FPGA Manufacturers and Their Products

You've likely come across the names of various FPGA manufacturers and their products.

But you might not be too familiar with these companies and their offerings, or perhaps you're unsure about which products belong to which manufacturer.

In this video, I'll provide a concise overview of these FPGA manufacturers and their product categories.

I'll also discuss their product performance and differences to help you make more informed choices when necessary.

Major FPGA Manufacturers: Xilinx and Altera

There are several companies worldwide that produce FPGAs, but an overwhelming 80 percent of the total production is dominated by two key players: Xilinx and Altera.

Now, you might have heard that in 2015, Intel acquired Altera, which means that Altera is now part of the Intel family.

As a result, their FPGAs are referred to as Intel FPGAs, and we no longer use the Altera name.

In a different turn of events, AMD acquired Xilinx in 2020.

However, despite this change in ownership, Xilinx's FPGA products are still referred to as Xilinx FPGAs, and you can access their website just as you did before.

This differs from the Altera scenario, where if you try to visit the old website URL, you'll be redirected to the FPGA product section of the Intel website.

According to statistics from 2020, around 52 percent of FPGAs were produced by Xilinx, while Intel accounted for 35 percent.

The remaining market share, which is relatively small, is distributed among other FPGA manufacturers.

That's why in most companies that work on digital projects using FPGAs, you'll find that they mostly use FPGAs from these two big companies.

Exploring Xilinx FPGA Products

Now, let's dive into Xilinx's product lineup.

Xilinx's products are organized into various families, each with its own set of members.

Starting with the older families, we have Spartan3, Spartan6, Virtex4, Virtex5, Virtex6, and more recently, Spartan7, Virtex7, Kintex7, and Artix7.

Among the newer FPGA families, we encounter Virtex Ultrascale and Kintex Ultrascale.

The most recent additions are Virtex Ultrascale+, Kintex Ultrascale+, and Artix Ultrascale+.

Within each of these FPGA families, there are different members. The primary distinction between these members lies in the quantity and capacity of hardware resources they offer.

Now, let's talk about a special Xilinx product called the ZYNQ chip.

Xilinx doesn't classify this product as a traditional FPGA; instead, they refer to it as an SOC or System On Chip. This is because the FPGA is just one component of this chip.

Within the ZYNQ chip, you'll find a well-known microprocessor- ARM. Yes, the same ARM microprocessor you're probably familiar with.

This means that alongside the FPGA, there's a microprocessor that you can program.

With this chip, you have the flexibility to design hardware using the FPGA and implement computational and control algorithms through a combination of hardware and software.

In essence, working with this chip presents a new challenge: deciding how to distribute processing tasks between software and hardware.

This topic itself warrants a detailed discussion.

Comparing Xilinx FPGA Families

Let's now take a closer look at the main Xilinx products when it comes to performance, power consumption, and pricing.

I've prepared a chart to illustrate this comparison.

In terms of performance and functionality, VIRTEX stands out as the top performer. Following that, we have KINTEX and then ARTIX.

When we talk about "highest functionality," it means that VIRTEX enables you to implement circuits with the highest speed.

Looking at the digital resources available in the FPGA, VIRTEX offers the most extensive feature set.

However, when it comes to power consumption, it's the reverse of the performance picture.

VIRTEX, with its high functionality, consumes the most power, while ARTIX consumes the least.

Now, let's discuss pricing. VIRTEX, which offers the highest functionality, also comes with the highest price tag.

On the other hand, ARTIX has a more affordable price point.

It's worth noting that SPARTAN, although not shown on this chart, is Xilinx's most budget-friendly FPGA.

However, compared to the FPGAs listed on this chart, SPARTAN has lower capabilities.

As I mentioned earlier, each Xilinx FPGA family has its own set of members.

To give you a better idea, I've displayed some of the members from the SPARTAN 6 FPGA family on the table below.

Keep in mind that there are many more members within each family.

For a more detailed understanding, I've summarized information about four specific members of the SPARTAN 6 FPGA family in the table above.

In the first column, you'll find the Part Number section, which lists the part numbers for these four SPARTAN 6 FPGA family members.

Moving on to the columns after the first one, I've highlighted some of the primary digital resources available in these FPGAs and compared their quantity and capacity for these four family members.

For instance, let's consider the number of Logic Cells, which, as you may recall from previous videos, are a fundamental component of FPGAs and contain LUTs (Look-Up Tables) within them.

In the case of SPARTAN 6 LX9 FPGA, you'll find 9,152 Logic Cells.

If we add kilo next to the number after LX, it's roughly equal to the number of Logic Cells, so it's about 9,000 Logic Cells.

In contrast, for another family member, SPARTAN 6 LX100, the Logic Cell count is 101,261, or roughly 100,000 Logic Cells.

This table also provides information about Block RAM capacity.

You can see the number of Multipliers available; for instance, SPARTAN 6 LX9 has 16 multipliers, while SPARTAN 6 LX100 boasts 180 multipliers.

Additionally, these FPGAs offer special pins beyond the standard input-output pins, which enable high-speed communications in the gigabit range.

These pins are known as Gigabit Transceivers, and they can achieve speeds of up to 3.2 GB/s in SPARTAN 6 FPGAs.

You might notice that some members of the SPARTAN 6 FPGA family have the letter "T" at the end of their name.

This indicates that they come equipped with these high-speed pins

For example, the SPARTAN 6 LX25T FPGA has two gigabit transceiver pins, while the SPARTAN 6 LX100 has eight.

Finally, in the last column, you can find the number of I/Os (Input/Output pins). For example, SPARTAN 6 LX9 offers 200 I/Os, while SPARTAN 6 LX100T has 498 I/Os.

Intel's FPGA Product Lineup

Now, let's explore the products offered by Intel.

Intel, like Xilinx, offers a range of FPGA products organized into various families, each with its own set of members.

Here's a list of the main products from Intel: Agilex, Stratix, Arria, MAX, and Cyclone.

Comparing Intel FPGA Families

In terms of comparison, if we look at performance and efficiency, Agilex leads the pack, followed by Stratix, and then Arria.

Conversely, when it comes to power consumption, the ranking is the opposite.

As for pricing, Agilex tends to have the highest price tag, given its superior performance and efficiency.

Understanding the Distinctions Between FPGA and CPLD

Now, let's briefly discuss Complex Programmable Logic Devices (CPLDs) and how they differ from FPGAs.

Without considering the internal structure of CPLDs, we can define them as a type of FPGA with significantly fewer internal resources.

To be more precise about the differences between FPGAs and CPLDs, consider a few key points.

First, as you may recall, FPGAs require configuration, where you specify the content of Look-Up Tables and the switch matrices to connect the wires.

For FPGAs, this configuration is volatile, behaving like RAM. If you configure an FPGA and then turn it off and turn it on, all the information will be erased, and the FPGA will return to its initial state.

As a result, FPGAs require an external permanent memory module for configuration storage.

In contrast, CPLDs come with built-in permanent memory, allowing you to store configuration content permanently.

This eliminates the need for external permanent memory.

Additionally, CPLDs are far less complex than FPGAs, resulting in significantly lower power consumption.

Another distinction between FPGAs and CPLDs lies in the array of hardware resources.

FPGAs boast various resources such as memory blocks, multipliers, clock resources, high-speed pins (aka gigabit transceivers), and, in some cases, embedded microprocessors.

CPLDs, however, lack these hardware resources and are primarily composed of a number of logic gates.

So, when should you consider using CPLDs?

Typically, they are suitable for very small designs that demand exceptionally low power consumption and rapid design development.

In most other cases, FPGAs are the preferred choice.