Who Is Using ESL and Why?
What strategy does Celoxica have to accelerate the ramp product wise or marketing wise?
We are working with very specific customers, particularly in the System C space. We work with several customers to develop our tools along with their System C methodologies, so that when they are ready to roll out System C to their design environment, we are part of that. As far as System C and System C synthesis goes, it is still an early adopter market. For the C synthesis at large the algorithm acceleration we are starting to see hitting various bowling pin markets. We are working to complete solutions in vertical markets. For example, imaging is one of those where we have worked. Algorithmic acceleration of visual computing is where we are working to create an integrated vertical solution, to hit the bowling pin - we tend to use Geoffrey Moore terminology. Thirdly, we have enough of those successes then we think the mainstream market will follow.
Are there any sweetspots for Celoxia? By geography, by industry segment, by application, ..?
If you had asked me a year ago about sweetspots, we probably would have told you Japan. As it turns out over the course of the last year the US is starting to catch up and we are seeing a lot of adoption here locally. Europe is still a little reticent although. We have had some good growth there as well. Growth is coming all over the board. Sweetspots? I would have to say in terms of industries that there are niches where you have for example algorithm development for HDTVs around various IP blocks like JPEG 2000 and HDOT 64. People are looking for solutions there that are flexible and optimal. They are often adopting this methodology in order to get into those markets. This provides a good opportunity for us. Defense has been very good in the US because they tend to be algorithmic in nature and also predisposed to suing programmable logic as opposed to a two year long custom ASIC development. We've done very well there. A lot of their problems are imaging in nature such as target acquisition, robotic vision. Things like that are very imaging oriented, so they lend themselves very well to acceleration.
Are your products more attractive to those applications with a large number of gates?
Not so much the number of gates that is the driver. It's the complexity of the algorithm. If you've got an algorithm that's so complex that you can not write it using a hardware methodology or it is so complex and the performance is so tricky that you have to accelerate it, if it is not something that you can just not run on a processor, then you have to accelerate using hardware. Those are the drivers for adopting this type of technology. We are providing a way to do algorithm acceleration by splitting things off from a processor and dropping programmable logic along side that but still using a C based methodology to program it. So there are applications of course where that won't necessarily fit. There are people doing just fine with processors only and those people have not adopted ESL yet. If we are coming to a place and someone says that we want to benchmark you against RTL, we say that's fine and we do a lot of that. The real answer is that if you can write your design in RTL then you probably don't need ESL yet. The real benefit is when you've got something you can't write in RTL or where you can get much better productivity and benefit as you can see from the survey. It's really performance more than productivity even. You get better performance by splitting it across both hardware and software. Those are the applications where we get in first.
Celoxica has both software tools and FPGA boards. Do most of your customers use both?
Probably a good half of our customers buy a complete solution where they will buy both software and boards. That's part of why we have that. It helps leverage a complete solution. Only about 20% of our business comes from the board side. It's significant but not the major driver. Software is the main driver of our business.
Over time do you expect the hardware side of the business to become smaller?
For us it will probably stay steady and then start to shrink off a little bit. We have a number of partners that provide that type of hardware. For example we did a press release earlier this summer with Silicon Graphics. Their hardware happens to be in the form of a FPGA board that ties into a supercomputer and we provide the programming environment. But basically it's the same sort of thing, they're providing the hardware and we are providing the software. We have other customers that have single board computer solutions like Matrox, SBS and Creko. Those companies provide FPGA cards. They are using acceleration but they use our software. That's fine with us. Our main focus is our software.
Are the boards of your own design and manufacture?
Yes. We do have boards that we manufacture. We have PCI based cards that have Xilinx and Altera chips, usually the latest and greatest. We have some Ethernet based cards or desktop cards that are used for development and prototyping that have a lot of bells and whistles like LCD display, audio/video in and out. Basically all the things you would need to prototype a system. The reason we do the cards ourselves is that they serve as great platforms for showing people the capabilities of prototyping or FPGA development using this type of software. They also make a great demo for us as well. We will always have some of our own cards because it completes the solution. We can offer and create cards that are specific for completing vertical market solutions. In the end it is leveraging other people's hardware that is probably even more powerful for us.
The top five articles over the last two weeks as determined by the number of readers were:
Synopsys Fires Again at Magma (Electronic News Magazine) In his ruling U.S. District Judge Maxine M. Chesney adopted Synopsys' positions outright on three of the four claim construction questions at issue in the patent lawsuit brought by Synopsys Inc. against Magma Design Automation Inc.
Survey Shows Hidden Market for ESL and FPGA; Electronic System Level Design and FPGAs Are the Big Winners in This Year's Worldwide Survey on Designer Trends See this week's editorial above.
AWR Acquires European EDA Developer APLAC Solutions Oy APLAC develops and markets simulation and analysis software for analog and RF design. APLAC's RF design technology has been widely used by Nokia Mobile Phones for years, and has been used in designing over 30 percent of all mobile phone RF ICs worldwide
Mentor Graphics Announces Product Certification for 64-Bit Red Hat Enterprise Linux Platform Mentor's entire line of Eldo and ADVance MS analog and mixed-signal products have been certified for operation on Opteron and EM64T processor architectures using the Red Hat Enterprise Linux 3 platform.
IEEE-USA COMMUNIQUE TO U.S. IEEE MEMBERS ON THE KATRINA DISASTER IEEE encourages U.S. IEEE members, who want to make personal donations to relief or reconstruction organizations, to give to any of the organizations identified by the Federal Emergency Management Agency (FEMA), listed online at: http://www.fema.gov/press/2005/resources_katrina.shtm#donate.
Other EDA News
DATE presents a new initiative on System Design Records
Berkeley Design Automation Joins the Cadence Connections Program; PLL Noise Analysis Capability Integrated into Cadence's Virtuoso Analog Design Environment
ITRI Tapes out Low-Power DVFS Test Chip With Cadence Encounter Synthesis and Implementation; Encounter Low-Power Design Methodology Helps Slash Power Consumption up to 40%
Be the first to review this article