generic polyphase filterbanks with cuda
play

Generic Polyphase Filterbanks with CUDA Jan Krmer DLR German - PowerPoint PPT Presentation

Apr 14, 2023 •581 likes •1.01k views

Generic Polyphase Filterbanks with CUDA Jan Krmer DLR German Aerospace Center Communication and Navigation Satellite Networks Weling 04.02.2017 r rr www.dlr.de Slide 1 of

  1. Generic Polyphase Filterbanks with CUDA Jan Krämer DLR German Aerospace Center Communication and Navigation Satellite Networks Weßling 04.02.2017 ❑♥♦✇❧❡❞❣❡ ❢♦r ❚♦♠♦rr♦✇ ❉▲❘

  2. www.dlr.de · Slide 1 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

  3. www.dlr.de · Slide 1 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

  4. www.dlr.de · Slide 2 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Once upon a time in a space project Multicarrier scheme with 15/30/45 carrier ❉▲❘

  5. www.dlr.de · Slide 2 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Once upon a time in a space project Multicarrier scheme with 15/30/45 carrier So let’s just use a PFB, right? ❉▲❘

  6. www.dlr.de · Slide 3 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Early Trouble 45 carrier means 45x the bandwidth Only 12-15 % guardband available At least 3x oversampling needed Up to 1500 tap filters needed ❉▲❘

  7. www.dlr.de · Slide 3 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Early Trouble 45 carrier means 45x the bandwidth Only 12-15 % guardband available At least 3x oversampling needed Up to 1500 tap filters needed ❉▲❘

  8. www.dlr.de · Slide 4 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Early Trouble CPU reference implementation 1000 taps 35dB rejection Originally 9x oversampling 2 Msamples/second achieved ⇒ 4 Msamples/second needed ❉▲❘

  9. www.dlr.de · Slide 4 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

  10. www.dlr.de · Slide 5 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 What is CUDA NVidias framework for GPGPU Used mainly to accelerate scientific computing Uses the massive amount of available compute cores inside a GPU ❉▲❘

  11. www.dlr.de · Slide 6 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 GPU Interior GPU consists of several Streaming Multiprocessors (SM) Each SM consists of numerous compute or CUDA cores Single-Instruction Multiple-Threads (SIMT) structure Several kinds of memory Global Memory (GDDR5 RAM) (slow) On-Chip (shared) Memory per SM (faster) Registers (blazingly fast) ❉▲❘

  12. www.dlr.de · Slide 7 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 CUDA Interior Builds a (up to) 3 dimensional Grid The Gid contains the (up to) 3 dimensional Thread Blocks containing the threads Groups of 32 threads inside a Thread Block are grouped together ⇒ Warp ❉▲❘

  13. www.dlr.de · Slide 8 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 CUDA Interior ❉▲❘

  14. www.dlr.de · Slide 9 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Thread Execution Each Block has a unique ID inside the Grid ⇒ Each thread has a unique global ID Thread Scheduler assigns each Thread Block to one SM and executed concurrently All threads in a Warp are executed concurrently inside the SM ❉▲❘

  15. www.dlr.de · Slide 10 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Performance Bottlenecks Uncoalesced loads from global memory ⇒ Several cache-lines to be loaded ❉▲❘

  16. www.dlr.de · Slide 10 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Performance Bottlenecks Uncoalesced loads from global memory ⇒ Several cache-lines to be loaded Bank conflicts when accessing shared memory ❉▲❘

  17. www.dlr.de · Slide 10 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Performance Bottlenecks Uncoalesced loads from global memory ⇒ Several cache-lines to be loaded Bank conflicts when accessing shared memory Branching ⇒ Which instruction should be executed? ❉▲❘

  18. www.dlr.de · Slide 10 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

  19. www.dlr.de · Slide 11 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Why PFBs and Channelizers/Synthesizers? Used to reduce computational complexity for resampling filters Used to separate small bandwidth channels Used to generate multicarrier ’broadband’ signals ❉▲❘

  20. www.dlr.de · Slide 12 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Extracting a channel with 1 N of the total bandwidth Mix Signal to Baseband Apply anti-alias filter Downsample the signal ❉▲❘

  21. www.dlr.de · Slide 13 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Extracting a channel with 1 N of the total bandwidth Mix Signal to Baseband Apply anti-alias filter Downsample the signal N -phase PFB splits one-dimensional filter in its N different phase shares ❉▲❘

  22. www.dlr.de · Slide 14 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Taps of the regular prototype filter ❉▲❘

  23. www.dlr.de · Slide 15 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Taps of the regular prototype filter Split into 4 polyphase partitions ❉▲❘

  24. www.dlr.de · Slide 16 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Taps of the regular prototype filter Split into 4 polyphase partitions Newly structured dataflow ❉▲❘

  25. www.dlr.de · Slide 17 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Taps of the regular prototype filter Split into 4 polyphase partitions Newly structured dataflow FFT separates all the channels ❉▲❘

  26. www.dlr.de · Slide 18 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Structure of a PFB Channelizer Oversampling can be achieved by manipulating the input commutator and FFT input To synthesize several incoming channels just the reorder the operations ❉▲❘

  27. www.dlr.de · Slide 18 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

  28. www.dlr.de · Slide 18 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 ❉▲❘

  29. www.dlr.de · Slide 19 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Identifying necessary operations Channelizer consists of 4 operations Shuffle the input stream Polyphase filtering FFT Shuffle the output stream ❉▲❘

  30. www.dlr.de · Slide 20 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Input Shuffling Input Commutator implemented as matrix traversal Number of threads needs to accomodate to the filter history ⇒ Grid dimension takes care of this Input buffer reads are coalesced ⇒ Block x-dimension same size as polyphase partition Intermediate buffer writes are therfore not coalesced ❉▲❘

  31. www.dlr.de · Slide 21 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Filter Operations Block X dimension computes several input samples Block Y dimension computes oversampled output samples Grid X dimension represents polyphase partitions Grid Y dimension provide additional concurrency (due to block thread limits) ❉▲❘

  32. www.dlr.de · Slide 22 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Filter Operations Each threadblock transfers memory from global memory to shared memory Each sample is accessed several times ⇒ shared memory offers faster memory transfers Register and shared memory spills are avoided ❉▲❘

  33. www.dlr.de · Slide 23 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 FFT and Output Shuffling FFT is the CuFFT of CUDA Output shuffling implemented as double loop done on Host CPU (for now) ❉▲❘

  34. www.dlr.de · Slide 23 of 27 > Generic Polyphase Filterbanks with CUDA > Jan Krämer > 04.02.2017 Outline 1. Motivation 2. Short introduction to CUDA 3. PFBs and the Channelizer 4. Translation to CUDA 5. Results 6. Release plans and future changes ❉▲❘

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Multi-rate Signal Processing 3. The Polyphase Representation Electrical & Computer

Multi-rate Signal Processing 3. The Polyphase Representation Electrical & Computer

3 The Polyphase Representation Appendix: Detailed Derivations Multi-rate Signal Processing 3. The Polyphase Representation Electrical & Computer Engineering University of Maryland, College Park Acknowledgment: ENEE630 slides were based on

793 views • 25 slides
Introduction to CUDA C What is CUDA? CUDA Architecture Expose general-purpose GPU

Introduction to CUDA C What is CUDA? CUDA Architecture Expose general-purpose GPU

Introduction to CUDA C What is CUDA? CUDA Architecture Expose general-purpose GPU computing as first-class capability Retain traditional DirectX/OpenGL graphics performance CUDA C Based on industry-standard C A handful of

1.31k views • 62 slides
Lecture 2.1 - Introduction to CUDA C CUDA C vs. Thrust vs. CUDA Libraries Objective To learn

Lecture 2.1 - Introduction to CUDA C CUDA C vs. Thrust vs. CUDA Libraries Objective To learn

GPU Teaching Kit Accelerated Computing Lecture 2.1 - Introduction to CUDA C CUDA C vs. Thrust vs. CUDA Libraries Objective To learn the main venues and developer resources for GPU computing Where CUDA C fits in the big picture 2 3

1.28k views • 12 slides
Lesson 7 Rate Conversion Filtering and Downsampling interchange Filtering and Upsampling

Lesson 7 Rate Conversion Filtering and Downsampling interchange Filtering and Upsampling

Lesson 7 Rate Conversion Filtering and Downsampling interchange Filtering and Upsampling interchange Polyphase decomposition Polyphase Filtering Polyphase Filtering Polyphase filtering Example Example Example Example -n -n Example

474 views • 16 slides
CUDA/Ada An Ada binding to CUDA Reto B urki, Adrian-Ken R uegsegger University of Applied

CUDA/Ada An Ada binding to CUDA Reto B urki, Adrian-Ken R uegsegger University of Applied

Introduction Bindings in Ada CUDA/Ada Design CUDA Binding Conclusion CUDA/Ada An Ada binding to CUDA Reto B urki, Adrian-Ken R uegsegger University of Applied Sciences Rapperswil (HSR), Switzerland 1/16/2012 Master seminar: Program

469 views • 36 slides
1 Definition of a simple generic class Why generic programming (cont.) class Pair <T> {

1 Definition of a simple generic class Why generic programming (cont.) class Pair <T> {

Topics background and goals of generic programming basics of generic classes = parameterized types generic methods for general algorithms inheritance rules for generic types Generic programming in Java bounded type parameters

261 views • 5 slides
Lecture 2.4 Introduction to CUDA C Introduction to the CUDA Toolkit Objective To become

Lecture 2.4 Introduction to CUDA C Introduction to the CUDA Toolkit Objective To become

GPU Teaching Kit Accelerated Computing Lecture 2.4 Introduction to CUDA C Introduction to the CUDA Toolkit Objective To become familiar with some valuable tools and resources from the CUDA Toolkit Compiler flags Debuggers

887 views • 34 slides
Signal processing with heterogeneous digital filterbanks: lessons from the MWA and EDA Randall

Signal processing with heterogeneous digital filterbanks: lessons from the MWA and EDA Randall

Signal processing with heterogeneous digital filterbanks: lessons from the MWA and EDA Randall Wayth ICRAR/Curtin University with Marcin Sokolowski, Cathryn Trott "Holy grail of CASPER system is multi-user system Outline Jack H

942 views • 27 slides
Outline Overview Parallel Computing with GPU Introduction to CUDA CUDA Thread Model

Outline Overview Parallel Computing with GPU Introduction to CUDA CUDA Thread Model

Outline Overview Parallel Computing with GPU Introduction to CUDA CUDA Thread Model CUDA Memory Hierarchy and Memory Spaces CUDA Memory Hierarchy and Memory Spaces CUDA Synchronization 2110412 Parallel Comp Arch CUDA:

449 views • 12 slides
GPU Programming Alan Gray EPCC The University of Edinburgh Overview Motivation and need

GPU Programming Alan Gray EPCC The University of Edinburgh Overview Motivation and need

GPU Programming Alan Gray EPCC The University of Edinburgh Overview Motivation and need for CUDA Introduction to CUDA CUDA kernels, decompositions CUDA memory management C and Fortran OpenCL 2 NVIDIA CUDA

1.24k views • 37 slides
A High-Level Intro to CUDA CS5220 Fall 2015 What is CUDA? C ompute U nified D evice A

A High-Level Intro to CUDA CS5220 Fall 2015 What is CUDA? C ompute U nified D evice A

A High-Level Intro to CUDA CS5220 Fall 2015 What is CUDA? C ompute U nified D evice A rchitecture released in 2007 GPU Computing Extension of C/C++ requires NVCC (CUDA Compiler) and NVIDIA Graphics Card Historical

507 views • 38 slides
Specification of APERTIF Polyphase Filter Bank in C aSH Rinse Wester a , Dimitrios Sarakiotis a

Specification of APERTIF Polyphase Filter Bank in C aSH Rinse Wester a , Dimitrios Sarakiotis a

Specification of APERTIF Polyphase Filter Bank in C aSH Rinse Wester a , Dimitrios Sarakiotis a , Eric Kooistra b , Jan Kuper a University of Twente, Enschede ASTRON, Dwingelo August 27, 2012 1 Monday, August 27, 12 Contents

327 views • 31 slides
CUDA 7 AND BEYOND MARK HARRIS, NVIDIA CUDA 7 Runtime C++11 cuSOLVER Compilation

CUDA 7 AND BEYOND MARK HARRIS, NVIDIA CUDA 7 Runtime C++11 cuSOLVER Compilation

CUDA 7 AND BEYOND MARK HARRIS, NVIDIA CUDA 7 Runtime C++11 cuSOLVER Compilation [&](char)c)){) ))for)(auto)x):)letters))) ))))if)(c)==)x))return)true;) ))return)false;) }) C++11 FEELS LIKE A NEW LANGUAGE Bjarne Stroustrup,

1.1k views • 37 slides
Generic Methods 36 What are Generic Methods? Generic methods = methods that introduce type

Generic Methods 36 What are Generic Methods? Generic methods = methods that introduce type

Generic Methods 36 What are Generic Methods? Generic methods = methods that introduce type parameters Similar to declaring a generic type but type parameter's scope is limited to method where it is declared The following are

584 views • 6 slides
How to Write a Parallel GPU Application Using CUDA and Charm++ Presented by Lukasz Wesolowski

How to Write a Parallel GPU Application Using CUDA and Charm++ Presented by Lukasz Wesolowski

How to Write a Parallel GPU Application Using CUDA and Charm++ Presented by Lukasz Wesolowski Outline GPGPUs and CUDA Requirements for a GPGPU API (from a Charm++ standpoint) CUDA stream approach Charm++ GPU Manager 2 General

562 views • 11 slides
Approaches to GPU computing Manuel Ujaldon Nvidia CUDA Fellow Computer Architecture Department

Approaches to GPU computing Manuel Ujaldon Nvidia CUDA Fellow Computer Architecture Department

Approaches to GPU computing Manuel Ujaldon Nvidia CUDA Fellow Computer Architecture Department University of Malaga (Spain) Talk outline [40 slides] 1. Programming choices. [30] 1. CUDA libraries and tools. [10] 2. Targeting CUDA to other

1.17k views • 49 slides
Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical Engineering Lecture 38: April

Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical Engineering Lecture 38: April

Mixed-Signal VLSI Design Course Code: EE719 Department: Electrical Engineering Lecture 38: April 10, 2018 Instructor Name: M. Shojaei Baghini E-Mail ID: mshojaei@ee.iitb.ac.in 1 2 2 Module 49 Resolution Enhancement in ADCs using

348 views • 11 slides
T witte r F e e ds Pr ofiling With T F - IDF Juraj Petrik & Daniela Chuda 1 T a sk

T witte r F e e ds Pr ofiling With T F - IDF Juraj Petrik & Daniela Chuda 1 T a sk

T witte r F e e ds Pr ofiling With T F - IDF Juraj Petrik & Daniela Chuda 1 T a sk Given celebrity Twitter feed (English not guaranteed) Determine: Fame level Occupation Age Gender 2 Motiva tion Our

501 views • 17 slides
Learning and Imbalanced Data January 28, 2019 David Rimshnick Data Science in the Wild, Spring

Learning and Imbalanced Data January 28, 2019 David Rimshnick Data Science in the Wild, Spring

Learning and Imbalanced Data January 28, 2019 David Rimshnick Data Science in the Wild, Spring 2019 What is data imbalance? Unequal distribution of data towards a certain characteristic Target variable Classification: Certain

414 views • 15 slides
Kolmogorov-Chaitin Complexity of Linear Digital Controllers Implemented using Fixed-point

Kolmogorov-Chaitin Complexity of Linear Digital Controllers Implemented using Fixed-point

Kolmogorov-Chaitin Complexity of Linear Digital Controllers Implemented using Fixed-point Arithmetic James F. Whidborne , John Mckernan , Da-Wei Gu dag@leicester.ac.uk Department of Engineering, University of Leicester

421 views • 29 slides
From Atoms to Bits Ahmet Onat 2018 onat@sabanciuniv.edu Layout of the Lecture Analog

From Atoms to Bits Ahmet Onat 2018 onat@sabanciuniv.edu Layout of the Lecture Analog

From Atoms to Bits Ahmet Onat 2018 onat@sabanciuniv.edu Layout of the Lecture Analog interfacing to sensors: Signal conditioning Sampling and quantization Bridge circuits and instrumentation amplifiers Linearization

972 views • 80 slides
Hate Speech Detection is Not as Easy as You May Think: A Closer Look at Model Validation Aym

Hate Speech Detection is Not as Easy as You May Think: A Closer Look at Model Validation Aym

Hate Speech Detection is Not as Easy as You May Think: A Closer Look at Model Validation Aym Arango, Jorge Prez and Brbara Poblete UNDETECTED ALMOST PERFECT HATE SPEECH VS STATE-OF-THE-ART IN RESULTS SOCIAL MEDIA UNDETECTED HATE

641 views • 47 slides
Welcome! Todays Agenda: Primitives (contd.) Ray Tracing Intersections

Welcome! Todays Agenda: Primitives (contd.) Ray Tracing Intersections

INFOGR Computer Graphics Jacco Bikker - April-July 2016 - Lecture 3: Ray Tracing (Introduction) Welcome! Todays Agenda: Primitives (contd.) Ray Tracing Intersections Assignment 2 Textures INFOGR

529 views • 51 slides
Multilinear Algebra Based Fitting of a Sum of Exponentials to Oversampled Data Lieven De

Multilinear Algebra Based Fitting of a Sum of Exponentials to Oversampled Data Lieven De

Multilinear Algebra Based Fitting of a Sum of Exponentials to Oversampled Data Lieven De Lathauwer Jean-Michel Papy Sabine Van Huffel K.U.Leuven FMTC /K.U.Leuven K.U.Leuven Catholic University of Leuven Flanders

398 views • 35 slides

More recommend