In case anyone else sees the same issue.
In MemTest86, V4.2 there is a bug in Test #8, which is the Random number sequence test.
If a single error is found by MemTest86 in Test #8, then there is a cascade effect and many thousands of false errors are rapidly produced.
The first error will be a real RAM fault. But the subsequent additional errors in Test #8 are not.
The bug has been fixed and will be available in V4.3 of MemTest86.
The problem was that some of the internal variables used in Test #8 where being overwritten by the code in the error handling function. So after displaying the first valid error, the test starts to produce false results.
There was also a related issue that the display of the average number of error bits was wrong when there is a large number of errors (> 65,000 errors). This was also fixed.
In MemTest86, V4.2 there is a bug in Test #8, which is the Random number sequence test.
If a single error is found by MemTest86 in Test #8, then there is a cascade effect and many thousands of false errors are rapidly produced.
The first error will be a real RAM fault. But the subsequent additional errors in Test #8 are not.
The bug has been fixed and will be available in V4.3 of MemTest86.
The problem was that some of the internal variables used in Test #8 where being overwritten by the code in the error handling function. So after displaying the first valid error, the test starts to produce false results.
There was also a related issue that the display of the average number of error bits was wrong when there is a large number of errors (> 65,000 errors). This was also fixed.
Comment