From SRAM to DDR3 for VFC-HPC V2 Consequences and performances.

Slides:



Advertisements
Similar presentations
Synchronous Static Random Access Memory (SSRAM). Internal Structure of a SSRAM AREG: Address Register CREG: Control Register INREG: Input Register OUTREG:
Advertisements

High Speed Data Acquisition Architectures. Some Basic Architectures Non-Buffered (streaming) FIFO Buffered Multiplexed RAM Ping Pong Multiplexed RAM Dual.
2009 Spring Errors & Source of Errors SpringBIL108E Errors in Computing Several causes for malfunction in computer systems. –Hardware fails –Critical.
Lorentz force detuning measurements on the CEA cavity
1 Fully Digital HF Radios Phil Harman VK6APH Dayton Hamvention – 17 th May 2008.
Ultrafast 16-channel ADC for NICA-MPD Forward Detectors A.V. Shchipunov Join Institute for Nuclear Research Dubna, Russia
CHL -2 Level 1 Trigger System Fully Pipelined Custom ElectronicsDigitization Drift Chamber Pre-amp The GlueX experiment will utilize fully pipelined front.
COMMUNICATION TECHNOLOGY by Shashi Bhushan School of Computer and Information Sciences.
Conversion Between Video Compression Protocols Performed by: Dmitry Sezganov, Vitaly Spector Instructor: Stas Lapchev, Artyom Borzin Cooperated with:
The first testing of the CERC and PCB Version II with cosmic rays Catherine Fry Imperial College London CALICE Meeting, CERN 28 th – 29 th June 2004 Prototype.
DATA ACQUISITION SYSTEM FPGA2 APEX20K200E SAMSUNG MICROCONTROLLER ARM - RISC CORE (50MHZ – 32 BIT, 8 KByte SRAM) BOOT FLASH 512K X 16 PROGRAM MEMORY SDRAM.
Ph. Farthouat CERN ELEC 2002 ADC 1 Analog to Digital Conversion  Introduction  Main characteristics –Resolution –Dynamic range –Bandwidth –Conversion.
3/7/05A. Semenov Batch-by-Batch Intensity Monitor 1 Two-Channel Batch by Batch Intensity Monitor for Main Injector BBI.
A New Reference Design Development Environment for JPEG 2000 Applications Bill Finch CAST, Inc. Warren Miller AVNET Design Services DesignCon 2003 January.
AUDIO COMPRESSION msccomputerscience.com. The process of digitizing audio signals is called PCM PCM involves sampling audio signal at minimum rate which.
The GANDALF Multi-Channel Time-to-Digital Converter (TDC)  GANDALF module  TDC concepts  TDC implementation in the FPGA  measurements.
Computational Technologies for Digital Pulse Compression
DLS Digital Controller Tony Dobbing Head of Power Supplies Group.
Data and Computer Communications
Data Handling Stephen Kaye Caltech Data Format in Pipeline 16 Bit data from ADC FPGA combines multiple conversions (subtract 5 reset, add.
PicoTDC Features of the picoTDC (operating at 1280 MHz with 64 delay cells) Focus of the unit on very small time bins, 12ps basic, 3ps interpolation Interpolation.
LIGO-G E Data Compression study with E2 data S. Klimenko, B. Mours, P. Shawhan, A. Sazonov March LSC 2001 meeting Baton Rouge, Louisiana LSC Session.
Status of the Beam Phase and Intensity Monitor for LHCb Richard Jacobsson Zbigniew Guzik Federico Alessio TFC Team: Motivation Aims Overview of the board.
1 Programming of FPGA in LiCAS ADC for Continuous Data Readout Week 3 Report Jack Hickish.
Outline Kinds of Coding Need for Compression Basic Types Taxonomy Performance Metrics.
Measurement opportunities with the LHC transverse damper W. Hofle, D. Valuch.
Dima Svirida (ITEP/BNL) Comments on Polarimeter Programming NEW IN RUN6  AGS polarimeter switched to 10 times faster readout with CMC100 USB controller.
Initial Performance Results of the APS P0 (Transverse Bunch-to-Bunch) Feedback System N. DiMonte#, C.-Y. Yao, Argonne National Laboratory, Argonne, IL.
Huffman Code and Data Decomposition Pranav Shah CS157B.
Digital Phase Control System for SSRF LINAC C.X. Yin, D.K. Liu, L.Y. Yu SINAP, China
Xiangming Sun1PXL Sensor and RDO review – 06/23/2010 STAR XIANGMING SUN LAWRENCE BERKELEY NATIONAL LAB Firmware and Software Architecture for PIXEL L.
Digital Phase Control System for SSRF LINAC C.X. Yin, D.K. Liu, L.Y. Yu SINAP, China
Alexei SemenovGeneric Digitizer Generic Digitizer 10MHZ 16 bit 6U VME Board.
Communications Equipment Communication Equipments: –Equipments that are, Not a direct part of the network (like node, client, terminal, server etc.) but,
Computer Architecture Lecture 32 Fasih ur Rehman.
CIS Data Communications1 CIS-325 Data Communication Dr. L. G. Williams, Instructor.
FPGA firmware of DC5 FEE. Outline List of issue Data loss issue Command error issue (DCM to FEM) Command lost issue (PC with USB connection to GANDALF)
Charge Measurement Using Commercial Devices Jinyuan Wu, Zonghan Shi For CKM Collaboration. Jan
1 07/10/07 Forward Vertex Detector Technical Design – Electronics DAQ Readout electronics split into two parts – Near the detector (ROC) – Compresses and.
Digital CFEB (an Update) B. Bylsma, EMU at CMS Week, March 16, Ben Bylsma The Ohio State University.
Making Connections Efficient: Multiplexing and Compression Data Communications and Computer Networks: A Business User’s Approach Seventh Edition.
SRS Activities at IFIN-HH: VMM2 Hybrid, FECv6 Firmware, High- Density Optical ATCA-SRS Mezzanine Sorin Martoiu, Michele Renda, Paul Vartolomei (IFIN-HH.
VFC-HD PROJECT STATUS AND OUTSTANDING SEPTEMBER 2015.
High Speed Digital Systems Lab Spring/Winter 2010 Project definition Instructor: Rolf Hilgendorf Students: Elad Mor, Ilya Zavolsky Integration of an A/D.
1 Programming of FPGA in LiCAS ADC for Continuous Data Readout Week 7 Report Friday 15 th August 2008 Jack Hickish.
A commercially available digitization system Fotiou Andreas Andreas Fotiou.
Block-Based Packet Buffer with Deterministic Packet Departures Hao Wang and Bill Lin University of California, San Diego HSPR 2010, Dallas.
1 Programming of FPGA in LiCAS ADC for Continuous Data Readout Week 6 Report Wednesday 6 th August 2008 Jack Hickish.
Status report 2010/9/24 Atsushi Nukariya. GEMFE2 outline (1) ・ FPGA controls 4 chips. ・ Each chip has 9 channels. ( 8 channels will be used in data acquisition,
Roman Kofman & Sergey Kleyman Neta Peled & Hillel Mendelson Supervisor: Mike Sumszyk Final Presentation of part A (Annual project)
09/09/2010 TDAQ WG - Louvain 1 LKr L0 trigger status report V. Bonaiuto, A. Fucci, G. Paoluzzi, A. Salamon, G. Salina, E. Santovetti, F. Sargeni, F. Scarfi’
TPC CRU Jorge Mercado (Heidelberg) Ken Oyama (Nagasaki IAS) CRU Team Meeting, Jan. 26, 2016.
Gunjeet Kaur Dronacharya Group of Institutions. Outline I Random-Access Memory Memory Decoding Error Detection and Correction Read-Only Memory Programmable.
Vanderbilt University Toshiba IR Test Apparatus Project Final Design Review Ahmad Nazri Fadzal Zamir Izam Nurfazlina Kamaruddin Wan Othman.
Readout controller Block Diagram S. Hansen - CD-1 Lehman Review1 VXO Ø Det Links to 24 SiPM Front End Boards Clock Event Data USB ARM uC A D Rd Wrt 100Mbit.
Lifecycle from Sound to Digital to Sound. Characteristics of Sound Amplitude Wavelength (w) Frequency ( ) Timbre Hearing: [20Hz – 20KHz] Speech: [200Hz.
1 Programming of FPGA in LiCAS ADC for Continuous Data Readout Week 4 Report Tuesday 22 nd July 2008 Jack Hickish.
Enhancement Presentation Carlos Abellan Barcelona September, 9th 2009.
MADEIRA Valencia report V. Stankova, C. Lacasta, V. Linhart Ljubljana meeting February 2009.
Digital Acquisition: State of the Art and future prospects
RF acceleration and transverse damper systems
Front-end Electronic for a neutrino telescope : a new ASIC SCOTT
96-channel, 10-bit, 20 MSPS ADC board with Gb Ethernet optical output
KRB proposal (Read Board of Kyiv group)
Ewald Effinger, Bernd Dehning
CoBo - Different Boundaries & Different Options of
CS Lecture 2 Network Performance
New PSB beam control rf clock distribution
Beam Current Monitoring with ICT and BPM Electronics
Operation and Control of 42 GHz Gyrotron system in ECRH
Presentation transcript:

From SRAM to DDR3 for VFC-HPC V2 Consequences and performances

From SRAM to DDR3: what would change? Memory total bandwidth reduced: from 11.5Gbps for the to 2.66Gbps (more than a factor 4) Memory storage increased: from 144Mb to 16Gb (more than a factor 100) Cost per board reduced of about 200CHF Note: 1 DDR3 at 1.33Gbps has the BW of a 16bit connected to a ARRIA V GX

DDR3 with ARRIA-V Vs. 12bit fast (400+Msps) ADC The issue: – DDR3 BW = 2.66Gbps – Data = 5.76Gbps per channel The possible solutions or compromises – Using an internal buffer and time multiplexing the acquisitions (TMX) – Data compression to a factor 2 (lossy or lossless) (CMP) – Data “rebinning” with smart algorithms to the bunch level leading to 40MHz (R2B)

Options: TMX Using an internal high speed buffer (FIFO) before the DDR3 The total memory is 15kb: reading the buffer at half the writing speed will allow us to store without loss 30kb before saturation. 30kb are 2.5ksamples or ~200 bunches (25ns slots) We could divide the accelerator in 18 batches of 198 consecutive bunches. Each turn we can acquire a subset of 9 of those and – Have a full accelerator every 2 turns (minus part of the abort gap) – Acquire for N turns half of the accelerator and the N turns after the other half – Use 2 boards to get to full rate

Options: CMP The patterns are well known and ad hoc algorithms could be studied to reduce the data rate (performance to be evaluated) Lossless compression could be used (non constant compression rate….) Lossy compression like – Wavelet compression (easy to implement in an FPGA) – Recoding like in simil-floating point. Using 4 bits to encode the exponent and 2 bits of mantissa we would introduce an error on the encoded sample of less than 12.5% even if we would use a 16bit ADC as data source (relative error is fixed)

Options: R2B With the knowledge of the pulse shape algorithms could be implemented to recover the amplitude of the pulses in the various 25ns slots even in the case of leakage of the signal into more than one slot. The signal can be in first approximation be estimated to have 16 bits starting from a 12 bit one going down to 40MHz with an equivalent BW of 640Mbps. This solution is compatible with 4 channels per VFC

Comparison TMXCMPR2B 400Msps 31k full turns Only half ring can be acquired each turn in slices of 200 bunches 62k full turns 280K turns Could be shared between 4 channels 440Msps 31k full turns Only half ring can be acquired each turn in slices of 200 bunches AG cannot be acquired in ping-pong mode 62k full turns 280K turns Could be shared between 4 channels 480Msps 31k full turns Only half ring can be acquired each turn in slices of 200 bunches AG cannot be acquired 62k full turns Requires compression factor of K turns Could be shared between 4 channels

References mem-index.jsp mem-index.jsp - memory/emi_intro_specs_select_memory.pdf - memory/emi_intro_specs_select_memory.pdf