The notebook memory-module flaw that HP identified last week as being an "industry wide" problem only seems to be affecting HP products.

Last Friday, HP announced it would offer a replacement for certain notebook memory chips from Samsung, Micron and WinBond found in about 900,000 of its laptops, and warned that many other manufacturers' laptops would be affected.

However, Samsung has not received any reports of problems or flaws from customers other than HP, a spokeswoman said. IBM has said it discovered the issue while testing memory chips in its Thinkpad notebooks, but said that it has disqualified those parts and had had no reported problems.

Likewise Gateway, which said it has not received any reports of the problem from customers, and its newly acquired subsidiary eMachines had discovered and eliminated the problem. Dell has not noticed this problem on any of its systems, said a spokeswoman. And Toshiba said it was investigating the problem, and plans to issue an update on the status of its notebooks in the coming days.

Representatives from the affected memory module manufacturers Micron and Winbond did not respond to repeated requests for comment.

This apparent failure of HP to recognise a problem before shipping its products - as its competitors claim to have done - has already got US lawyers excited about the prospects of a class-action suit, according to one news report. One example of the terrific US legal system in action can be seen here, although whether those affected will be able to get online is another story.

HP has put a program up on its website so customers can determine whether they have a flawed memory chip. The affected HP notebooks cut across several product lines and were shipped over the last two years.

The problem
The flaw occurs when a PC attempts to re-enter an active mode from a sleep mode. Under certain conditions, the memory module can hang and cause the system to crash, resulting in the loss of data.

The sleep mode that must be enabled to induce the flaw is known as C3. It is the deepest level of sleep that a processor can enter in order to save power, according to documentation for the Advanced Configuration and Power Interface (ACPI) standard used by PC and processor vendors to manage power consumption. Some notebook vendors put the processor into this state when the user presses the sleep button, or when the notebook is left idle for an extended period of time.

Some notebook vendors also use the C3 sleep mode to actively manage power consumption by putting the processor into C3 mode thousands of times a second during gaps in application activity. The likelihood of the system crashing is much greater when the C3 mode is entered and exited so frequently, said Ronald Kasic, director of customer engineering and sustaining marketing for HP.

In order for those crashes to occur, the C3 sleep mode must be used in conjunction with the 845, 852, or 855 mobile chipsets from Intel, processors that support the C3 state, and certain memory chips from Micron, Samsung, Winbond and Infineon, HP said.

The Infineon chips affected by the flaw were a limited number of 256MB DIMMs (dual inline memory modules) shipped in the first part of 2003, Infineon said in a statement. The chips made up a very small portion of the notebook DIMMs shipped by Infineon during that period and no other memory modules have been discovered with the flaws since then, the company said.

Most PC vendors don't use the C3 sleep state because the lighter power-saving states provide the greatest overall benefits, said Roger Kay, vice president of client computing with IDC. For example, the C3 state requires the operating system to manage certain parts of the chip while in that state, while the lighter states don't require any intervention on the part of software, according to the ACPI documentation.

However, HP described the combination of the C3 sleep state and the Intel chipsets as a "very prevalent architecture used in the industry" on its list of frequently asked questions about the memory issue on its website.

HP deserves credit for going public with the problem and making it easy for customers to find out if they might be affected by the problem, said Stephen Baker, director of industry analysis for NPD Techworld. But the fact that no other vendors have reported problems and that two vendors discovered similar problems in testing raises questions as to whether HP simply missed the issue in its own testing, and attempted to portray the flaw as an industry issue to avoid that perception.

"Best case scenario, they thought this would be a problem for a lot of people. Worst case scenario, they recognized they were the only ones doing this and wanted to deflect attention from the fact that they missed it," Baker said.

HP didn't discover the flaw during preliminary testing because the probability of the issue occurring is extremely low and dependent on the user's configuration, it said on its memory recall Web page. "We stand 100 percent behind the claim that this is an industry-wide problem. If you're using those core components, you have the potential to be impacted by this," said Mike Hockey, an HP spokesman.