|
|
#1 | |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
After installing 260.19.06 I ran qvdpautest and got the error:
The following error occured: Can't create vdp device : unsupported GPU? With versions before 260 I have no problem. vdpauinfo gives: Error creating VDPAU device: 23 |
|
|
|
|
|
|
#2 | |
|
Moderator
Join Date: Aug 2005
Posts: 1,327
|
JoJo19,
Please can you run the following command in a terminal, and then run your VDPAU application in the same terminal. This will print some extra debugging information that should help diagnose the problem. Also, can you please say which driver version you had installed before that worked. Thanks. export VDPAU_NVIDIA_TRACE=1 export VDPAU_NVIDIA_DEBUG=3 |
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
Working driver versions were 195.xx and 256.xx.
For everyday use I take only 195.30, because of the known problems of later versions with xine. I did the exports before starting the first bug report, but not in the same terminal. So now I attach bug reports for again qvdpautest and for xine with the exports in the same terminal. qvdpautest: VDPAU nvidia: Version: NVIDIA VDPAU Driver Shared Library 260.19.06 Mon Sep 13 07:05:35 PDT 2010 VDPAU nvidia: Error detected 0 462 VDPAU nvidia: Backtrace: --: /usr/lib/libvdpau_nvidia.so [0xb5f64000] DSO load base 00: /usr/lib/libvdpau_nvidia.so [0xb5f7445c] 01: /usr/lib/libvdpau_nvidia.so [0xb5f7730c] vdp_imp_device_create_x11 vdpauinfo: VDPAU nvidia: Version: NVIDIA VDPAU Driver Shared Library 260.19.06 Mon Sep 13 07:05:35 PDT 2010 VDPAU nvidia: Error detected 0 462 VDPAU nvidia: Backtrace: --: /usr/lib/libvdpau_nvidia.so [0xb72c1000] DSO load base 00: /usr/lib/libvdpau_nvidia.so [0xb72d145c] 01: /usr/lib/libvdpau_nvidia.so [0xb72d430c] vdp_imp_device_create_x11 Error creating VDPAU device: 23 display: :0.0 screen: 0 xine: VDPAU nvidia: Version: NVIDIA VDPAU Driver Shared Library 260.19.06 Mon Sep 13 07:05:35 PDT 2010 VDPAU nvidia: Error detected 0 462 VDPAU nvidia: Backtrace: --: /usr/lib/libvdpau_nvidia.so [0xb57ef000] DSO load base 00: /usr/lib/libvdpau_nvidia.so [0xb57ff45c] 01: /usr/lib/libvdpau_nvidia.so [0xb580230c] vdp_imp_device_create_x11 vo_vdpau: Can't create vdp device : unsupported GPU? main: video driver <vdpau> failed |
|
|
|
|
|
#4 | |
|
Moderator
Join Date: Aug 2005
Posts: 1,327
|
JoJo19,
Are you sure that this failure is 100% correlated with the new driver? The failure in question is a call to malloc() - the system's memory allocation function. In 260.*, this is probably the very first allocation that VDPAU ever makes, whereas in earlier drivers, this allocation was made later during initialization. This implies that this particular allocation is more likely to work successfully in 260.* than earlier drivers, not less. This indicates that your system is running low on memory. Last edited by Stephen Warren; 10-11-10 at 10:39 AM. Reason: Slight rewording, typo. |
|
|
|
|
|
|
#5 |
|
Moderator
Join Date: Aug 2005
Posts: 1,327
|
JoJo19,
I think we've found the source of the problem. The size of that particular allocation significantly increased between 256.* and 260.* as part of support for Xinerama. Given the amount of physical RAM in your machine, it's unlikely the new size can be allocated on your machine. We'll look into ways of making this more dynamic to reduce memory usage. |
|
|
|
|
|
#6 | |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
Does this depend only on physical RAM (I have 512 MB), or would increasing the swap partition help? And if yes, would it make sense or would it be too slow?
|
|
|
|
|
|
|
#7 |
|
Moderator
Join Date: Aug 2005
Posts: 1,327
|
Adding a couple GB of swap might help. The access to the allocated RAM is fairly sparse, so you might get acceptable results.
|
|
|
|
|
|
#8 |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
Could you please tell how much swap is necessary as a minimum?
It is not easy for me to make some free space on that hard drive for that, so I would like to know. And thank you for all the information. And btw Xinerama is disabled in my xorg.conf, is this not honored? Last edited by JoJo19; 10-11-10 at 04:03 PM. Reason: adding question |
|
|
|
|
|
#9 |
|
Moderator
Join Date: Aug 2005
Posts: 1,327
|
We've identified and fixed this bug. The fix will be included in a future driver release.
You'll probably need 1-2G of free swap (free, not total size) with this bug. However, it's probable that performance will significantly suffer due to this bug, so it'd be simplest to use an earlier driver, and wait for the bug fix to be released, rather than risking re-partitioning your disk. |
|
|
|
|
|
#10 |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
I see. So IŽll wait for the next release.
Thanks a lot. |
|
|
|
|
|
#11 | |
|
Linux kernel hacker
Join Date: Feb 2009
Location: Ottawa, Canada
Posts: 173
|
Quote:
Eg. Create/use a 2GB swapfile as follows: hdparm --fallocate 2000000 /swapfile ## (kilo-bytes) mkswap /swapfile swapon /swapfile |
|
|
|
|
|
|
#12 |
|
Registered User
Join Date: Oct 2010
Posts: 16
|
Yes, with a 1 GB swapfile it works (as far as I can tell after few tests only).
|
|
|
|
![]() |
| Thread Tools | |
|
|