I had a chance to speak by phone with Guido Arnout, President and CEO of PowerEscape, about the various announcements out of the company this week. Arnout is also Chairman of CoWare and Co-Founder and President of the Open SystemC Initiative (OSCI). He was in Japan on business when we spoke:
“We're here at the EDA Faire in Japan. Because of SARS last year and bird flu this year, these shows are becoming even more important. We're excited about our announcements this week. From digital communications to cell phones, PDAs, digital boxes, camcorders, and base stations, these products manipulate a lot of data from analog to digital. I've observed that more than 80 percent of the power is consumed in the cache memory on-chip or off-chip. Some people disagree, but then I say they should include off-chips memory. Then they say, 'Oh, then 90 percent!' In battery operated devices, it's obvious that something needs to be done. Many devices are so noisy that people worry about both the
ecological and environmental noise.”
“I obtained the CoWare technology from IMEC when I started PowerEscape; both Philips and STMicro have already confirmed the technology. So, we've built two products - PowerEscape Analyzer and PowerEscape Analyzer+Cache analyze the energy bottleneck in C algorithms. In a MPEG decoder, for instance, which is written in C and then partially implemented in hardware and partially in software, it doesn't matter if it's done in hardware or software. Our flow is taking the original algorithm and analyzing it. In a product design that may have 20,000 lines of code, there may only be 400 lines that control how the dynamic energy is consumed. The good thing is that when you do this power analysis
[with our tool], you're also helping with the energy. And it can be done very quickly by people who don't even have to know the structure.”
“We have an OEM relationship with CoWare; our products are a perfect companion to CoWare products. Our target audience is the people who write the code. A digital camera company, for instance, may have 500 programmers on a project. These people have no idea whether their software will be energy efficient or not. Now they can see as their code evolves exactly how it will perform from an energy and performance standpoint. This will help them to become more conscious [of their coding decisions]. I founded this company in February 2003, but we've believed that it's most important to have customers and product before announcing. We already have a customer here in Japan, and we have a lot of customers who want to get their hands on this product and try it out.”
“[Meanwhile], I'm also President of OSCI [Open SystemC Initiative]. We've had an incredible amount of downloads off of our website. Every time we have a new version, there are thousands of additional downloads. The people who use SystemC tend to do more than just hardware. They build platforms and models that people can use to program against. I believe SystemC and SystemVerilog can co-exist. At OSCI, we're engaged in an IEEE process to get the language to be an IEEE language. That will happen in 2004. [Meanwhile], the user base is spreading rapidly in Japan and Europe, although the U.S. is still pretty RTL committed.”
“How do I manage to keep up with everything while I'm traveling? Well, we do a lot of teleconferencing. On Friday, I've got a CoWare Board meeting, and we'll do it from Japan via teleconferencing. But, eventually you just have to be there. When you have a brand new product, you want to look people in the eye. Once you're well established, teleconferencing will do very well. Particularly after 9/11, when you couldn't travel, if you put your mind to it you found that teleconferencing could be effective in combination with travel. But in the early stages of a company, you need to bring the reality to the customers. And that's what we're doing.”
In the category of ...
The caffeine diaries
Speaking of gurus and gadflys - John Cooley's long-awaited DAC 2003 report is out there for all to see. When I got John's e-mail blast notice that it was a bazillion pages long and 7 months after the fact, I said maybe that one better wait for a rainy day.
Well the rainy day finally came, so I decided I needed to go do the deed and plow through the thing. But first I went out and got my fix - a Grande double-shot non-fat extra hot white chocolate hazelnut mocha with whipped cream. Thus armed and ready, I sat back down and settled in to deal with the ESNUG DAC 2003 report.
First thing I saw was that John was right, it's humongous! But the second thing I saw was, I was wrong about the coffee. Because the more I clicked through - and it takes time because there are 43 different sections (no joking) that you have to go to in order to actually see the whole thing - the more I read John's header notes and scrolled through the supporting commentaries, the more I laughed and made those odd nodding noises we all make when we're reading and agreeing with somebody else's caustic criticisms, the more that I thought that this was the best thing since sliced bread.
And the more the caffeine took hold, the more I thought that maybe DAC should just be relocated outright, put right out there on the web where real men speak their minds, and where all of the money it takes to get to DAC, to eat at DAC, to be housed at DAC, and to put spit and polish on (oftentimes less-than-perfect) technology and business strategies so that they'll be fetching/eye-catching once a year - all that money could be so much better spent on a year's worth of crunchy mochas that we could all imbibe in the comfort of our own desk chairs while reading through this kind of stuff on-line.
But, of course, that was just the caffeine talking. It's obvious that there's got to be a DAC. After all, the companies who weren't there - as in, they didn't have a booth - were explicitly pointed out in the report and suspected of being in some kind of trouble. Clearly, if there isn't a DAC, we won't have any way of knowing who's making a go of things, and who isn't.
So here's my recommendation. Sit down and take the time to read the entire ESNUG DAC 2003 report. But skip the coffee. Get a beer instead, and maybe some chips and salsa. Because you'll see it's really not an ordeal at all. It's a party. It's a DAC party.
(My favorite quote from the report? Jim Kemerling of Triad Semiconductor: “As with most high end EDA tools, the product still has a way to go before it is ready for commercial distribution. I noticed a few bugs during the demo. This usually translates into a lot of bugs when the customer is trying to drive the tool on their own. Hopefully they'll not follow the typical EDA industry trend and [will] get it cleaned up before release.”)