nV News Forums

 
 

nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   TV-input support (http://www.nvnews.net/vbulletin/showthread.php?t=31847)

michalw 07-12-04 10:03 AM

TV-input support
 
Hello,

Are there any plans to include TV-input support for cards with VIVO?

Thanks in advance for your comments,
Michal

Thunderbird 07-12-04 11:38 AM

Re: TV-input support
 
In general tvin is something added by the board manufacturers themself (asus, msi ..). It is not really an official nvidia feature. To get tvin support there's a project called rivatv at rivatv.sf.net that has working tvin (not bugfree). The only problem is that it doesn't work with the 1.0-6106 drivers. The new drivers don't allow external drivers to access the nvidia card.

michalw 07-13-04 02:37 AM

Re: TV-input support
 
I am aware of the rivatv, but as you stated it does not work with the new drivers, so it can be treated only as a temporary solution.

From the user perspective it is the same hardware, as it is for the drivers, which don't allow external control of the VIVO chip. In my opinion TV-input support should be part of the NVidia driver.

Anyway, I thought the company that claims to be the "leader in Linux driver quality, stability, performance and support", has enough resources to write the driver for the add-on chip found in so many boards.

Thunderbird 07-13-04 03:50 AM

Re: TV-input support
 
Tvin support is not that simple. In short there's a special bus on the chip called the "i2c" bus to which you can connect extra chips like hardware monitoring ones but also tvin ones. First a driver is needed to talk to this external chip. Once you have that you still need a driver to do the tvin work. The problem with that is that there are lots of different chips.

hotbelgo 02-17-05 03:22 AM

Re: TV-input support
 
Quote:

Originally Posted by Thunderbird
Tvin support is not that simple. In short there's a special bus on the chip called the "i2c" bus to which you can connect extra chips like hardware monitoring ones but also tvin ones. First a driver is needed to talk to this external chip. Once you have that you still need a driver to do the tvin work. The problem with that is that there are lots of different chips.

In practice there are only a limited number of the external chips (all from philips) as the RivaTV project demonstrates, and it is only one of them that still does not work (the saa7174hl). (The tuner chips are all controlled by the bttv drivers.)

So, as far as I understand it, video-in is a part of the nvidia chip and it is surely something that nVidia has to take responsibility for and to provide an driver to control video-in that is compatible with the latest nvidia drivers and interoperable with i2c.

In an ideal world, nvidia would also recognise that the "personal cinema" cards in question are sold with the nVidia "NVTV" branding very much in evidence and the philips chips are not mentioned at all. I bought one because I perceived nVidia as being serious about Linux and did not realise at the time that the support was very circumscribed. It may not be for nVidia to get the philips chips working, but you clearly have more leverage than the average consumer to persuade them to ship a Linux driver along with the all the Windows ones (and SDK) that they already have.

Simon

jlo 03-19-05 02:23 PM

Re: TV-input support
 
Quote:

Originally Posted by Thunderbird
To get tvin support there's a project called rivatv at rivatv.sf.net that has working tvin (not bugfree). The only problem is that it doesn't work with the 1.0-6106 drivers. The new drivers don't allow external drivers to access the nvidia card.

So is anything being done to this problem that new nVidia drivers do not allow external drivers to access the nvidia card? rivatv is currently not working and TVin therefore neither because something was changed in nVidia linux driver. rivatv folks say, they cannot fix it if nvidia won't help them.

Anyone know, has ATI tackled this problem better? I am going to set up a new system soon, and I would like to know a working 3dGL+VideoIn hardware+drivers for Linux.

jlo 03-20-05 04:00 AM

Re: Video-input support
 
Quote:

Originally Posted by Thunderbird
Re: FX5700 Personal Cinema on Linux?
Please don't try to get multiple old topics regarding tvin high on the list again...

OK, sorry. I just first tried to search explanation with google, then in rivatv mailing list and then here using nvnews.net Search-tool. I couldn't find any explanation why in older nvidia drivers rivatv support was possible, but now with the latest ones the support has been dropped.

The correct title to the thread would be "Video-input support", because the most nvidia based VIVO-cards don't actually have TV-tuner, just Siemens SAAXXXX Video-In chips. For example I have Leadtek FX 5600 with SAA7114H.

The fact that nvidia driver don't allow other drivers access the card hardware at the same time is an after-effect, not the reason. What is the reason this change was made? Could there be an option in xorg.conf file for nvidia driver to enable access to rivatv driver? Say if the reason is the method is somewhat buggy, then the default behaviour would be to prevent the access, but if the user knows what (s)he is doing, one could enable rivatv working beside nvidia driver?

Is there an explanation (a link) somewhere what is the basic reason for this substandard working of nvidia boards? As it is said, the issue is getting old already, but still there is weekly/monthy questions why VIVO doesn't work, so clearly the explanation is missing. Usually Linux users are not satisfied with an explanation like "it is broken".

Thunderbird 03-20-05 05:46 AM

Re: TV-input support
 
Rivatv needs access to the i2c bus of the nvidia gpu and it needs to modify lots of card registers. These two drivers really can't share those things as it is very difficult to do (both would have to track changes the other made) and it further isn't great for stability either. The only nice way would be for nvidia to create a driver themself that works together with their own driver.

jlo 03-20-05 12:51 PM

Re: Video-input support
 
Quote:

Originally Posted by Thunderbird
Rivatv needs access to the i2c bus of the nvidia gpu and it needs to modify lots of card registers. These two drivers really can't share those things as it is very difficult to do (both would have to track changes the other made) and it further isn't great for stability either. The only nice way would be for nvidia to create a driver themself that works together with their own driver.

Well they did share it couple of versions back in nvidia driver family.
I don't quite see it realistic, nVidia would start to support all current and future Siemens SAAXXXX (and other) video-in chips, which are included in nvidia chip based graphics cards. But callback hooks nvidia could do very quickly, which would allow rivatv to use GPU and card registers through nvidia driver.

This really is not rocket science which would have to take months and years to tackle. nVidia could be kind and focus to this problem as soon as possible and provide a socket to rivatv driver to attach through it in a controlled way. It would take a max one week, if the problem would be taken seriously.The much better method technically of course would be to open source the driver, but that is not the only option to make it work.

This situation seems to lead nowhere. nVidia claims it is not in their responsibility to support Siemens SAAXXXX Video-in chips, but at the same time their driver blocks any other driver to get through to SAAXXXX chips and support them. As a gatekeeper, nVidia is in charge and should frankly and sturdy carry its responsibility and not let the users down making excuses.

Thunderbird 03-20-05 04:47 PM

Re: TV-input support
 
It is not just the communication link between the Philips SAAXXXX chip that rivatv needs access too. If it was just this it wouldn't be a very big issue. Also access to lots of display registers is needed to dump video images (that stuff isn't done using i2c as i2c is very slow and mainly for controlling chips and reading information). All these registers need to be accessed using dma to make stuff fast and sharing this all is the problematic part. In the old driver rivatv indeed work a bit but had lots of issues for instance dma didn't work and other problems. (because it is about impossible to implement)

hotbelgo 03-21-05 07:03 AM

Re: TV-input support
 
Quote:

Originally Posted by Thunderbird
In general tvin is something added by the board manufacturers themself (asus, msi ..). It is not really an official nvidia feature. To get tvin support there's a project called rivatv at rivatv.sf.net that has working tvin (not bugfree). The only problem is that it doesn't work with the 1.0-6106 drivers. The new drivers don't allow external drivers to access the nvidia card.

The RivaTV website has been updated explicitly to note that it does not support the newer NVTV personal cinema cards.

It would be a big pity if people misunderstood nVidia's Linux "support" and accidentally increased nVidia sales before learning that the support does not strech very far and, in particular, does not include recent TV cards. That much I know from personal experience! :(

jlo 03-21-05 08:04 AM

Re: Video-input support
 
Quote:

Originally Posted by Thunderbird
these registers need to be accessed using dma to make stuff fast and sharing this all is the problematic part. In the old driver rivatv indeed work a bit but had lots of issues for instance dma didn't work and other problems. (because it is about impossible to implement)

I haven't ever used the video card in MS XP or on other MS Windows, but was it "about impossible" there too? Or does it (3D + Video-IN) work there? Or is this clearly a design flaw, the video-in chips can never work in the same PCI/AGP-card as nvidia GPU chips? If it is a design flaw, I believe many have right to return their card and get a refund.

I know there is so called "impossible" tasks in computing science. They usually are those NP-problems or due to some missing hardware or design flaw. I am sure this is not a NP-problem, so the two other choices are in question.


All times are GMT -5. The time now is 01:47 AM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.