You might want to get your hardware checked, when random programs that should not fail start randomly acting in a weird way, it can be an indication that something is marginal (power supply, RAM, ... even the CPU possibly.) I know you doubt that it is possible, but it might be.
The egrep core happened on both machines, which are very different machines. The CS20 could have components which might be marginal, since it runs pretty hot and therefore pretty close to its shutoff temperatures. The DS25, on the other hand, has three power supplies and enough fans to keep everything very comfortably cool.
I can't imagine the same kind of problem happening on two different physical machines being hardware issues which cause the same core dumps in the same binaries. It's possible, sure, but I really can't imagine it's likely.
It is also possible there's some kind of error in the Alpha VM implementation (perhaps brought about by other changes which violate some assumption which used to be OK, but now is no longer valid.)
I wonder. I've considered trying older NetBSD versions on one of the machines, but I think it'd take a good week of compiling without a core dump to really say for sure. I might just prepare a half a dozen NFS exported versions and netboot to automate tests...
John