SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Rambus (RMBS) News Only
RMBS 107.76+1.2%Nov 7 9:30 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: REH who wrote ()1/9/1999 8:49:00 AM
From: REH   of 236
 
Buyers gird for Rambus revolution
M. Jamal Arif

Density and technology have evolved rapidly from the earlier single-in-line memory modules to today's 168- and 200-pin dual-in-line memory modules for the current PC-100 SDRAM devices.

Moving from one generation of SIMMs and DIMMs to another has proven to be evolutionary, with no problematic areas to speak of for either the DRAM supplier or OEM customer. However, as the purchasing community prepares for the Rambus generation targeted at PC main memory, it will learn that the enabling effort for implementing Rambus-in-line memory modules (RIMMs) will be revolutionary in nature.

Going from the PC-66 to PC-100 SDRAM required about 6 to 12 months of effort for the industry. But, in the case of Rambus and its revolutionary implementation, purchasing executives can expect an extended transition.

There will be lots of issues. Making sure of the reliability and quality of the Rambus product will require a considerable amount of hand-holding between Rambus and its licensees, such as Hyundai. The shift to Rambus will also require DRAM vendors and OEM customers to work together closely.

Despite what the most optimistic and well-intentioned pundits are saying, Rambus devices and RIMMs are expected to run through a gauntlet of cost, time, and yield concerns.

You can expect to pay more for a RIMM than what you've been paying for the conventional DIMM. There will be a premium for each Rambus device, and there will be eight of them on a RIMM. Moreover, each device comes in a micro-BGA, which introduces some challenging and expensive packaging issues.

Testing the RIMM will also incur additional cost, and assembly will be expensive at the start. RIMM testing by the DRAM vendor will be of paramount importance to the purchasing community to ensure the elimination of potential pattern sensitivities and noise problems. That's one major reason it won't be easy for third-party module makers to attempt to build RIMMs. Factory resources for RIMM testing are critical-and third-party houses just don't have those resources. Altogether, these factors equate to a relatively expensive module compared with a DIMM.

Time and yield are the other two major hurdles. By the nature of the Rambus technology and the lengthy interaction that has occurred between Rambus and its licensees, and between Rambus DRAM vendors and their OEM customers, it will take time for the product to become pervasive in the market. The yield aspect of Rambus will play a major role as well in bringing down the cost and making the technology more acceptable to PC OEMs.

However, as most veteran purchasing executives know, the DRAM market can shift from what appears to be a forerunner to a better price/performance/availability technology within a matter of weeks.

Other modules for other market segments will continue to evolve with movement in density and newer configurations. There are various types and flavors depending on the targeted market segment.

The PC, server, workstation, and graphics segments of the DRAM market each have unique and different requirements. For instance, the small-outline DIMM is built for notebooks and graphics systems because of height requirements.

Most standard DRAM modules are developed and built by major DRAM suppliers. However, some module vendors buy loose components and build the modules themselves, and then, in some instances, subcontract the work to third parties to assemble the modules.

Some third-party module-assembly companies make quality memory modules, but the majority do not. In most cases, the printed-circuit boards are of low quality, due to the general practice of cutting corners to reduce cost.

Although some price points are lower than others, price is generally a function of the module's quality.

But the major DRAM vendor's pricing is generally higher than other module makers. That's because the DRAM vendors extensively test these modules in the assembly line, whereas other module makers usually apply a functional go/no-go test.

If the module works, that's fine. If not, the user can exercise the vendor's 10-day return policy-no questions asked.

So the purchasing executive should ask for additional information regarding the total package associated with a memory module before buying it. The following questions should be asked:

- Is the memory module guaranteed to meet the specific requirements of the system?

- Will the module work reliably with the system? By what standards and to what degree?

- What type and level of testing has been applied to the module?

- Who built the module?

Finding answers to these questions will prevent many executive and buyer headaches down the road.

-M. Jamal Arif is senior manager of tactical marketing at Hyundai Electronics America, Semiconductor Division, San Jose.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext