XC7V2000T, the perfect Thanksgiving gift

I guess they are only getting one good chip per wafer... after you add in the cost of development the chips are a little on the expensive side.

I worked for a test equipment maker once and they were using the largest chip Xilinx made at that time which was a bargain at $1000. But then the box it went in sold for $150,000 so the chip cost was lost in the noise. I think they only expected to sell ten or so which means they barely covered their development costs. The funny thing was they were only using 10% of the chip, the rest was for "future expansion". That's how little they cared about the cost.

Rick

Who can afford these things? The NSA? How many of these are they going to sell a year, a dozen?

Jon

That sounds like a good project for a MBA type case study.

I could easily believe that $10K is cheap insurance. How much would it cost to figure out what size chip to use? Don't forget lost opportunity cost as well as the more obvious costs. Do you want your best guys spending time on figuring out what size chip to use or getting the project ready to ship?

The prices there are probably worthless. Make a serious approach to a Xilinx distributor, and you will get a completely different price. Still expensive, but a lot lower. That happens with both Altera and Xilinx, as they have full control of customers and prices. Each customer gets its own price, based on forecast, risk and a few other parameters.

And I should add, the next chips will progably be even cheaper. It is normal to give a step price model to customers.

This is why I always buy FPGAs in the jumbo economy pack.

Bob Perlman Cambrian Design Works

That's a pretty good price for parts that are made out of pure unobtainium.

The same people that spend that kind of money on a scope or logic analyzer. I guess these devices are very handy for prototyping ASICs.

The most expensive Virtex I ever worked with was $1000. At that time my employer let us re-design the product using a couple of Spartan 2 devices.

Am 24.11.2011 12:33, schrieb Tim:

Think of how much money you have to invest, in order to fill 2.4 million LUTs with useful logic. Then the price appears in another light.

Matthias

You design just once, but maybe want to sell many units... But at least the prices are precisly calculated, look at the 6 cents of the "low-end" part.

But seriously: When you look what you can do with a 40kLUTs part, then designs that large are either bad/inefficient, brute-force-approaches ("do the same thing 100 times") or incredible huge projects. So I guess, the real useful application is ASIC-prototyping, so you are right...

Thomas

Hi,

Matthias Alles skrev 2011-11-28 12:24:

That D´depends on how you code :)

Imagine using a schematic tool and primitives only exept for a few ROMs for microcode THEN the V7 price becomes a bargain;)

/michael