is well placed to look after
both the inputs and outputs of the system. Bruce will hopefully continue
with readout, RODs etc.

Calibration is less of a priority.

We need to ensure that other groups can extend/implement their modules
in whatever framework we can provide. Needs more contact with them!!


Report from UCL meeting
-----------------------

Shortly before christmas, Murrough attended a small meeting at UCL with
Bob Cranfield (level2 & ROS), John Hill, John Butterworth and John Lane
(SCT), Barry Green (level2 & ROBins) amd Gordon Crone (ex-ATLAS level 2).

We discussed issues around the ROS in connection with both SCT and Level1
RODs and our short and long term requirements.

Gordon has some experience of the ROS software and may be able to look
into what would be involved in integrated readout of ROD event fragments
into the ROS - ie towards a DetDAQ solution which could be used to read
data from more than one ROD crate.

No great decisions were taken but we felt it was useful for the UK based
ROS developers and ROS users to keep in touch.


Software organisation
---------------------

At a previous meeting, Steve had outlined his experiences of software
development in industry. Murrough is writing a document with some
suggestions for how we should proceed: eg splitting into packages with
one person with overall responsibility for each package (also like the
Online group); organsation of code repository; use of Doxygen for code
documentation; also requiring a general document describing each package.

This document will be circulated for comment when its a bit less drafty!


Actions
-------

ACTION: Norman to assemble all eight different Glink and Slink data
        formats going into and out of the RODs and produce a little document.

ACTION: Bill get CPM ROI slice event fragment format from Norman/Viraj

ACTION: Bill talk to Bruce about use of his test vector program

ACTION: Murrough & Norman agree and circulate a single proposal
        for TTCrx addressing

ACTION: Norman to produce a short document on VME addressing

ACTION: Murrough to provide Norman with a specification for the
        Concurrent VME PC board

ACTION: Steve to make a crate and disk audit at both RAL and Bham


Next meeting
------------

Will be on Tuesday 30 January at RAL
CR3 (R61) in the morning, CR2 (R1) in the afternoon.