Is it normal that successive passes by memtest86 are taking longer?
It took 35 min to complete one pass (the default config was for four passes) so I took a leap and assumed that the test would be over in ~105 minutes.
I came back 3 hours later, expecting it to be finished; however, it's only halfway through the third pass! (Zero errors, thankfully). What's the explanation?
Spec-wise: am running memtest86 off of a bootable USB on a Lenovo X220, 64-bit i5, 8gb RAM.
From memtest86 v4.00 onwards, the first pass runs half the number of cycles of subsequent passes. The idea is to first perform all the different tests as fast as possible, so that if you have a problem which only crops up in one of the tests you get the error as soon as possible. Subsequent passes are more targeted towards stress/endurance testing.
Memtest86+ V4.00 changelog
- Major Architectural changes
- First pass twice faster (reduced iterations)
- Detect DDR2/3 brands and part numbers
- Corrected detection for Intel "Lynnfield" CPU
- Added detection for Intel "Clarkdale" CPU
- Added detection for Intel "Gulftown" CPU
- Corrected detection for AMD 45nm K10 CPU
- Added detection for AMD "Magny-Cours" CPU
- Solved crash with AMD Geode LX
- Added detection for Intel XMP Memory
- Added for CPU w/ 0.5/1.5/3/6/12/16/18/24MB L3
- Added "clean" DMI detection for DDR3/FBDIMM2
- Better detection of Integrated Memory Ctrl
- Complies with SMBIOS 2.6.1 specs
- Fixed compilation issues with gcc 4.2+
- Many others bug fixes
More here.