What is you longer term vision besides Moor's Law?
There are some really interesting things going on. An alternate way of looking at FPGAs is as sort of the next generation computer architecture. Some sort of blend of today's traditional processors and FPGA. There is a big hole in the memory architecture around FPGAs to be used in the future computer. It is clear that for special purpose applications that people who design those kind of interfaces get tremendous results out of FPGA based computing. So there are still problems to crack there over the next 5 to 10 years that's going to be a very interesting growth area.
Hole in memory architecture?
You've got so much computational power in FPGA but feeding it with data to process is the biggest problem. For specific problems you can custom design a system to feed it. If you have radar or something, you can bring in all that data over some custom designed interface but using it in a more general computing environment the memory interface is the biggest remaining challenge. Having some kind of general solution to that problem solved.
Memory technology or memory interface?
The memory technology says how you manage the interface between this high speed parallel computing engine that an FPGA can be looked at as and the large amount of data, it's going to process. Where is it stored and how do you get at it in parallel? Are you getting the same data over and over or are you streaming? There are a lot of interesting questions in how you get access to all that data as fast as you can crunch on it. For specific applications people design things today and make it work.
There would be significant value in a more general solution?
Yes. I think it is going to take a fair number of years to get solved.
Are there any other topics related to being a CTO that we haven't covered that my readers might be interest in?
The other aspect of it is staying connected to different projects and technologies in the company, trying to get synergies out of the organization as a whole. We have an FPGA team, an ASIC team. Trying to avoid duplicate work going on in the company if for what ever reason they don't talk. There are political issues. People are naturally a bit territorial. So how do you as an outsider stay on good terms with everybody but still give your input over the long haul. You have to have a light touch. They have a lot of pride in their work. You can't just go in and say No, you are doing it wrong. You have to give people a chance to figure things out for themselves.
Have you thought about this rather than you should be doing this
Is the Synplicity development team centralized or spread over the planet?
About 80% is here in Sunnyvale but we have a good group in France, another group in Oregon (small group there) and a fairly good size and growing group in India. We have actually a little group in Turley of all places. Kind of where we found the talent.
How does this impact your activities?
You have to make a bit of extra effort when you have a remote group to spend face-to-face time with them. I have found that if you don't have face-to-face time as a foundation then all the phone calls in the world won't work.
Do you travel to these remote sites frequently?
I go France or people come here. You can still get face-to-face time in different ways. I think you need it. France is a good place to go.
Thanks for an informative interview
Synplicity provides FPGA, Structured/Platform ASIC and cell-based/COT ASIC designers with synthesis, verification, and physical implementation software solutions. The company serves approximately 3,000 direct customers and over 40,000 customers through its OEM partners. Headquartered in Sunnyvale, the firm employs over 280 people in its 20 facilities worldwide.
In earlier commentaries I covered several problems during 2004 regarding Intel with respect to product recall, product delay and cancellation, an so on. The year is now over, so we can look at the bottom line results.
On January 11 Intel Corporation announced record quarterly revenue of $9.6 billion, up 13 percent from the third quarter and up 10 percent year-over-year. For the year, Intel achieved revenue of $34.2 billion, up 13.5 percent from 2003 and higher than the previous record of $33.7 billion set in 2000.
Fourth-quarter net income was $2.1 billion, up 11 percent sequentially and down 2 percent year-over-year. Earnings per share were 33 cents, up 10 percent sequentially and flat with the fourth quarter of 2003. Results for the third quarter of 2004 included tax-related items that increased earnings by approximately 3 cents per share. Intel used $2B to buy back 89 million shares of stock during the fourth quarter.
"We ended 2004 with record revenues and robust demand for Intel architecture products across all geographies and channels," said Intel CEO Craig Barrett. "Our investments in manufacturing capacity, innovative new products and global presence have allowed us to post double-digit gains in both revenue and profits two years in a row. In 2005, we look forward to continued growth as we ramp our 65nm process technology and introduce our first dual-core microprocessors across a range of new platforms."
The company cut inventory levels by more than 18 percent to $2.6 billion, compared with $3.2 billion at the end of the previous quarter. The $560 million reduction was significantly more than the $350 million to $400 million expected by analysts. In fact, Intel executives said during the conference call that inventory levels, which had surged to $3.2 billion in mid-2004, had gotten a little too low during the quarter.
For the year, net income of $7.5 billion was up 33 percent from $5.6 billion in 2003. Earnings per share were $1.16, up 36 percent from 85 cents in 2003. Intel paid record cash dividends of $1 billion, announced two doublings of the company's cash dividend and used a record $7.5 billion to repurchase 300.5 million shares of common stock.
|Table 1 Intel Financial Results for 4Q and the Year 2004|
The architecture unit grossed $3.48 billion in operating profit, up 25 percent, accounting for all of Intel's profits. The communications unit -- which includes flash memory products and its popular Centrino line for laptops -- lost $196 million, narrowing from a loss of $250 million in the previous period.
|Table 2 Quarterly Revenue Breakdown|
On a geographic basis the Americas accounted for 21% of total revenue, AP 46%, Europe 24% and Japan 9%.
For the first quarter of 2005, Intel forecast sales would be in the range of $8.8 billion and $9.4 billion -- in line with analysts' consensus target of $8.9 billion.
On January 11th IBM Corp. announced that it will give open source developers free access to 500 of its software patents. The patents cover areas such as storage management, simultaneous multiprocessing, image processing, database management and e-commerce. IBM said its offer is available to any individual, group or company working on or using software complying with an industry initiative's definition of open source software. The company also said it hopes the move will serve as the basis for an industry-wide "patent commons" that would lead to greater innovation.