4 Nov
2005
4 Nov
'05
9:19 a.m.
Allison wrote:
There is also the matter of fault tolerence. Parity
doen't provide that
unless the OS can map out the offending bank and the offending bank
isn't in a critical location. ECC can keep you going if you have a
stuck bit but it's up to the user/maintence to fix asap.
But does any software even consider being able to be rolled back from a
check point nowdays?
With a well designed virtual memory system ( no gui stuff I guess ) a
process that takes several
days? or more to process needs to interuptable and saveable.