Skip to content
DenseChip
Learn
PracticeCode labInterview

DenseChip

Navigate

Menu

LearnPracticeCode labInterview

Tools

All toolsSV editorCode repoResume
DenseChip

Practice, courses, and chip-design tools in one focused workspace.

LearnPracticeCode labInterviewToolsLearning paths

© 2026 DenseChip

AboutContactPrivacyTerms
Skip to lesson

Course contents

  • How to Use This Course
  • Reading Paths
  • Quick Start: Your First Green Build
  • Part 1: How Chips Get Made6
    • Idea to Silicon
    • What RTL Is
    • ASIC, FPGA, SoC and IP
    • Front-End and Back-End: Who Does What
    • PPA Primer
    • What Happens After RTL
  • Part 2: Your Workstation5
  • Part 3: Specify and Architect3
  • Part 4: The Design Loop6
  • Part 5: The Verification Ladder6
  • Part 6: Automation and Reproducibility5
  • Part 7: Hand-Off to the Back-End3
  • Part 8: When Things Go Wrong3
  • Part 9: Project and Next Steps3
  • Reference4
Loading lesson

‹Part 1: How Chips Get Made
  1. Topics
  2. Start Here: RTL and Verification
  3. Part 1: How Chips Get Made
  4. ASIC, FPGA, SoC and IP

Part 1 · How Chips Get Made · Foundation (F)

ASIC, FPGA, SoC and IP

ASIC, FPGA, SoC and IP are not four exclusive boxes: what each name answers, how a SoC can be an ASIC, and what IP really means.

On this page

    Note

    Level: Foundation (F) · Answers: Is a SoC an ASIC, and is an FPGA a chip you program like a CPU?

    Before this: What RTL Is

    A job ad asks for SoC experience. A board advertises an FPGA with an embedded processor. A vendor sells a UART as IP. Three labels, and they overlap in ways the words alone never explain. Newcomers usually file all four terms as four kinds of chip, and then every sentence about them sounds like a contradiction.

    Is a SoC an ASIC, and is an FPGA a chip you program like a CPU?

    Four names that answer four different questions

    An ASIC is an application-specific integrated circuit: a chip designed for a purpose and fabricated in silicon. Fabrication means the transistors are printed permanently, so after manufacture the hardware cannot change. A smartphone application processor is an ASIC: it was designed for phones, built in a factory, and it will never become anything else.

    An FPGA is a field-programmable gate array: reconfigurable hardware whose function is set after manufacture. Loading a configuration bitstream connects its lookup tables and block RAM into the circuit you described, and loading a different bitstream builds a different circuit on the same chip. RTL can target an FPGA, which is why the register from the previous lesson could be tested on one without changing what the code means.

    A SoC is a system on chip: many functions on one chip, such as a processor, memory, and peripherals working together. SoC answers how much sits on the chip, not how the chip was built. That is why the question in the title has a yes inside it: a SoC built in a factory is an ASIC, and a SoC prototyped on reprogrammable hardware is not.

    IP means intellectual property: a reusable design block such as a UART (universal asynchronous receiver-transmitter), a FIFO (first in, first out buffer), a CPU (central processing unit) core, a bus controller, or a memory macro. A vendor delivers it as RTL source, as a netlist of gates, or as a hardened macro ready for layout, and a chip team drops it into their design instead of writing it again. The other reading of the letters, an internet address, never applies in chip design, so when a lesson here says IP it always means the block.

    One chip, two ways to build it

    The figure below shows how the four names fit together. Read the top box first: a SoC holding a CPU core, a UART, and memory, each one an IP block. Then follow the two arrows down. The same IP delivered as RTL can be fabricated as an ASIC SoC or wired onto an FPGA prototype, and the bottom row names what each path is physically made of.

    How the four terms relate. Look at the top box first, then follow both arrows down and compare what each built chip is made of and whether it can change.

    Which confusion goes with which term?

    The table below pins each term to its meaning, its typical use, and the confusion to drop. Read the last column as carefully as the first: each row names the exact mistake this lesson wants you to stop making.

    Term

    What it means

    Typical use

    Common confusion

    ASIC

    A chip designed for a purpose and fabricated in silicon; fixed after manufacture

    Phone processor, fixed-function video encoder

    That every ASIC is small or simple; a phone processor is a large ASIC

    FPGA

    Reconfigurable hardware wired up by a configuration bitstream after manufacture

    Prototyping a design, products that must change in the field

    That the bitstream is a program the chip runs like software; it rewires hardware instead

    SoC

    Many functions combined on one chip: processor, memory, peripherals

    Smartphone chip, embedded controller with memory on board

    That SoC names a build method; it names how much is integrated, and a SoC can be an ASIC

    IP

    A reusable design block delivered as RTL, a netlist, or a hardened macro

    UART (serial port controller), FIFO (first-in, first-out buffer), CPU (central processing unit) core, USB (universal serial bus) controller, memory macro

    That IP means an internet address; in chip design it always means the block

    Three traps hidden in plain sentences

    • Calling an FPGA a processor you program. A program tells a fixed CPU what to do step by step, while a bitstream rewires the FPGA into different hardware. Treating the two as the same leads you to expect software flexibility from a hardware prototype, and to miss that every path through the rewired logic still has to meet its speed target.

    • Reading IP as an internet address. Datasheets and vendor catalogs say IP core, and they always mean a reusable block. If you carry the networking meaning into a meeting about licensing a USB controller, every sentence about delivery format stops making sense.

    • Sorting every chip into exactly one bucket. An FPGA that contains an embedded processor acts as both reconfigurable hardware and a system on chip. A memory macro is IP and also the physical memory inside a SoC. The buckets overlap by design, so ask which question each name answers instead of picking one label per chip.

    Rule: ASIC says how a chip is built, FPGA says it can be rewired, SoC says how much is on it, and IP is the reusable block inside.

    Check yourself Sort these eight into ASIC, FPGA, SoC, and IP, allowing overlaps: a smartphone application processor; a UART controller design; a prototyping board with reprogrammable logic; a fixed-function video encoder chip; the CPU core inside a SoC; a vendor USB controller delivered as a netlist; an FPGA that contains an embedded processor; a memory macro.

    Smartphone application processor: SoC and ASIC, since many functions share one fabricated chip. UART controller design: IP, the reusable block. Prototyping board: carries an FPGA, which is the reconfigurable chip on it. Fixed-function video encoder: ASIC, fabricated for one job. CPU core inside a SoC: IP used as a component. USB (universal serial bus) controller as a netlist: IP in netlist form. FPGA with an embedded processor: FPGA that is also SoC-like, since it holds a whole system. Memory macro: IP in hardened form. Most items take two labels, which is exactly the point.

    Check yourself Is a SoC always an FPGA? Can a SoC be an ASIC?

    No to the first, yes to the second. A SoC names integration, not the build method: a phone processor is a SoC and an ASIC at once, fabricated in silicon. An FPGA prototype of that same system is reconfigurable hardware holding SoC-like content, but most shipped SoCs are fabricated ASICs.

    Check yourself Can the same UART RTL be used in an FPGA and an ASIC without changes?

    Sometimes. The logic transfers, but each target carries technology-specific assumptions: how clocks and resets are structured, whether memories infer correctly for each flow, and whether the code leans on vendor primitives that exist on only one side. Plan for small target-specific adjustments around an unchanged core.

    Interview

    Asked: "What is IP in chip design?"

    Weak answer: It is the address of a block on the network, like an IP address.

    Strong answer: It is intellectual property: a reusable design block such as a UART, FIFO, CPU core, or memory macro. A team licenses or reuses it instead of writing it again, and it arrives as RTL source, as a gate netlist, or as a hardened macro. I would name which form we received, because that decides what we can still change.

    Follow-ups: (1) What is the difference between soft IP and a hardened macro? (2) Why would a team buy a USB controller instead of writing one?

    Turning RTL into gates is where the ASIC and FPGA paths split in practice, and the course that owns that detail is the synthesis course:

    Related topics

    • Synthesis course: the synthesis flow overview

    • Lesson 1.2, What RTL Is

    Keep going

    Next: Lesson 1.4, Front-End and Back-End: Who Does What. You can name the pieces. The next lesson says who builds each one: where the front end ends and the back end begins.

    Key takeaways

    • ASIC names a fabricated chip, FPGA names reconfigurable hardware, SoC names integration, and IP names the reusable block.

    • The names overlap: a SoC can be an ASIC, an FPGA can hold a system, and IP lives inside both.

    • The same RTL can target an FPGA or an ASIC, with small adjustments for clocks, resets, memories, and vendor primitives.

    Previous

    6 / 47

    Next: Front-End and Back-End: Who Does What

    Up next

    Up nextFront-End and Back-End: Who Does What

    Sign in to save progress

    Practice this lessonReport an error

    See also

    • Idea to Silicon
    • PPA Primer
    • What Happens After RTL