AtlasPublicTopicHeader.png

L1 Muon Hardware Public Results

Introduction

Approved plots that can be shown by ATLAS speakers at conferences and similar events. Please do not add figures on your own. Contact the responsible project leader in case of questions and/or suggestions.

SROD for L1 Endcap Muon Trigger

Performance of Run 3 Level-1 endcap muon trigger readout system: PLOT-TDAQ-2021-02 (May 2021)

The processing speed of SROD EventBuilder application for various data size at the nominal L1A rate of 100 kHz. One SROD receives TTC data from one TTC Fanout Board and trigger data from 12 New Sector Logic (SL) boards. EventBuilder application builds event data using these data fragments and sends them to ROS. For this plot, fixed data were sent from 12 New SLs. The processing speed corresponds to the actual L1A rate considering the deadtime due to busy assertion by the trigger readout system. Average of 10 million events has been taken for each point. It has been observed that EventBuilder can process readout data of size up to 5000 bits without any deadtime. This corresponds to about three times the average data size expected in Run 3 (1750 bits/event), which is shown by the red dashed line. .png
png pdf
contact: Kaito Sugizaki, Stefano Veneziano
The processing speed of SROD EventBuilder application for various L1A rate for fixed readout data of 1760 bits per event, which is approximately equal to the average data size expected in Run 3. Using fixed data sent from 12 New Sector Logic (SL) boards, event data were built by EventBuilder and then sent to ROS. The processing speed corresponds to the actual L1A rate considering the deadtime due to busy assertion by the trigger readout system. Average of 10,000 events has been taken for each point, and the red dashed line corresponding to “processing speed = L1A rate” has been drawn for reference. It has been shown that EventBuilder can stably process readout data of expected size at nominal L1A rates in the range of 100 Hz to 170 kHz without any deadtime. .png
png pdf
contact: Kaito Sugizaki, Stefano Veneziano
The relation between the event-building time per event in SROD EventBuilder application and the size of readout data sent from SROD to ROS at the nominal L1A rate of 100 kHz. Average of 10 million events has been taken for each point. The event-building time is proportional to the size of data and contributes directly to the SROD performance. The event-building process consists of simple one-step procedures which only change the order of data, thus inducing this simple linear relationship. .png
png pdf
contact: Kaito Sugizaki, Stefano Veneziano
The relation between the data transfer time per event in SROD EventBuilder application and the size of readout data sent from SROD to ROS at the nominal L1A rate of 100 kHz. Average of 10 million events has been taken for each point. The data transfer time is in linear relation with the size of data and contributes directly to the SROD performance. The 3 μs offset is caused by the checking procedures of S-LINK busy channel and will be further optimized for better performance. .png
png pdf
contact: Kaito Sugizaki, Stefano Veneziano
The ratio of “pad words” in data sent from one New Sector Logic (SL) board to SROD. Pad words are dummy data sent between each event fragment at low L1A rates to reduce the readout delay caused by TCP Nagle buffering. They are not sent at high rates and do not affect the trigger readout performance. In this configuration, TCP Maximum Segment Size (MSS) of SL was set to 360 bytes = 90 words. Pad words were input to complement the event data in SL TCP TX buffer whenever further event data was not input for a time interval of 1 ms. Average of 200,000 events has been taken for each point. It has been proved that the pad word implementation works as expected. .png
png pdf
contact: Kaito Sugizaki, Stefano Veneziano

SROD performance at CERN test bench: ATL-DAQ-PROC-2018-032 (June 2018)

The measured the performance of the SROD system using the CERN test-bench environment. The setup used: 12 SL boards, 1 TTC-Fanout board, 1 SROD PC and the basic trigger systems. This test does not use the ROS or SFO systems of the full ATLAS Level-1 end-cap muon trigger system.

Measured output rate is shown as a function of the input data size. The Level-1 trigger rate was fixed to 100 kHz and the input data size was changed. The x-axis means the input data transfer rate per a SL board.

.png
png pdf
contact: Kosuke Takeda, Stefano Veneziano
The measured the performance of the SROD system using the CERN test-bench environment. The setup used: 12 SL boards, 1 TTC-Fanout board, 1 SROD PC, 1 ROS PC, 1 SFO PC and the basic trigger systems. This setup reproduces the ATLAS Level-1 end-cap muon trigger system.

Measured output rate is shown as a function of the Level-1 trigger rate. The Level-1 trigger rate was changed and the input data size was fixed. The x-axis means the Level-1 trigger rate, and the y-axis means the SROD processing rate.

.png
png pdf
contact: Kosuke Takeda, Stefano Veneziano


Major updates:
-- StefanoVeneziano - 19-Nov-2018
Responsible: StefanoVeneziano
Subject: public

Topic attachments
I Attachment History Action Size Date Who CommentSorted ascending
PDFpdf SRODPerformance-200kHz.pdf r1 manage 13581.9 K 2018-12-04 - 13:39 MarkStockton  
PNGpng SRODPerformance-200kHz.png r1 manage 55.7 K 2018-12-04 - 13:39 MarkStockton  
PDFpdf SRODPerformance-SlinkLimit.pdf r1 manage 13581.9 K 2018-12-04 - 13:39 MarkStockton  
PNGpng SRODPerformance-SlinkLimit.png r1 manage 46.7 K 2018-12-04 - 13:39 MarkStockton  
Edit | Attach | Watch | Print version | History: r35 < r34 < r33 < r32 < r31 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r35 - 2021-05-10 - JumpeiMaeda
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Atlas All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback