New port: pvrxxx for Hauppauge PVR150/500
Ian Smith
smithi at nimnet.asn.au
Tue Oct 17 04:58:41 PDT 2006
On Mon, 16 Oct 2006, Rick C. Petty wrote:
> On Mon, Oct 16, 2006 at 05:52:21PM +1000, Ian Smith wrote:
> >
> > Even bit-banging, I expect that it should be able to service interrupts
> > between bytes - though not between bits, obviously - at least while
>
> Why obviously? There's a clock line, so I can't see why pausing even
> between the bits would hurt anything... so long as your transfer rate is
> below the 400 kbit/s throughput. I've done a lot of Atmel AVR i2c myself,
> even with direct bit-banging it seems to work just fine. Of course, if you
> could offload the bus timing you would see a huge performance gain, but
> that's not possible AFAIK in this case.
Looking at the I2C timing diagrams, I suspect you may be right, though
you'd likely only want to stretch timing with SCL high (ie not mid-bit).
Also, while the master controls SCL, the slave may or may not delay the
clock on acknowledge; I think dealing with it bytewise would be simpler.
Anyway, just in addressing this nasty 20-plus second delay reported on
loading firmware, enabling interrupts between bytes sounds like much
easier work and would reduce latency to one nine-bit byte-time, or
~22.5uS @400kHz, which surely beats 20 seconds of unresponsiveness :)
[..]
> > does appear that iicbb is pretty long in the tooth ('98?)
>
> The iicbb could use some cleaning up. I could take a look at it if no one
> else has the time/ability.
It sure won't be me, on either count .. back to AVR bit-twiddling ..
Cheers, Ian
More information about the freebsd-multimedia
mailing list