[libre-riscv-dev] [Bug 117] RISCV FCLASS instruction needed

bugzilla-daemon at libre-riscv.org bugzilla-daemon at libre-riscv.org
Sun Jul 28 13:30:51 BST 2019


--- Comment #12 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #11)

> > does test_fpclass_pipe.py look reasonable, i.e. conformant to the FCLASS
> > spec?
> yes, though the conditions are quite redundant as written and it needs to be
> updated to match FPFormat.

i originally wrote FPClassMod in that format, from the spec, in an "i can
read and understand this according to the RISCV spec" way.

i then looked at the rocket-chip chisel3 code, and went "i get that this
is optimal as far as hardware is concerned, so let's go with it... however
i simply can't understand easily why it works".

i'm more concerned, in the *testing*, about readability and understandability
(not about "performance" of the testing code).

however in the hardware, gate count is clearly more important, even at the
expense of readability.

You are receiving this mail because:
You are on the CC list for the bug.

More information about the libre-riscv-dev mailing list