Subscribe by Email

Your email:

RocketBlog - a discussion about all things related to FPGA verification and debug

Current Articles | RSS Feed RSS Feed

FPGA projects dominate by a 30 to 1 margin

Share on Twitter Twitter | Submit to Digg digg it |  Add to delicious  delicious |  Submit to StumbleUpon StumbleUpon | Submit to Reddit reddit 

Gartner announced today that FPGA projects are dominating ASIC projects by a 30 to 1 margin.  The global economic climate will result in a 22% decrease in ASIC projects in 2009.  This is the second year in a row with significant declines in ASIC projects. 

The general wisdom as discussed in the EE-Times article is that for all but the highest volume projects FPGAs is the technology choice. In my earlier post titled "Mamouth FPGAs Require New Tools", the FPGA technology trend is shows the increase in capacity and performance in the most recent two years.  This change is a key factor for the shift in design projects because, without an option the design teams would have stayed with ASIC technology but now that they have an option - they are going with the lower risk, lower cost and faster time-to-market chip technology choice.  

FPGA tools are constantly changing.  They have to in order to optimize the performance of the design on each specific FPGA architecture.  Unlike ASIC where one tool fits all, FPGA synthesis or Place and Route for example must make effective use of the FPGA resources in order to maximize performance of the design on the FPGA.  The constant change in these tools is the source for errors which are only found in the lab, the very bugs that require countless re-spins of the FPGA in the lab to discover them.  

GateRocket is focused on the growing FPGA market and specifically solving the verification and debug problem.  We minimize the risk of using the large FPGAs by identifying issues before you get into the lab, when they are the least costly and time consuming to fix.  Check out our whitepaper to learn how.  


Comments

I tend to concur. Every year an increasing percentage of my introductory & advanced digital design & verification w/ SystemVerilog students are more interested in FPGAs than custom ICs. Mask costs are becoming prohibitive, unless you are building, say, general purpose memories, CPUs, or cell phone chips.
Posted @ Wednesday, April 01, 2009 8:20 PM by John Eldon
I think this will only get worse from now on. As it is, few companies can afford a fab even at 90nm. FPGAs will be the way to go for virtually all but high volume producers. I guess the question is: how much of a performance/power degradation will we be willing to put up with? Will this mean the end of full blown ASIC for smaller companies or lower volume projects?
Posted @ Wednesday, April 15, 2009 10:55 AM by Patrick Ndai
30:1 seems very high to me, but I've not seen historical data of how this has tracked over time. If you (or Gartner) have that info, it would be interesting to see. Also, does this include FPGAs for prototyping or just FPGAs for production? That would be interesting. 
 
Nonetheless, the larger FPGAs are the size of ASICs of 3 years ago, so I imagine that the same tools applied to ASICs will make their way into the FPGA flow (floorplanning, SI, constrained random verification). One good thing is that the tools for FPGAs are less expensive :-) 
Posted @ Saturday, April 18, 2009 1:03 AM by Harry Gries
FPGA technology will still improve, higher speed, more options on size (smaller as well as bigger ones are needed), and very easy availability is still improve.. ASIC and FPGA will surely co-exist, atleast till quantum devices rule after a decade or so..
Posted @ Wednesday, April 22, 2009 6:10 AM by Ashok Verma
Post Comment
Name
 *
Email
 *
Website (optional)
Comment
 *

Allowed tags: <a> link, <b> bold, <i> italics