FPGA Design from Scratch · The Silicon Fabric

Part II: Inside the FPGA

What a field-programmable gate array actually is: look-up tables, flip-flops, programmable routing, and the hard blocks of the Zynq-7020 on the PYNQ-Z2.

The fabric at a glance

Processing System2 × ARM Cortex-A9(Part VII)MMCMCLB: LUTs + FFsChapter 5block RAMChapter 6DSP48E1Chapter 6clock networkChapter 6I/O blocksChapter 5
The PL fabric of the Zynq-7020, schematically. Logic tiles (CLBs) hold the look-up tables and flip-flops; programmable routing joins them; columns of block RAM and DSP slices sit among them; clocks travel on a dedicated global network from clock management tiles; I/O blocks connect to the pins. The ARM processing system occupies its own corner of the same die.

What you will learn

Part I built digital logic from gates and flip-flops on paper. Part II opens up the chip that will carry your designs: the Zynq XC7Z020 on the PYNQ-Z2. Chapter 4 places the FPGA among CPUs, GPUs and custom chips and explains how a bitstream turns generic silicon into your circuit. Chapter 5 shows that every piece of logic ends up as a small truth table in a look-up table, next to flip-flops, carry chains and a sea of programmable wires. Chapter 6 covers the hard blocks — block RAM, DSP multipliers, clock synthesisers — and introduces the ARM processors on the same die.

After this part you will be able to look at a line of HDL and predict, roughly, what hardware it becomes and what it costs: how many LUTs, whether it uses the carry chain, a block RAM or a DSP slice, and what limits its speed.

Chapters in Part II

Share:XRedditLinkedIn