https://www.theregister.com/2021/06/04/google_chip_flaws/

Today's computer chips are so advanced, they are more 'mercurial' than precise – and here's the proof
Rarely seen miscalculations now crop up frequently at cloud hyperscale

 

https://www.theregister.com/2021/06/04/google_chip_flaws/

Today's computer chips are so advanced, they are more 'mercurial' than precise – and here's the proof
Rarely seen miscalculations now crop up frequently at cloud hyperscale

Loved the last paragraph of the article:

"In our recent experience, roughly half of these human identified suspects are actually proven, on deeper investigation, to be mercurial cores – we must extract 'confessions' via further testing (often after first developing a new automatable test)," Google's researchers explain. "The other half is a mix of false accusations and limited reproducibility."

 

While cleaning up some old stuff I found my old 368 (original Compaq Deskpro 386) that didn't count or run 32-bit software.
https://www.pcjs.org/documents/manuals/intel/80386/
http://chipdb.org/cat-16-289.htm

I think the one on the left has the bug.
The 386 on the right has the double sigma mark to it's been tested for 32-bit operation.

There are three distinct species:
- no marks
- "16 bit s/w only"
- double sigma

 

https://www.theregister.com/2021/06/04/google_chip_flaws/

Today's computer chips are so advanced, they are more 'mercurial' than precise – and here's the proof
Rarely seen miscalculations now crop up frequently at cloud hyperscale

When he says "age", does it mean literally time elapsed after being produced or time actually powered ON (running)?

 

It usually means active run time. You can accelerate 'age' by using elevated temps and voltages.
https://www.asset-intertech.com/resources/blog/2013/08/silicon-aging-and-signal-integrity/