<< BACK TO SUMMARY

Using Micro-Scope for Memory Testing (Part 2)

To follow up last week’s tech tip that discussed the general principles of memory testing - this week we will look at the specific tests available in the Micro-Scope diagnostic.

Base Memory – The first test to consider, for the lower 640KB of RAM, must be selected during Micro-Scope’s boot-up. To enter the test, press ‘0’ when the Boot Selection menu appears. As we mentioned last week, the Base memory test uses a separate boot-up because the Micro-Scope operating system normally resides in base memory, as do many system values that must be moved and restored, and this simplifies the procedure.

The Base Memory option is a very compact program that only takes up 2KB of RAM, but of course it is not as comprehensive as the suite of tests available in the full diagnostic. However, it does run 3 different routines: a Refresh test, a Walking Ones test and an Address Line test. On those rare occasions when the main Micro-Scope diagnostic program is unable to boot, a hard error in base memory may be the cause, and this test provides a convenient way to locate the problem.

Main Memory Tests

To test RAM above 640KB, boot up to the main Micro-Scope diagnostic (option 1 or 3 on the Boot Selection menu). Select Memory Tests from the Diagnostic menu, and then choose from one of three options: Cache, Expanded Memory or Extended Memory. The same tests are available for each of these. Expanded memory was a work-around for the 640KB barrier and does not apply to any recent systems, so you will normally be concerned only with cache or with extended memory.

The tests described below work essentially the same in both cases, except for extra steps that must be taken with extended memory to ensure that system RAM and not cache is being tested, and consequently the tests will take longer. If time is an issue, cache can be temporarily disabled in CMOS and the extended memory tests will then run faster. It is also possible to select a range of addresses to test, rather than testing all of memory at one go.

Pattern Test – This is actually a collection of 3 individual tests. First, a Walking Ones test is run, which moves an FFFF pattern from one location to the next, doing an immediate read after each address is written. Next is an Inverse Walking Ones, followed by a Checkerboard pattern. These patterns are good at quickly finding stuck bits, but they do not detect cross-linked cells or address lines. They are also subject to line charges, as we discussed last week.

Pseudo-Random Test – Each section of memory is filled with semi-random data and then checked. As second data pattern is then written to each address and that address is checked along with the previous address. This catches cross-linked address lines.

Xor’d Address Test – For a more thorough check of both addresses and data, a test byte is combined with each address, and then stored at that address. Eight passes are made with different test byte patterns so that all bits are checked.

Bit Test – This one sequentially changes each bit of each word in memory. There is a short and long version of the Bit Test. The long version includes an extra toggle, with a cache flush in between. It is quite a lengthy test, but nothing else gives as thorough a check of each cell in memory.

Grid Test – The test pattern detects physically adjacent cells that are linked together, which is the most common example of cross-linking that occurs within a memory chip. Like the Bit Test, it too comes in a long and short version.

One or another of these tests will catch whatever type of memory error that you have. If you’re not sure what type of error to look for, the first choice on the menu is All Tests. It will use the short versions of the Bit and Grid tests, but be aware that it can still take quite some time in a system with large amounts of RAM. For intermittent errors, any or all of these tests can also be selected from the Batch menu, and set to run in a loop until the failure occurs.

About the only memory failure that Micro-Scope can’t find are the one-time errors due to cosmic rays. Fortunately these aren’t common, but about the only way to prevent them entirely is to work in a lead-lined bunker under a mountain somewhere. Thanks just the same, but I’ll stay in the sunlight and use Micro-Scope on the other 99.99%.

Disclaimer - The Micro 2000 Tech Tip is a free service providing information only. While we use reasonable care to see that this information is correct, we do not guarantee it for accuracy, completeness or fitness for a particular purpose. Micro 2000, Inc. shall not be liable for damages of any kind in connection with the use or misuse of this information.