September 12, 2005
Who Is Using ESL and Why?
Please note that contributed articles, blog entries, and comments posted on are the views and opinion of the author and do not necessarily represent the views and opinions of the management and staff of Internet Business Systems and its subsidiary web-sites.
Jack Horgan - Contributing Editor

by Jack Horgan - Contributing Editor
Posted anew every four weeks or so, the EDA WEEKLY delivers to its readers information concerning the latest happenings in the EDA industry, covering vendors, products, finances and new developments. Frequently, feature articles on selected public or private EDA companies are presented. Brought to you by If we miss a story or subject that you feel deserves to be included, or you just want to suggest a future topic, please contact us! Questions? Feedback? Click here. Thank you!

What are people doing with ESL tools?

They tend to focus around algorithm design, co-design, verification and modeling of systems using these high level languages. Those are still the most popular uses. Certainly in our business we are seeing a lot of adoption of C based synthesis by non-traditional customers. 80% of our customer base would probably fall into the system and algorithm designer category with a smattering of software. Only about 20% of our business comes from traditional hardware design. In our business we are seeing every day the growth every day in non-traditional EDA. A lot of what people are using these tools for is the ability to take an algorithm and split it out between a processor and custom hardware, particularly programmable hardware but it could also be structured ASICs design in order to get better performance of their design in terms of speed, power, throughput and just the efficiency of usage. A lot of the hits on FPGAs in the past as programmable logic devices are for example that they are more expensive and have lower performance than an ASIC. While that's still going to be true, if you look at the cost of getting into an FPGA design in the future, they are minimal compared to an ASIC design. If you look at the cost of getting into an FPGA versus say another processor, the FPGA is much, much cheaper and actual has the capability of accelerating the performance
because of the nature of the custom logic you put in there.

We're seeing the applications reflect that. Instead of people taking a design and adding another Pentium or ARM processor or very expensive chips, they are actually going to augment that by putting in an FPGA device. An FPGA might be $100 per part but it is 10x cheaper or one-tenth the price of that processor they were going to use before because it tends to accelerate things in a more parallel fashion than the processor to give you a much better performance increase. We are seeing 10x to 100x performance increases in algorithm acceleration and much lower power. It's a better power picture than you get from a processor. Because of the ESL tools we give them the capability to program it
with software. So it provides the advantages of flexibility. That's why we are seeing more adoption among non-standard hardware people. We are making this custom logic available as a custom coprocessor to design. The numbers we are seeing here are kind of showing that.

What are some of the additional things that are required in order to encourage ESL adoption?

Certainly training, methodology and standards showed up big time as things that needed to be put in place. I think as the industry matures it will do that. The industry also has to recognize that the software people obviously need a different type of training than the hardware people when you allow them to do this type of design. Those are the challenges we face.

Why would people choose ASIC versus FPGA?

We are seeing that the FPGA versus ASIC trend still holds basically due to the non-recurring engineering costs and performance. If you need something with high performance and you can't get away with an FPGA, then you choose ASIC. The NRE might drive you from ASIC towards FPGA. We looked at some additional factors as well like reconfigurability, prototyping and software acceleration for why people might choose FPGA over ASIC. One of the encouraging things for the FPGA guys is that very few people, maybe 10%, consider FPGAs as glue logic. Most people are considering them as integral processors in their designs. On the design side the trend shows that there are very few people in our survey that are using FPGAs as programmable logic devices that are very small. In fact there are a lot more people using ASIC designs that are very small. The FPGA guys can feel secure that if they out big FPGAs there will be people there to use them because there is demand for as many Gigs as you can get in these coprocessor type developments. Performance wise FPGAs have obviously less performance than ASICs but surprisingly we are seeing pretty good correspondence between the two. There are still more people hanging around the 100MHz and 250MHz node and FPGAs are well able to handle that. It's not necessarily always performance that is driving the switch from FPGA to ASIC. In fact
people are often able to move the other way because of the costs and still get their performance at this point.

How does the cost of FPGA tool compare to versus ASIC tools.

We still track that because that is important for our business. ASIC people are still spending a lot more money than FPGA guys and that holds true in our survey. We're encouraged to see that a good majority of people are still willing to spend over $25K. If there are 10X more people that are able to use FPGAs even at $10K and above, that represents a huge market for FPGA that we can tap into. It's a new type of market. You have to get your value proposition aligned with what these people are able to pay. It's good news from the survey as well.

Use of embedded cores

The majority of designs are using some form of core inside the embedded system. It doesn't necessarily mean that it is an embedded on-chip core. Most people in the survey use or plan to use some form of embedded core, hard core of soft core.

FPGA, off chip microprocessors and DSPs are still the most popular. The majority of people are using some form of processor in their system which is what we expected to see.

One of the things that surprised us is that we are not seeing too much pickup of other things than FPGAs in terms of reprogrammable fabric. It's still kind of early in the day where somebody can expect to see some form of custom reprogrammable silicon like Elixent, IPFlex or where say Toshiba has a reconfigurable fabric on their device. We are not seeing a whole lot of demand yet for those. Only 8% are using and only 24% are intending to use. So early days for things other than FPGAs as far as reprogrammability is concerned.

That bodes well for the FPGA guys. It means that there a tough road for those people doing reconfigurable fabrics or reconfigurable processors. We are not seeing the adoption there.

Why would someone augment their microprocessor or DSP processor with an FPGA?

We are finding good adoption across the board here. Flexibility and performance are the main issues. People want the capability of getting the best performance they can flexibly across the processor and the hardware. That's why they are augmenting their processors with FPGAs.

Compared to last year's survey what were the most significant changes?

The most significant in terms of what we were looking for was that the adoption of ESL is much greater than it had been. It's shown very good continued growth. The movement towards more algorithmic and systems people using custom logic because they were enabled by ESL. There was also a continued growth pattern. The continuing of the trends si really what we are seeing here.

So extrapolation of prior trends rather than any big changes?

There hasn't been any step function here, no sudden switch. The trend is continuing.

Were there any surprises?

I can't say that there were any real surprises. We were a little surprised to find out that there are more people looking at reconfigurable processors or reconfigurable fabrics that companies have but in the end proprietary solutions are always a tough sale. Those will be proprietary until they get more adoption. I guess that was a bit of a surprise.

The fairly significant dip in planned HDL use was a bit of a surprise over last year. In last year's survey they expect it to basically hold steady. The same people that were using HDL expect to continue to use it. This time we saw a kind of dip down from 52% to 45% or so, a fairly significant dip. A group of people now using HDL that are now expecting to use ESL languages, presumable System C since that one had the biggest increase in planned usage. That's actually a surprise.

« Previous Page 1 | 2 | 3 | 4 | 5  Next Page »

You can find the full EDACafe event calendar here.

To read more news, click here.

-- Jack Horgan, Contributing Editor.


Review Article Be the first to review this article

Featured Video
Senior Electrical Engineer for Allen & Shariff Corporation at Pittsburgh, Pennsylvania
Upcoming Events
SEMICON Southeast Asia 2018 at New Malaysia International Trade & Exhibition Centre (MITEC) Kuala Lumpur Malaysia - May 22 - 24, 2018
Methodics User Group Meeting at Maxim Integrated 160 Rio Robles San Jose CA - Jun 5 - 6, 2018
IEEE 5G World FOrum at 5101 Great America Parkway Santa Clara CA - Jul 9 - 11, 2018
PCB WEST 2018 Conference & Exhibition at Santa Clara Convention Center Santa Clara CA - Sep 11 - 13, 2018
TrueCircuits: IoTPLL

Internet Business Systems © 2018 Internet Business Systems, Inc.
25 North 14th Steet, Suite 710, San Jose, CA 95112
+1 (408) 882-6554 — Contact Us, or visit our other sites:
AECCafe - Architectural Design and Engineering TechJobsCafe - Technical Jobs and Resumes GISCafe - Geographical Information Services  MCADCafe - Mechanical Design and Engineering ShareCG - Share Computer Graphic (CG) Animation, 3D Art and 3D Models
  Privacy PolicyAdvertise