Re: [PATCH] [MIPS] [Loongson] Specific integer instructions added ( How to test âprefetchxâï)

[Richard Sandiford <rdsandiford at googlemail dot com>]

"Ruan Beihong" <ruanbeihong@gmail.com> writes:
> I'm sure that load to $0 doesn't raise any exception.

OK, thanks.  Thought I'd better make sure ;)

> As to the division insns, (d)div(u).g is faster than (d)div(u).When
> both quotient and remainder are needed, a (d)div(u).g followed by a
> (d)mod(u).g is slower. But when sequence of quotient and remainder are
> needed, the new instructions are a little faster. Considering the case
> that single div or mod takes the most share of usage of (d)div(u),
> these new instructions is useful.

Oh, I realise the new instructions are useful for the single-result case.
I just wanted to know what was the right thing when "both" results are
needed.

I think you're saying that (d)div(u) is faster when both the quotient
and remainder are needed simulatenously.  Is that right?  If both
results are needed, I think it's likely that they'll be needed at the
same time.  Division and modulus have such a long latency that we're
seldom able to fill the pipeline while waiting for the results.  I think
there are going to be few cases in which we can wait almost twice as long
for one of the results to be ready.  (Sorry for the over-generalisation.)

Which is fine if true.  The point of my question was simply:
what's the intended behaviour in the dual-result case?  And what
does the patch do?  It's useful to have tests in gcc.target/mips
to test this sort of thing.

>>Do you have a copyright assignment on file?  We'd need one before we could
> accept the patch.
>
> I don't because I don't know how to. I think the files I modified
> already have a copyright assignment, and I agree with them. If I
> really need to add a assignment, please tell me how and to where.

The problem is that you need to assign copyright for the changes you've
made yourself.  The assignment applies to authors (and the stuff they
write) rather than files.  I'll send the forms privately.

Thanks,
Richard