Ultra-X, Inc. has created a very impressive software memory diagnostic tool, called RAM Stress Test (or RST, as they refer to it). The diagnostic comes on a single floppy disk, and is self-booting, which allows the software to access memory at the lowest level possible.
RST has two modes of operation – ‘normal’ and ‘burn-in’. These modes are selected by the position of the write-protect tab on the floppy disk. With write-protect on, the software defaults to normal mode, where the user can select which region of memory to test (base or extended), which address ranges to include, the number of loops and which memory tests perform. When the disk is not write-protected, the software automatically starts in ‘burn-in’ mode, which rigorously tests extended memory with most of the possible memory patterns.
There are over 20 industry standard memory tests, including data bus, address lines, refresh, stuck fault and stuck low, and also includes test patterns such as walk left and right, inverted walk left and right, checkerboard, inverted checkerboard, parity, inverted parity, ATC, marching C and others. There is also an option for user defined patterns, and a proprietary address leak test. The address leak test is extremely exhaustive and can run for up to 4 hours per megabyte of memory!! Fortunately, the standard tests will complete a loop in only a few minutes, even with 128MB or more of memory installed.
For our evaluation, we tested numerous FPM, EDO and SDRAM modules that we have available. While not one of these had any bad memory cells (confirmed with a hardware tester which is also under review), we were able to find a motherboard with one bad SIMM slot. Using the burn-in mode, we found that any module placed into SIMM slot 2 on one of our test motherboards would receive intermittent errors, and never in the same place. At first we thought it was the memory, but one of the very cool features of RST is that it displays the actual address (or address line, depending upon the test) that has the error. In this case the address was different every time, but always on the same bank. Swapping modules did not prevent these errors from occuring, but using the modules in a different slot gave good results.
All-in-all, we were very impressed with this software. We were under the impression that no software could possibly give accurate memory test results, but this has turned out to be incorrect. In fact, we were informed by the QA manager of a large memory manufacturer (who asked to remain anonymous) that they use RST to catch errors that even their sophisticated hardware tester sometimes missed. Very impressive indeed.
The main drawback of RST is that you must have memory which functions at least well enough to boot with, so it will not be able to verify ‘dead’ modules. For this, you will need a hardware tester. The good news, however, is that RST is very affordable (our copy was under $100 at the time we received it), and much less expensive than even a ‘cheap’ hardware tester.
We feel that RST is a must for any PC technician or hobbyist that builds or upgrades systems regularly. We also feel that even memory vendors and systems integrators will benefit by helping to reduce RMAs, and thereby improving customer satisfaction. Even overclockers may benefit by testing the integrity of their memory at higher bus speeds. During our own tests, we tried several different PC100 modules (by different manufacturers) running at 112MHz. Though the system booted each time, one of the modules exhibited intermittent errors at this speed, though it performed flawlessly at 100MHz.
As a final note, we have been using products from Ultra-X for over a year (hardware and software). Their products always seem to be of high quality, and their support is superlative (not a comment made lightly!). For more information on RST, and other products by Ultra-X, Inc. check out http://www.uxd.com.
As always, we are interested in your thoughts on this review and this product. Please send any feedback to email@example.com
Be the first to discuss this article!