Note that the number is 6% of machines, so your machine is very likely to be fine. And all of Gary's explanations are true, too.
-
-
-
I used to work at a co with 1k physical machines and we saw lots of data corruption, some segfaults, consistent with the numbers above.
- 2 more replies
New conversation -
-
-
I think that's a side-effect of working on CPUs: simulator bugs are potentially disastrous, so you track/classify all failures
-
And since you spend all day staring at bits anyway, looking at bits in memory is a pretty natural thing to do.
End of conversation
New conversation -
-
-
BTW the background here is that 95% of mem in maybe 80% of machines is checking a software model vs. hardware "RTL" (code).
-
Data corruption is likely to show up as some kind of test failure because it's unlikely that a bit flips to match the other model.
- 1 more reply
New conversation -
-
-
When we looked at segfaults, they would often obv. be from corruption (e.g., load addr into reg, deference, faulting address has 1bit flip).
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
BTW, your desktop and laptop most likely don't have ECC. How lucky are you feeling today?
-
@danluu eight bits or just the seven? Do you feel lucky, punk?
End of conversation
New conversation -
-
-
@danluu Very much looking forward to the ARM/POWER unsuitability rant... -
@andywarfield Trying to decide if I should wait for rtrment before I pen that,but if in Seattle let me buy you a beer and talk your ear off. - 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.