Re: PCI_LATENCY_TIMER

Gerard Roudier (groudier@club-internet.fr)
Tue, 29 Sep 1998 23:36:30 +0200 (MET DST)


On Tue, 29 Sep 1998, Doug Ledford wrote:

> Gerard Roudier wrote:

> > Doug,
> >
> > Thanks for your explanation of Adaptec interpretation of PCI.
> >
> > The initial post was about the 7880U that looks like an Ultra Narrow
> > controller and, so, is only able to use 15 % of the PCI BUS bandwitch.
> > This controller is specifying a MAX_LAT of 1.92 micro-seconds.
>
> The same numbers are used on Ultra narrow or Ultra-Wide cards. At 40MB/s,
> Ultra wide uses roughly 30% of the available bandwidth. But, I had that in
> my last email, I also stated in my last email that the value of 8 in the
> MAX_LAT register was wrong on the 7880 chips :)

I had 312 messages in my mailbox this evening and each of yours must be
counted as sized as 10 times the average message size. :-)
So, I may have missed that. I remember you spoke about old boards and
I did'nt know that the 7880 was so old.

[ ... ]

> > My opinion is that any acceptable interpretation of MAX_LAT leads to
> > either this value being bogus, or that controller not being able to
> > share efficiently the PCI bus with _common_ PCI devices.
>
> The MIN_GNT and MAX_LAT values aren't about sharing. They aren't about
> fairness. They are about what it takes for that particular device to

No fairness requirement leads to PCI devices being allowed to speficy 254
as MIN_GNT and 1 as MAX_LAT that makes these PCI parameters irrelevant.

> operate at 100%. Anything else in these registers would be a lie and would

A device telling that it may achieve 100% when the PCI bus is dedicated
to itself is not a lie but _the_ truth, since any other device, even slow
will have some impact on performances.

> actually make it *harder* to get the correct performance out of the PCI
> bus. You don't want your cards being fair about PCI usage when specifying
> what they *want*, you want them to ask for what they need. Whether or not
> they get it is another matter.

You are playing with words in a insane way.

Here are my needs:

1) If I get 1000 M$ I will be 100% happy.
2) If I get 10 M$ I will be 99% happy.
3) If I get 1 M$ I will be 98% happy.

I _want_ 1000 M$, but I _need_ only 1 M$ to be quite happy. The
additionnal 999 M$ that I want are just wasting money for you with regards
to their efficiency to make me happy. So 3) is a lot more relevant for
you if you want to make me extremally happy. ;-)

In my opinion, PCI devices are expected to be as fair as I am when
specifying their needs, in order to allow the PCI BIOS to make a
reasonnably good work.

[ ... ]

Regards,
Gerard.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/