Why do you run memtest for xxx number of hours/passes?

[Grenage]

...and how did you come up with the number?

It was the length of the couple of films I watched with the other half.

[mikewhatever]

...anyone else wants to share insights?

[chili555]

Because if your RAM has bad blocks it won't necessarily show up every time that the specific address is accessed. You need to do at least a few passes to be sure that the errors are detected.

The more passes you do the more likely you are to detect any anomalies.

Indeed. And for me, 'more' translates to the time it takes to sip a nice glass of red wine, have a nice eight-hour sleep, wake up, visit the restroom, brush my teeth, check out the market futures on CNBC, pour my coffee and then turn off memtest.

[mikewhatever]

So, what I learn from this very limited discussion is that RAM testing is kind of like religion, lots of opinions, zero knowledge or evidence. Isn't it surprising?

[chili555]

zero knowledge or evidence.

Well, not exactly zero:

less /usr/share/doc/memtest86+/README.gz

9) Execution Time
==================
The time required for a complete pass of Memtest86+ will vary greatly depending on CPU speed, memory speed and memory size. Memtest86+ executes indefinitely. The pass counter increments each time that all of the selected tests have been run. Generally a single pass is sufficient to catch all but the most obscure errors. However, for complete confidence when intermittent errors are suspected testing for a longer period is advised.

Obviously, what exactly is 'longer' depends on many factors.