Chipmaking: June 2008 Archives

On Semiconductor is pushing more of its products out to foundry, naming Israel-based Tower Semiconductor as its supplier, even though On Semi has a foundry operation of its own. What's more, the deal seems to go further than a classic foundry deal as Tower will "co-develop and manufacture multiple lines of products and technologies".

On Semi already uses Tower as a foundry for analogue products — they originally signed a five-year supply deal in 2001 and Tower picked up a supplier award in 2005 — so it looks as though this deal will see the Israeli foundry do more of the actual product development work, presumably concentrating on process tweaks and perhaps back-end design, while On Semi provides the product definitions:

Tower and ON Semiconductor are to jointly develop new, best in class, manufacturing technology platforms, leveraging each company's core expertise to achieve a highly competitive, cost-effective product offering. These products are to be manufactured in Tower's Fab1 facility.

In its existing deal with Tower, On Semi uses Fab1 to produce a range of analogue parts. The 2001 deal gave On Semi the ability to use up to a quarter of Fab1's capacity. Itzhak Edrei, executive senior vice
president for worldwide product lines and sales at Tower Semiconductor, said in the release: "This added layer of collaboration between both companies should significantly increase our manufacturing volume of On Semiconductor products."

With the big guns such as TSMC and UMC controlling much of the market for high-integration digital products, the smaller foundries need to concentrate on less flashy areas, such as mixed-signal with a big 'A' and small 'd'. However, this type of work needs more process tuning, which is why fabless companies have found it difficult to make headway against the integrated device manufacturers. Getting closer to the foundry should make the fab-light approach work better, although it effectively locks the customer in and results in risk for the foundry as well, as it demands that the manufacturer invest in tuning processes for devices that might not sell so well.

Intel came close to giving the idea of having a fixed clock-speed rating on its upcoming Nehalem the heave-ho, according to Intel fellow Rajesh Kumar, speaking to journalists ahead of the VLSI Circuits Symposium in Hawaii this week. The people who were going to be putting the processor into PCs didn't care for the idea, it seems.

The company has radically altered the way that Nehalem is clocked compared with its predecessors in order to improve both memory bandwidth and power consumption. It means that the core, memory buses and I/O run almost independently.

The bigger change is internal, where it seems that the concept of a fixed clock running at several gigahertz has been discarded in favour of letting the logic run at its own speed. This is something that people such as former ARM architect Professor Steve Furber have been advocating for years. The concept of a system clock is entirely artificial and exists largely to make life easy for chip designers and simplify the job of testing chips as they come off the production line. Chips such as the Amulet don't run off any kind of clock: the logic inside finds its own speed.