Scribd
Upload
45 views24 pages

PDC: Introduction To Fgpa-Based Acceleration Architecture: Bcs6A FAHAD 18F-0181 AHMER 18F-0336

The document provides an introduction to FPGA-based acceleration. It discusses how FPGAs can accelerate data-intensive applications by providing parallel processing capabilities. FPGAs offer flexible and adaptable architectures that allow them to straddle multiple domains. They consist of logic blocks, I/O cells, and interconnect resources that can perform operations quickly in parallel. FPGAs are well-suited for applications like image and video processing. While slower than CPUs for general-purpose processing, FPGAs excel at tasks like high-speed data conversion that benefit from parallelism. Bittware designs FPGA accelerators for various uses like computing, networking, storage and sensor processing to further increase performance.

Uploaded by

Awais Shahid
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
45 views24 pages

PDC: Introduction To Fgpa-Based Acceleration Architecture: Bcs6A FAHAD 18F-0181 AHMER 18F-0336

The document provides an introduction to FPGA-based acceleration. It discusses how FPGAs can accelerate data-intensive applications by providing parallel processing capabilities. FPGAs offer flexible and adaptable architectures that allow them to straddle multiple domains. They consist of logic blocks, I/O cells, and interconnect resources that can perform operations quickly in parallel. FPGAs are well-suited for applications like image and video processing. While slower than CPUs for general-purpose processing, FPGAs excel at tasks like high-speed data conversion that benefit from parallelism. Bittware designs FPGA accelerators for various uses like computing, networking, storage and sensor processing to further increase performance.

Uploaded by

Awais Shahid
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 24

PDC: INTRODUCTION

TO FGPA-BASED BCS6A
FAHAD 18F-0181
ACCELERATION AHMER 18F-0336

ARCHITECTURE
INTRODUCTION

-DATA GENERATION QUICKER THAN


ABILITIY
-DATA INTENSIVE APPLICATIONS I.E.,
BIG DATA, ML, IOT AND HIGH SPEED
COMPUTING
-NEED OF ACCELERATION DEVICE
ACCELERATION DEVICES

-GPU
-ASIC
-FPGA
FPGAS
-FIELD PROGRAMMABLE GATE ARRAYS
-EXCELLENT PERFORMANCE IMPLEMENTATION
-FLEXIBLE
-ADAPTABLE ARCHITECTURE
-ABILITY TO STRADDLE MULTIPLE DOMAINS
-NEEDS SERVER TO FUNCTION
-ACCELERATES OR BOOSTS CPU SERVER ENGINE
CONSTRUCTION OF FPGA’S
-AN IC COLLECTION OF
-LOGIC BLOCKS
-I/O CELLS
-INTERCONNECTION RESOURCES
-PERFORMS AND & XOR OPERATIONS LIKE FLIP FLOP
-MODERN FPGA’S ARE ON CHIP SYSTEM
FGPA’S APPLICATIONS
•Aerospace & Defense
•ASIC Prototyping 
•Automotive
•Medical
•Video & Image Processing 
•Wired Communications
• Wireless Communications
FGPA’S APPLICATION
•Broadcast & Pro AV Consumer
•Electronics 
•Data Center High
•Performance Computing
•and Data Storage 
•Industrial 
FGPA’S MACHINE VISION
-Used A Lot In Machine learning
-Image Camera by Sony
-Takes Signals & Converts It Into Data
-Data Conversion
-Non-Crictical Process Such As Camera Control
-Data Conversion & Image Enhancement
COMAPRISON WITH CPU
PERFORMANCE
-CPU IS SPECIFIED AS HOW LONG FUNCTIONS TAKE PLACE
-FGPA IS BASED ON THROUGH PUT
-SPEED IS SLOWER THAN CPU
PARALLEL PROCESSING
NATURE
-RUN VERY FAST AND DIVIDES PROGRAMS IN PARALLEL
ADVANTAGES
-FOR A PINPOINT PROCESS
-SMALL SECTION OF DATA
-MANAGING HIGH RATES
-IMAGE PROCESSING
LIMITATIONS
-NOT FOR WHOLE
-DONOT HAVE LARGER MEMORY
-DONT HAVE BANKS OF REGISTERS
-REQUIRED EXTRA HARDWARE
COMPARISON
CONCLUSION
-WIDE SPREAD DATA CONVERSION
-OFF PROCESSION POWER FORM THE CPU
-ENABLE EXTREMELY HIGH BANDWIDTHS
GROWING DEMANDS FOR
FGPA ACCELERATOR
-HIGHLY FLEXIBLE FINE GRAINED PARALLELISM OFFERS
-DATA
-TASK
-PIPELINE PARALLELISM
-LOW LATENCY DATA PROCESSING DUE TO NON-INSTRUCTION
ARCHITECTURE
& DATA FLOW DESIGN
-EXCELLENT FLEXIBILITY
-QUICKLY DEVELOPING PROTOTYPES OF CIRCUIT
GROWING DEMANDS FOR
FGPA ACCELERATOR CONT.
-LOGICAL BLOCS
-I/O CELLS
-INTERCONNECTION RESOURCES
-ADVANCEMENT IN FGPA
-SHIFTING FROM LOW LEVEL TO HIGH LEVEL LANGUAGES & OPEN CL
-DATA CENTRES, AUTONOMOUS DRIVING, ROBOTICS & INDUSTRY 4.0
-BETTER FOR BATTERY HOSTED DEVICES AND CLOUD SERVICES ON
MASS SERVERS
BITTWARE FGPA’S
ACCELERATORS
-POWERFUL IN CIRCUITRY, DEVICES & INTERCONNECTS
-ACHRONIX, INTEL AND XILINX
-PARTITIONING WORKLOADS
-UNDERSTANDING MEMORY
-NETWORK BANDWIDTH REQUIREMENTS
-MANAGING THERMALS IN COMPLEX DATA CENTER
PROBLEM SOLVED BY
BITTWARE FGPA’S
-COMPUTE ACCELERATORS
-NETWORK ACCELERATORS
-STORAGE ACCELERATORS
-SENSOR PROCESSING ACCELERATORS
COMPUTE ACCELERATORS
-HETEROGENEOUS PLATFORMS EMPLOYING ACCELERATORS
-COMPUTATION PLATEFORM GENERATING HIGH PERFORMANCE
-FLEXIBLE & ENERGY EFFICIENT OPERATIONS
-520N-MX
-TRADITIONAL HLD & HIGHER ABSTRACTION
NETWORK ACCELERATORS
-NETWORK SPEED ISSUE
-MULTIPLY INFORMATION FLOW SPEED
-NETOWRK SECURITY FUNCTIONS
-REAL TIME PACKET CAPTURING
-DPI ABOVE 10 GB ETHERNET
-XUP-P3R BUILT WITH XILINX
-SMART NIC SHELL
STORAGE ACCELERATORS
-DATA PROCESSING IN PHYSCIAL PROXIMITY
-ARRANGMENT OF DATA STORAGE & COMPUTE PLANES
-TRADITIONALLY, DATA ALTERNATES BETWEEN STORAGE & COMPUTE
PLANES
-CPU OFFLOADING
-250 SERIES
SENSOR PROCESSING
ACCELERATOR:
-CORE MARKET FOR FPGA’S
-EMBEDDED APPLICATIONS LIKE RADAR,SDR AND ELECTRONIC
WARFARE FOR REALTIME DATA ACQUISITION, FILTERING & SIGNAL
PROCESSING
-XUP-VV9
-IDEAL FOR MULTICHANNEL SENSOR PROCESSING APPLICATIONS
CONCLUSION
-ACCELERATES HPC CENTERS, HYPERSCALERS, EDGE COMPUTERS,
DATA SCIENCTIST & CLOUD BUILDERS
-MICROSOFT PROJECT BRAINWAVE
-AMAZON
-NEXT GENERATION
REFERENCES
https://www.xilinx.com/products/silicon-devices/fpga/what-is-an-fpga.html

https://www.arrow.com/en/research-and-events/articles/fpga-basics-architecture-appl
ications-and-uses

https://www.mdpi.com › journal › FPGA_Architectures

https://www.mdpi.com › pdf

You might also like