Pass Completion Time

[Station]

I've been using 8.1 for the fast few days to find errors in my RAM, which I suppose exist.

Since the free version is limited to 4 passes, I always run 3 * 4 passes per module which amounts to about 12 hours of testing.


What I find odd is that the first pass is always much quicker to finish than the other three ones. Reading the forum, I noticed a statement confirming this, but WHY is this the case?

Data points, for 4 passes on my machine (actual specifications shouldn't matter that much):

Pass No	Time To Complete
1	36 minutes
2	1 hour 4 minutes
3	1 hour 5 minutes
4	1 hour 5 minutes

Passes 2 to 4 feature a quite common completion time, but pass 1 is way shorter. What's the reason for this discrepancy?

I figure there is no way to disable whatever is being skipped in the first run (if something is skipped, that is)?

[David (PassMark)]

It is by design.
If gives people the option of a quick test (36 min in your case), while still cycling through all the patterns.

[Station]

May I ask how this is achieved? Obviously it must be leaving something out, like jumping over specific memory regions (since it takes about half the time of the other tests, it sounds like it would just stripe half of the memory space as by using modulo 2?) and only testing a subset of them or using only a subset of the normal testing patterns in the actual tests?

I'm not complaining, merely wondering and trying to understand what the technical difference is.

I also think this should be mentioned more prominently, because I've only read that in the forums so far - but not documented anywhere else. To get a thorough check, more than one pass is pretty much a given.

[David (PassMark)]

All the ram is still tested, but there are less internal iterations in the 1st pass.