[mythtv-users] Mystique SaTiX-S2 Dual PCIe (stv0900 + stv6110)

Devin Heitmueller dheitmueller at kernellabs.com
Mon Apr 12 15:13:44 UTC 2010


On Mon, Apr 12, 2010 at 11:03 AM, Andre <mythtv-list at dinkum.org.uk> wrote:
>
> On 12 Apr 2010, at 15:39, Devin Heitmueller wrote:
>>
>> Be forewarned:  there are significant problems with this card when
>> doing capture on both inputs simultaneously.  Be prepared for timeouts
>> when starting streams and sometimes waiting tens of seconds for
>> streaming to start (it's an issue with the ngene bridge chipset).
>>
>> You can read more about it on the linux-media mailing list.
>
> I've been following linux-media (since you directed me to it) and I've been expecting problems but they haven't materialised (for me), I do start recordings several minutes early and can't remember the last time I watched live TV so maybe I've just not noticed. What I have noticed that in the 10+ days I've had the card I have a lot fewer glitchy recordings (none from the new card, except when I've purposely pushed it) and the odd completely failed recording has always been with the HVR-4000, I'd suspect a faulty card (which would be the second) but my HD-S2 behaves exactly the same.
>
> I get the impression that a lot of the linux drivers are flaky in differing ways, then again looking at the Windows forums for various cards they don't seem to be getting along especially any better! I'm happy to work around the "issues" in Linux, wouldn't have a hope of doing half of what I now take for granted on any other operating system :-))
>
> Thanks for the warning, I see from your posts on linux-media and your blog that your opinion is to be valued.

Hi Andre,

YMMV, of course.  I've spent about three weeks working on the ngene
driver and have about 90 patches pending for it.  Given the crap I've
seen so far in the current driver, I wouldn't recommend the card for
production use, although I am hopeful that we can get it into a
reasonable state.

I agree with your basic statement though that the Linux drivers do
vary quite a bit in quality.  Getting the card basically working is
only about 30% of the time required.  The other 70% is finding all the
weird edge cases and intermittent problems, and many developers simply
don't have the time to invest to deal with those cases.

Spending two weeks tracking down a one line register change that
addresses a 3% tuning failure rate just isn't worth their time.

Cheers,

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com


More information about the mythtv-users mailing list