Pages: 1 2
In my experience, defective, poor quality or slow memory modules are one of the main reasons for unstable systems (rivaled only by power supplies). I have personally troubleshot many systems where the problem ended up either being memory that was working in another system (but not quite fast enough to operate reliably in the new system) or a module that failed intermittently but passed all ‘normal’ memory tests. If the user had access to a good memory diagnostic, much time and trouble (as well as money) could have been saved. Similarly, vendors would likely find their costs reduced via the use of a reliable diagnostic, if they don’t already have one.
There are a number of memory diagnostics available that claim to effectively test system memory, but I’ve seen no independent evaluations that I would consider reliable. I have used a few professional level diagnostics from Ultra-X for my own purposes, but have seen several other products recommended in newsgroups and on message boards. While professionals doing a lot of troubleshooting will generally not balk at spending some money on a good tool, end users and some vendors will usually prefer free software. If these tools are not be up to the task, they may end up causing more trouble than if no diagnostic was used at all, because the technician (or user) is convinced the memory is good. I’ve actually seen vendors ‘prove’ a memory module is working by simply booting it to DOS!
When Ultra-X notified me that they had a new memory diagnostic PCI card for the professional (called RST Pro), I decided that this would be a good opportunity to test out several other products on the market, including other tools from Ultra-X. Note that the intent of this test was not to compare features, but to find out which diagnostics would reliably detect a variety of memory errors. RST Pro has a number of advantages over other memory diagnostics, such as ECC validation, that will be looked at in a separate review.
I identified a total of seven different diagnostic products (freeware, shareware and commercial) that have been recommended in various places. These include RST, RST Pro, PHD PCI and QuickTech Pro (all commercial products from Ultra-X http://www.uxd.com), DocMem (freeware from CST, Inc. http://www.simmtester.com/PAGE/products/doc/docinfo.asp), BCM Diagnostics (fom BCM Corp http://www.bcmgvc.com/utilities/windiag.zip), and, MemTest86 (freeware from Chris Brady http://reality.sgi.com/cbrady/memtest86/ ). I also decided to put Burn-In Test (commercial product from Passmark http://www.passmark.com) into the mix. Though not a memory diagnostic, per se, a good burn-in product should detect obvious memory errors. With the exception of BCM Diagnostics, all of these tests used the fully functional product, either supplied by the manufacturer or purchased. The BCM Diagnostics are shareware, and come with a 30-day evaluation, which I took advantage of for purposes of this comparison.
Two of these (PHD PCI and RST Pro) are PCI cards that take over the system after POST, allowing them to communicate directly with the chipset and access areas of memory normally unavailable to an application running under an OS. MemTest86, QuickTech Pro and RST are self-booting utilities, meaning they load their own operating system, which allows much the same access to memory, but without the low-level hardware access. DocMem is a DOS utility, with the associated limitations, while BCM Diagnostics has both a Windows and DOS mode. Burn-In Test is also a Windows application. Of course, the limitation of a Windows application is that a significant amount of memory is reserved by the OS, making it unavailable to the diagnostic program.
Be the first to discuss this article!