On Thursday 29 Jul 2004 4:29 pm, Johannes Stezenbach wrote:
Tim wrote:
It appears that boards with the Intel SATA controller built in to the
south bridge don't have any problems.
With the NF7-S and VP DVB-t, some fiddling with PCI latency settings
improves matters a little, but it is still unwatchable when
recording. I
think the VisionPlus is a lost cause.
I'm still trying to find a DVB-t card that is likely to work with the
NF7-S V2.0 and SATA. Any more comments? I'm thinking of shelling out
for
a DVICO card in the hope that it will be more reliable.
There's the possibility that the power supply isn't
stable enough. DVB frontends are quite sensitive in
that area.
Another possibility which was discussed recently on lkml is
that large IDE DMA transfers are a source of high irq latency
(i.e. the DVB card misses some irqs). But I guess the
DVB driver would spit out error messages in this case.
You could try to lower max_s from 2048 to e.g. 128 in
linux/drivers/ide/ide-disk.c:idedisk_setup() to find out.
Before making a final conclusion, why not check up the connection to
the LNB
? There can be problems on bad connectors as well -- if the shield
does not
have a good connection with the connector. Earlier somebody had a
similar
problem with the same symptom, the only difference in that case was,
when the
machine was getting heated up.
The entire cable run could be checked up for possible problems before
jumping into conclusions.
I had CRC errors when i had bad splitters. Just make sure it is the
IDE DMA
controller itself. I had problems when i tried to reduce latency to
less than
64. Eventhough Twinhan suggests 256.
One more case is there, the vision plus cards are claimed to have
problems
with some VIA chipsets and some Triton chipsets by the manufacturer
itself. I
don't know which motherboards do carry these chipsets. What info i got
was it
was due to the bad timing on the chipset.
Regards,
Manu
Johannes