[Libre-soc-dev] cray-style vector of 40 years setting VL=0 at runtime

[Luke Kenneth Casson Leighton]

jacob can you please read in full and absorb the commit comment here?
https://git.libre-soc.org/?p=openpower-isa.git;a=commitdiff;h=b7e64d86c80eaec5ed30a54b1cb354af4183a6a2

complaints about how 40-year Cray-style Vectors work is very
inappropriate to put into the source code that we are using as a
show-case for the capabilities of Simple-V.  it could be taken as
a sign that we have absolutely no idea what we are doing, are
incompetent, and that the entire exercise of the past four years
has been on a false premise and that the OPF ISA WG should
take the fact that SV is so deeply "flawed" - because according
to those source code comments "VL=0 has not been thought
through properly" - that it should in no way be accepted into
Power ISA.

the planned SVP64Single is the correct way to solve the problem,
reducing code size in the process (no need for a setvl), but it
was completely inappropriate to say "i don't like that i have to do this,
VL=0 should be banned" because it basically is saying that you
have absolutely no idea why VL=0 was permitted in the original
Cray Vector ISA in the first place.  after four years working with SV
all this time i have to say that i am really taken aback that you don't
know why VL=0 is permitted.

therefore can you please rather urgently study the commit message
and understand this vital capability, as well as understand the
catastrophic consequences of prohibiting VL=0 at runtime.

l.