Difference between revisions of "Talk:Problem with DVI throughput"
(→Corruption and Flickering on T43) |
|||
Line 53: | Line 53: | ||
[[User:Wyrfel|Wyrfel]] 01:43, 27 Oct 2005 (CEST) | [[User:Wyrfel|Wyrfel]] 01:43, 27 Oct 2005 (CEST) | ||
+ | |||
+ | I haven't, as far as I know, tried the newest IBM drivers for the Radeon X300 - dated 2005-02-11 on Lenovo's website. I just did an install of ATI Catalyst 5.11 with the included control center. This installation actually requires a hack (http://www.driverheaven.net/patje/) for mobile devices. But in any case, the NEC LCD200VX still requires "Reduce DVI Frequency" to be unchecked in the ATI Control Center to avoid artifacts and instability. I would guess that the IBM drives also have this option enabled by default, so probably they would not work on this particular display without changing the settings, but I don't know for sure. I did take a look at the monitor's OSD info. In both cases, with and without "reduce DVI frequency" checked the OSD reports running at 1600x1200 at 75Hz horiz, 60Hz vertical. I'd really like to know what this setting does, then. | ||
+ | |||
+ | --[[User:Aeyea|Aeyea]] 21:07, 15 Nov 2005 (CET) | ||
+ | |||
---- | ---- | ||
Revision as of 21:07, 15 November 2005
Here's some information I found:
Most DVI devices today use single-link DVI which supports a maximum bandwidth of 165 MHz [1]. I suppose that ThinkPads provide a single-link DVI signal.
To calculate the bandwidth of your signal, in theory you just need to multiply the resolution with the vertical frequency. For example: 1600x1200 @ 85 Hz => 1600*1200*85 Hz = 163 MHz. This looks good, but it is outside the specs - because you need to take into account some extra time. CRTs need a so-called blanking time between the data for two display lines. This time is needed for something similar to a carriage return on a typewriter: The electron beam needs to be returned to the start of the next line. Additionally, some extra time is needed to transmit information about the border area around the real picture. About 25% of the bandwidth is used for these additional data. [2]
This means, that there is no specific ThinkPad problem with the throughput - you are just trying to transfer video data at a rate that is outside the specs of DVI. Maybe the DVI transmitter inside the ThinkPad even works at this data rate. But the DVI receiver inside the monitor might be "overclocked".
TFTs don't really need extra time. So there is a chance of writing a display driver with a reduced blanking time. This might explain, why some people were successful with other drivers.
The effective maximum resolution for single link DVI should be around 1600x1200 @ 65 Hz, or @ 60 Hz to be on the sure side. Has anyone solved his problems by lowering the vertical frequency?
--137.226.40.2 14:23, 1 Mar 2005 (CET)
Contents
No problem on T41p
I have a T41p with 1400x1040 LCD and a Samsung SyncMaster 213T connected through the Port Replicator II with DVI
After some windows registry hacks I can run the Samsung in 1600x1200 in single, clone display and extend display mode.
Linux requires no such hacks and will happily run at 1600x1200
The picture through DVI is perfect with either desktop applications or running DVD video.
This is good news. Do you run Linux in DualHead, MergedFB or CloneMode? In Windows you have a clear video overlay all the time (like after every boot)? Which drivers are you using in Linux and Windows, please list the with version numbers.
217.230.177.126 01:31, 10 Mar 2005 (CET)
Question about DVI/LCD switching on T4x / Radeon M9 / Linux
Has anyone been able to successfully switch back and forth between the LCD output and the Port Replicator II DVI output on Linux? How do you do it? My problem is described below:
[[3]]
No problems here
I installed the new catalyst drivers 5.6 using Patje's Mobility Moddingand it works for all resolutions. My laptop runs at 1400x1060 and my dvi screen at 1920x1200. No problems that I can tell. The omega drivers (based on older catalyst drivers) blue screened my windows xp on boot.
Thanks for the info, it is a while since i tested with Catalyst and back then it didn't allow 1400x1050.
Corruption and Flickering on T43
No problems changing to 1600x1200 @60Hz on DVI passthrough using stock ATI drivers (6.14.10) and IBM presentation director on T43 and port replicator II. However the image is not stable - green dot artifacts flicker in approximately horizontal lines, while the screen blanks in and out. Completely stable at 1280x1024. Changing to the Omega drivers did not help at all. Also replacing ATI with Omega drivers somehow removed the presentation director key binding Fn-F7 and I had to reinstall it to get it back. Monitor is NEC AccuSync 200VX. Tried what appeared to be both single and dual-link DVI cables (not sure about monitor or port replicator support for dual-link).
I just tried the same setup while replacing the NEC monitor with a Planar PL2010M. The Planar LCD gives a completely stable image at 1600x1200.
UPDATE: I was able to get a stable 1600x1200 image on the NEC panel by going into the ATI Catalyst Control Center (downloadable as standalone program from ATI webpage) and unchecking "Reduce DVI frequency on high-resolution displays" which was on by default. The "Alternate DVI operational mode" option seems to have no effect either way. I'm guessing "Reduce DVI frequency" changes to 50 Hz or something, and "Aleternate DVI operational mode" is the reduced blanking time, but who knows with ATI's ambiguous descriptions. In any case it's interesting that the fixes mentioned on this page actually cause trouble on the NEC display, while taking away the "fixes" makes the problem go away.
It's interesting, indeed, since it's quite illogical. Did you try the newest official IBM drivers? They worked out of the box for me and at least one other person. On a T41p with a NEC 2180UX that is. I think the "Reduce DVI frequency" changes to 60 Hz, which is supposed to be ideal for DVI TFTs, but that may in fact vary.
Wyrfel 01:43, 27 Oct 2005 (CEST)
I haven't, as far as I know, tried the newest IBM drivers for the Radeon X300 - dated 2005-02-11 on Lenovo's website. I just did an install of ATI Catalyst 5.11 with the included control center. This installation actually requires a hack (http://www.driverheaven.net/patje/) for mobile devices. But in any case, the NEC LCD200VX still requires "Reduce DVI Frequency" to be unchecked in the ATI Control Center to avoid artifacts and instability. I would guess that the IBM drives also have this option enabled by default, so probably they would not work on this particular display without changing the settings, but I don't know for sure. I did take a look at the monitor's OSD info. In both cases, with and without "reduce DVI frequency" checked the OSD reports running at 1600x1200 at 75Hz horiz, 60Hz vertical. I'd really like to know what this setting does, then.
--Aeyea 21:07, 15 Nov 2005 (CET)
Single-link DVI
No thinkpads are currently capable of supporting dual-link DVI natively.
Single link is limited to 165MHz, while dual link has double that.
1600x1200x60=115200000Hz or 115.2MHz (assuming no blanking space), so in theory that should not be a problem, but in practice there might be additional limitations, such as the quality of the video controller and the passthrough to the dock or port replicator.
The IBM video drivers have always limited DVI output to 1280x1024, probably due to these issues. Some people have had success hacking the drivers (or using alternative drivers), and going up to 1920x1200, but other people get corruption already at 1600x1200.
1920x1200x60=138240000Hz or 138.24MHz
X40?
The article says the X40 is affected by this issue---how is that? Are there any X40 docs or adapters that give the X40 DVI capability at all?
Right, removed them from the listing. Wyrfel 22:08, 25 Sep 2005 (CEST)
List of affected models is totally incorrect
Just because a ThinkPad is compatible with a Dock or Port Replicator that has a DVI port, does not mean you can actually use it.
For example, NONE of the T2x or X series ThinkPads can do DVI, and there are probably others.
--Tonko 02:04, 26 Sep 2005 (CEST)