Retest Counters for RAMCHECK and RAMCHECK LX

INN-8668-APN33
--Created: 8-07-09 (first published for RAMCEHCK)
--Last Updated: 1-07-19

The retest counter is a new feature added to the Basic and Extensive Test. It is designed to monitor transient (or non-fatal) errors that may occur intermittently in memory devices. These errors do not repeat themselves in subsequent tests, and therefore cannot be considered as permanent errors. Sometimes they may be the result of background radiation (also known as "Soft Errors" in the literature), and sometimes it might be just an artifact of the tightness of the timing parameters. Therefore, in contrast to regular memory errors, the RAMCHECK test is not terminated with an error message when such intermittent errors are encountered. It merely counts these intermittent errors in the test log as RETEST counts.

The retest counts are displayed in the test log after you complete each test as "BT RETEST=" for the Basic Test and "ET RETEST=" for the Extensive Test. Each counter is completely independent of each other, meaning that the BT RETEST counter only displays the count for the Basic Test and the ET RETEST counter only displays the count for the Extensive Test.

It should be emphasized again that having some retests does not necessarily mean that the module is bad. It is only a comparison tool used by advanced users to compare modules. For example, you may have four modules removed from a computer after intermittent failure. If all modules pass the RAMCHECK tests but one has an ET RETEST count of 10 while the others have less than 5, an advanced user may replace the module with the high retest count. Note that the replaced module still passed the RAMCHECK tests and therefore may still work well in another motherboard.

The following is a brief description of the retest mechanism. When the RAMCHECK / RAMCHECK LX encounter an error while testing a block of memory, it tries to verify the error by repetition of the test of the same block several times (depending on the actual test phase). Every repeated test increments the retest counter. If the memory block passes the test during these repetitions, the test proceeds to the next memory block. If the error persists, the tester determines whether this is a fatal error or if the frequency is too tight and needs to be reduced. Often, retesting the same block at the reduced frequency is successful and the full RAMCHECK test completes successfully.

For example, if you are testing a module in Basic Test and you get a retest count of three or more then there are two possibilities.

• The tester had to retest several memory blocks without the need to reduce the frequency.
• All three retests were on the same memory block causing a step down in frequency but no additional retests were needed at the reduced frequency. Subsequent tests may revert back to the original frequency as needed. Note that you can tighten the test by setting a fixed frequency via the main setup or the Change-on-the-Fly setup.

The DDR4 adapter uses the retest counters in a different way than the DDR3/2/1 adapters. It will be described in a seperate application note soon.

CONTACTING OUR TECHNICAL SUPPORT DEPARTMENT

For more information, please call us at 1-281-879-6226 M-F 9:00-5:00 US Central Time, or send your e-mail to support@innoventions.com, or fax your message to 1-281-879-6415. Please remember to include your phone and e-mail.