Can't reproduce on win64, try starting with --factory-startup and see if that helps.

Running with --factory-startup makes no difference in the behavior that I can detect. Out of the box it just exits silently.

Ah, it did that four times in a row and then finally started correctly on the fifth try.

It's definitely some kind of non-deterministic uninitialized variable kind of thing.

With today's 2.80 32-bit download I still get the same behaviors. Silently exits unless you add --debug, and then it's working sort of OK, though one file fails to render objects in Cycles, and I get various errors in the process of opening random .blend files (some of which may be old 2.79 format), and it generally crashes when I exit.

Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Memoryblock reconstruct: end corrupt
Error : EXCEPTION_ACCESS_VIOLATION
Address : 0x3F800000

Again, this is just be trying out 32-bits on a 64-bit Windows because other people reported issues with it. I don't have a true 32-bit Windows system to test on, and have no personal need for 32 to work on 64, so my only concern was whether this represents an actual issue which would affect users on 32-bit platforms.

@Ray Molenkamp (LazyDodo) can you take a look at this?

I have, i can't reproduce

@Brecht Van Lommel (brecht) do you guys have any windows computers to reproduce this on?

If it doesn't reproduce for LazyDodo then I'm not sure it's worth worrying too much about at the moment, unless and until a real 32-bit user comes along with issues, so I have no problem if you want to close this one.

There might be some fundamental 32-bit environment issue here that I don't normally see because I don't run similar 32-bit applications. As long as it works for someone then I'm not going to worry that it's completely broken which was my original concern.

seems similar to T59926
Haven't realized, this may be only present in 32bit build. i switched to 64, haven't noticed bug anynore and closed it.
I can still reproduce also T59927 on 32-bit

Does it work if you run the 32bit version in compatibility mode with win7 (32bit) ?

I've had an other user report that on win7 64bit he also got this issue, but after running it in compatibility mode, this doesn't happen (even when switching it off afterwards).

Seems like people had problems like this with java too https://stackoverflow.com/questions/8921607/jvm-32-bit-crashes-on-64-bit-machine

Quick tests don't show any apparent difference with Windows 7 compatibility mode.

I never bothered setting to Win7 compatibility.
The default target in Win7 is WinXp SP3.
For my Win10 (laptop) test, I used the default Win8 target.

The compatibility switch on/off thing didn't stick on Win10.

as for me the "compatibility with windows 7" trick works

The 32 bit build also crashes sometimes on startup for me, but not consistently.
Sometimes I also get

Memoryblock reconstruct: end corrupt

in the log.
Would it help if I did a 32 bit debug build myself to see where it crashes?

works great for me now, no crash anymore (so far).

No crashes on start-up for me.

I can confirm that it appears fixed here too! 32-bit running on top of Windows 10 64-bit.

Thanks Brecht!