S-LINK News 2001

COMPASS has taken 28 TByte of data over S-LINK

5 November 2001. Compass has finished data taking for 2001. Typically beam intensities of greater than 2*10^8 muons/spill (design value) were used. The 25 ns period is now used for systematic studies of the DAQ and some detectors. The last two weeks of data taking were particularly successful. All detectors promised in the May meeting of the SPSC were installed and operational.

The DAQ operated reliably at 17 000 events/spill, 30 kbyte/event. In total 14 TByte of data were taken with the complete 2001 spectrometer, while
about the same amount of data was taken earlier this year with an incomplete spectrometer. Data was taken over 85 CATCH read-out driver modules connected via ODIN S-LINKs to COMPASS Read-out buffers of which several are installed in a PC. Those PCs in turn are read out over Gigabit Ethernet.

The analysis of the data has started. COMPASS is set for physics in 2002.

Quizz answer: 60 cards for COMPASS

27 August 2001. The picture below presents 60 ODIN cards for the COMPASS experiment. Mike Deliman from Windriver Systems had the most accurate answer. He kindly suggested to donate his prize (an S-LINK resque kit) to physics research. In the meantime COMPASS has received another 36 cards which will be installed in September.

Quizz: what is this? ==>

4 July 2001. Tell us what the picture shows (what it is, how many) and what the stuff will do soon. The person sending in the most accurate and original answer will receive a free S-LINK resque kit containing: twenty spacers, fourty 2.5 mm screws and a airspray to clean fibre optics. Answers should be sent to Erik.van.der.Bij@cern.ch before July 28, 2001.

CERN Stores: 115 cards ordered in three months time; another 96 for COMPASS

8 June 2001. Since March this year HEP users can buy all S-LINK equipment directly from the CERN stores. Until now, the experiments have ordered 80 different cards (links and test tools) from the Stores, while the Stores has ordered 115 cards from the manufacturing companies. Apart from those orders, the COMPASS experiment has ordered 96 cards.
Although you can buy now from the Stores, you still should order a long time in advance. To reduce costs, only one or two cards of each type will be kept in stock, while the delivery times from the manufacturing companies may be three months or more. We are in discussion with the companies about shortening the delivery times.

ATLAS Tilecal receives S-LINK VME64x transition module

7 June 2001. The University of Geneva built the ATLAS Liquid Argon read-out driver (ROD) module in VME64x format. This module will receive the data from the front-end links over a transition module, a module that plugs in at the back of the crate. The Liquid Argon detector can use for this either the passive S-LINK transition module (with two S-LINK inputs and one output), or another module that has a custom input and one S-LINK output.
With another, active transition module, the ATLAS Tilecal detector can use the same ROD as the one built for the Liquid Argon detector. This active transition module, dubbed TM4plus1, contains FIFO buffers for the four input links and can reformat the incoming data so that it becomes suitable for the ROD. The output link is an ODIN link integrated on the board. Three boards have been built and are sent to the Universidad de Valencia where the Alteras will be programmed.

COMPASS receives batch of 60 ODIN link cards; 36 more to come

7 June 2001. The COMPASS experiment has received a batch of 60 Single ODIN link cards (for thirty links) for the readout system. Those cards are of the 128 MByte/s version. COMPASS will receive soon another 36 Double ODIN link cards that make up eighteen 160 MB/s links. COMPASS has used in the previous year fourty FCS-LINKs and now upgrades to the faster and cheaper ODIN links.

ATLAS integration tests of Level-1 and Dataflow successful

19 April 2001. In the three weeks starting from 26 March and ending on 12 April just before Easter, integration tests of Level-1 and Dataflow have been carried out. The following tests have been performed: MUCTPI -> ROS; MUCTPI -> ROIB; CPROD -> ROIB and CPROD -> ROS. Please contact Ralf Spiwoks for details about the results of the test.
Apart from some problems with reset and flow control as some devices under test (not the link cards) didn't implement those protocols correctly, the tests have been succesful. When you design boards that use S-LINK, you should pay particular attention to the reset protocol and flow control. It's simple to use, but that in fact may be the reason that during the design not enough attention is paid to it. If you like, you may contact Erik van der Bij who will review the design for you.

S-LINK equipment available from CERN Stores catalog 

19 March 2001. If you have a CERN budget code and want to  S-LINK equipment, don't look further. All links, test tools, PCI and PMC interfaces are now available from the CERN Stores. S-LINK is now as easy to get as an office chair, inkjet cartridge or a resistor.

Now it's in the Stores, you can order S-LINK equipment coming from three different countries in just a single Material Request placed via EDH . There is no need to pass via the S-LINK team to place your orders or to send out external orders.This will ease your own tracing of the orders as well. Ordering via the Stores will liberate time of the S-LINK team members so they'll have more time to help you in deciding which equipment you need or to review designs that use S-LINK.

You can consult the list of equipment available from the CERN Stores, you can go in the Stores Catalog to group 07.87 or search for common terms like SLINK, ODIN, SLIDAS, SSP or PCI.

ATLAS LArg ROD sends data to ROB

2 February 2001. The ATLAS Liquid Argon Read-out Driver has been able to send data over an S-LINK to a RIO2. This setup is very similar to that what can be used in the final ATLAS system. The data to the ROD was generated by a SLIDAS data generator mounted on an S-LINK VME64x transition module, processed by the ROD and then sent out over an ODIN Link Source Card that was mounted on the same transition module. Trigger rates upto 30 KHz could be handled with an event size of 1.2 KByte. Only the LArg front-end link should be connected to this setup to have a full slice of the data acquisition system operating.
 During the tests the following S-LINK boards have been used: SLIDAS (for data generation as no front-end input is available), S2VME64X transition module (to get the data into the ROD), ODIN (link). For debugging a SLITEST with a SLIDAD (for having a dead-end on the link) and a SLIBOX (for spying on data sent) have been used.

S-LINK64 extension specification

10 January 2001. S-LINK allows data transmission speeds of up to 160 MByte per second. This is limited by the specification of a 32-bit wide data channel that may run up to 40 MHz. This speed is a nice match to gigabit serialisers and optical transceivers that are reasonably cheap because of the Gigabit Ethernet and Fibre Channel markets. This speed may not match the requirements of all detectors in future high-energy physics experiments. The CMS experiment is the first one who may need transmission speeds of up to 800 MByte per second. As it is logical to extend the S-LINK specification to a higher speed by going wider and faster, together with CMS the S-LINK team has written the S-LINK64 extensions specification.

This S-LINK64 extension specification has to be used in conjunction with the original S-LINK specification (which has been stable since March 1997). Basically S-LINK64 defines a second connector that carries an additional 32 data lines and increases the maximum clock speed up to 100 MHz. Both specifications can be found at the EDMS S-LINK project page.

The CMS Trigger and Data Acquisition Group TriDAS group has presented already in June 2000 the ideas of using the S-LINK "data funnel" for the detector. No S-LINK64 link designs are started yet. Attila Racz from CMS will be the first one to design an S-LINK64 board, but it will not be before 2002 that boards are available. The S-LINK team has no plans to develop S-LINK64 links, PCI interfaces or test tools in 2001.

Old S-LINK News

CERN - High Speed Interconnect - S-LINK
Erik van der Bij - 15 April 2002