Three passes with memtest86+ will weed out errors, at least in my experience and usually it only takes the 1st pass to find something wrong. I know some folks like to run it for hours and hours and hours and I suppose if you're a diehard for stress testing that's fine but in my own experience it doesn't require nearly that much to get a failure to pop up if the RAM actually has some defect(s).
There's a "Windows" memory diagnostic that's part of any Windows installation media aka DVD for the past few years since Vista came out, it works as well as can be expected if you don't happen to have a CD/DVD or USB stick with memtest86+ on it handy. Pretty much every Linux distro LiveCD or DVD (from ISO sources) includes the ability to boot and run memtest86+ as well so that's something to remember as well.
I think I saw something recently that said memtest86+ might have compatibility issues with DDR4 depending on the hardware platform and chipset, I'm not 100% sure on that since I haven't used any DDR4 based hardware yet (desktop or laptop) but it's something to consider. I can't really see any reason to update something that does basic operational testing unless there's an issue of compatibility so the developers of memtest86+ might push one at some point if the DDR4 issues turn out to be actually happening.
I have found at least 1/2 dozen cases (between work, home and others) where no errors were seen in the first 3 passes but the ram was definitely the problem and memtest86+ took day(s) to find the problem. However these all occurred before Memtest86+ had the miltithreaded mode.
2 different diagnostics, like Gold Memory (shareware) and MemTest86.
Many years ago, one person ran MemTest86 or MemTest86+ for days, but neither reported any errors. Then he tried Gold Memory, and it reported a bad bit in 76 minutes, but it had to run for 9 more hours before showing the error again. I'm not saying Gold Memory is better, only that no diagnostic will find every defect, so you should use more than one program.