Why are FPGAs so expensive?

Question

I mean compared to ICs (ASICs) with similar complexity, speed etc. Let's compare Ethernet switches to Kintex FPGAs (note that the most expensive switch from the list is circa as expensive as the cheapest Kintex):

• FPGAs are well structured ICs (like RAMs). They can be scaled and developed easily.
• The design tools (Vivado, Quartus, etc.) are expensive too, so I think the price of an FPGA is the price of the IC (and development) itself excluding the cost of support and the tools. (Some non-FPGA vendors give free tools whose development cost includes the IC price.)

Are FPGAs produced in lower quantities than other ICs? Or is there any technological harness?

Answer 1

FPGA chips include both logic and programmable connections between logic elements, while ASICs include only the logic.

You'd be amazed at how much chip area is devoted to the "connection fabric" in an FPGA — it's easily 90% or more of the chip. This means that FPGAs use at least 10× the chip area of an equivalent ASIC, and chip area is expensive!

It costs a certain amount to do all of the processing on a given silicon wafer, no matter how many individual chips are on it. Therefore, to a first approximation, the chip cost is directly proportional to its area. However, there are several factors that make it worse than that. First, larger chips mean that there are fewer usable sites on the wafer to begin with — wafers are round, chips are square, and a lot of area is lost around the edges. And defect densities tend to be constant across the wafer, which means that the probability of getting a chip without a defect (i.e., "yield") goes down with chip size.

Answer 2

Another key driver of cost is verification.

FPGAs need to be individually tested before sale. This is partly to ensure that all of the thousands to several million routing interconnects and logic cells are functional. The verification however also involves characterisation and speed grade binning - determining how fast the silicon can operate and that the speed and propagation delays of all the many interconnects and cells are suitably matched to the timing models for its grade.

For ASIC designs, testing is typically simpler - a yes-no does the design perform as expected. As such the time required for verification is likely far less, and thus cheaper to perform.

Answer 3

There is one (more) important point which is usually overlooked, process technology.

FPGAs that have high market share are manufactured with cutting edge technology. To be more specific, Kintex-7 FPGAs have TSMC 28nm process and their shipment started in 2011^[1]. TSMC had started mass production of 28nm in the same year^[2].

[1] Xilinx ships first 28nm Kintex-7 FPGAs (By Clive Maxfield, 03.21.11)

[2] Chang said: "Our 28-nm entered volume production last year and contributed 2 percent of 4Q11's wafer revenue."

I don't know the process of the ethernet switches, but most of the ASIC design companies don't follow the cutting edge technology. It doesn't make sense for foundries as well.

The following chart shows TSMC's revenue by technology (1Q18). Even in 2018, 39% of the revenue comes from technologies older than 28nm. If we think about the number of chips, it is not hard to imagine that more than half of ASICs are today manufactured with technologies older than 7-year-old Kintex-7.

As a conclusion, process technology is one of the factors that make FPGAs more expensive. I don't claim it is a dominant factor, but significant enough to be considered.

Answer 4

I'm going to go out on a limb and say that this is by far dominated by simple supply and demand. Ethernet switches are mass produced with huge economies of scale and sell at discounts over chips that are not so widely used. FPGAs, I'd say, are not nearly so widely deployed as ethernet switches and so they cost more because the development and infrastructure costs are spread over fewer customers.

This isn't about process or die size or anything like that. Consider the Xilinx Virtex-7 (only because I could more readily find data for it) and let's compare to a few contemporaries :

• Virtex7 (2011), 28nm, ~6.8 billion transistors, $2500USD (popular models) to $35,000USD (higher end models)
• NVIDIA Kepler GK110 (2012), 28nm, ~7.1 billion transistors, Tesla K20 cards ~$3200USD at launch (chip price some smaller fraction of that)
• XBoxOne SOC (2013), 28nm, ~5 billion transistors, $499 USD for whole XBox at launch
• Xeon E5-2699 v3 [18 core] (2014), 22nm, ~5.6 billion transistors, ~$4500USD

So overall the Virtex FPGA seems reasonably priced (more popular models) compared to other silicon of a similar transistor count, generation, and sales volume. The XBox SOC sticks out as something which was widely deployed in a consumer device and the cost is likewise much lower.

NVIDIA's compute GK110 was much less widely deployed than similar consumer chips that ended up in gaming cards and was similarly more expensive, even given the architectural similarities and the fact that the chips were made in the same factory.

As for the Virtex chips, there isn't a 10x difference in the complexity of the $2500 chips vs the $35000 chips - the latter are simply much less popular and, with lower sales volumes, the cost per unit is necessarily higher.

The market is full of this. Anything you can sell a hundred million of you can always make cheaper than something you will maybe sell a hundred thousand of.

Answer 5

Short answer: because they can be.

FPGA's did not always exist. For them to be practical, in that they could be used to prototype logic for later production as ASICs, they had to be much larger and faster than the logic that would replace them---but they paid for themselves by being programmable. The life cycle of a product that started as FPGA before moving into ASIC was much quicker and cheaper than one that did not. FPGA marketeers were well aware of this, and charged accordingly.

I once worked on a product whose ASIC was anticipated to cost in the $100/each range, and we looked into prototyping on FPGA to shake out the bugs and features in the design, along with providing early customer demos at reduced operational speed. It would have taken at least three of the biggest hottest FPGA's available at the time, at some $2,000 each, to perhaps fit the logic. (Disregarding inter-package considerations, which were considerable. We decided against doing FPGA as the significant additional work to partition into multiple FPGA's would have delayed the actual product release. The costs for a handful of ultra-expensive FPGA boards would have been acceptable, but the delays were not.)

Cheaper/smaller FPGA's might be used for lower-volume productions of modest-sized designs, where an ASIC was never considered due to their substantial (and ever-increasing) NRE costs. These would be much further from the bleeding edge, and much less costly, but still have to be larger and faster than a base ASIC would need to be in order to make up for the radically decreased space and speed efficiency of FPGA logic. Again, the marketeers would definitely be doing their homework and pricing accordingly.

Any piece price for an FPGA that is more expensive than the base silicon costs given the die size and process are due to the above marketing factors. Welcome to capitalism! (Of course, FPGA dies are larger, and on more expensive [faster] processes than would otherwise be necessary. This is inherent. Think of FPGA's as an analog of those snap-circuit educational kits. Much larger and more expensive than a dedicated circuit, but highly flexible.)

Yield is also an issue, as FPGA's are terribly inefficient in terms of logic utilization for their size. It all has to 100% work or there would be 'programs' for them that could not run. At one time Xilinx had a program where, once you had a production design locked down, you could provide them with the configuration and test vectors and they would fit them onto yield failure dies, packaging the ones that passed and charging substantially less than the thousands of dollars each for perfect dies. Of course, the "FP" part of the name no longer applied, as there was no guarantee that any change to the program could run on any particular unit. I do not know if this is still offered by anybody, my guess would be not.

Answer 6

One peculiarity of FPGA pricing is that have seen an assembled board including a FPGA plus memory, clocks and power supply cost less than what distributors are listing.

As an example ALINX SoM ACKU5 FPGA Evaluation Boards & Kits on the Xilinx website has a price of $660.00 and the description contains:

ALINX SoM ACKU5P based on the AMD Kintex UltraScale+ XCKU5P-2FFVB676I, composed of FPGA + 2GB DDR4, 32bit + 64MB QSPI Flash, including all the basic components of hardware.

If I click the "Buy from Partner" link on the above Xilinx web page the price is ￡540.00 (which I think includes VAT).

Whereas if search for just the XCKU5P-2FFVB676I FPGA on some reputably UK distributors get higher prices than for the complete board:

1. DigiKey XCKU5P-2FFVB676I : Unit Price without VAT: £2,690.17000
2. Farnell AMD XILINX XCKU5P-2FFVB676I : Price (ex VAT) £2,825.00
3. Mouser XCKU5P-2FFVB676I : £2,824.97 (ex VAT)

The above prices are all for a quantity of one.

I'm not sure what causes the discrepancy in price between an assembled board and just the FPGA. The XCKU5P-2FFVB676I is supported by the free version of Vivado, so selling such a an assembled board at below cost price wouldn't allow Xilinx to necessarily get an additional income from software development tool licences.

Answer 7

Are they? Just checked at Mouser, a small Lattice ICE40 is exactly EUR 2 @25pcs. The cheapest SoC FPGA (with a hard Cortex M3 core) from QuickLogic is below EUR 6. If you need something bigger, MachXO, Igloo2, SmartFusion2 are quite affordable, too. I'd say, these are ompetitive prices. Same ballpark as microcontrollers with similar performance.

On the other hand, they also list a Xilinx Virtex7 device for EUR 19k! I guess these are produced and used in very low volume, in very special industrial or military applications (or in research projects), where these prices are not unusual and perhaps not significant considering the cost of the whole system. They sell it for EUR 19k because there is demand for that chip at this price. It's certainly not as complex and does not require the same high tech foundries as the Apple M3, the most recent x86 processors from Intel and AMD, and the NVIDIA GPUs and so on.