Yes.
Thank you very much. I was able to run with nv-codec 8.2
$ git checkout sdk/8.2
and recompile ffmpeg.
Regards,
Mahmood
On Fri, Dec 20, 2019 at 10:43 PM Dennis Mungai wrote:
> On Fri, 20 Dec 2019, 21:57 Mahmood Naderan, wrote:
>
> > >recommended R440 (with Turing).
>recommended R440 (with Turing).
Dennis,
I am using TitanV which is Volta.
Regards,
Mahmood
___
ffmpeg-user mailing list
ffmpeg-user@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-user
To unsubscribe, visit link above, or email
ffmpeg-user-requ.
I don't have root access. What should I do to use older ffmpeg and existing
nv-codec?
Regards,
Mahmood
On Fri, Dec 20, 2019 at 9:51 PM Dennis Mungai wrote:
> On Fri, 20 Dec 2019, 20:58 Mahmood Naderan, wrote:
>
> > Hi
> > I use this command which fails
Hi
I use this command which fails with this error message
$ ./ffmpeg -init_hw_device cuda=0 -filter_hw_device 0 -i vid1080.mp4 -vf
format=nv12,hwupload,scale_npp=1280:720 -c:v h264_nvenc 720p2.mp4
ffmpeg version N-96097-g99f505d2df Copyright (c) 2000-2019 the FFmpeg
developers
built with gcc 7
OK.
--extra-cflags=-I,-I
is not correct. The correct syntax is
--extra-cflags=-I --extra-cflags= -I
Regards,
Mahmood
On Fri, Dec 20, 2019 at 6:40 PM Mahmood Naderan
wrote:
> >You're going to have to install the full CUDA SDK.
> >The headers I see missing in your config are
>You're going to have to install the full CUDA SDK.
>The headers I see missing in your config are indeed present in the CUDA
SDK.
I actually have that in my home. I mean cuda toolkit and samples.
I can see npp.h and its libraries there.
$ cd ~/cuda-10.1.168/
$ ls include/npp*
include/nppcore.h
I have attached config.log.zip.
At the end, it is looking for npp.h and some other things. I can not find
in which package npp.h exists. All I see are library files.
gcc -D_ISOC99_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE
-D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600 -DPIC
-I/home/mahmood/c
>ERROR: libnpp not found
I even extracted
https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-npp-10-1_10.1.168-1_amd64.deb
and copied all libnpp files to my own cuda directory.
$ ls
libnppc.so.10 libnppicom.so.10libnppig.so.10
libnppisu.so.10
libnppc.
> Remove yasm/include from extra-cflags, remove yasm/lib from extra-ldflags
>and use --x86asmexe=/home/mahmood/yasm-1.3.0/bin/yasm
Thank you. I got it. also fixed nv-coded-headers. However, libnpp is not
found although I have manually downloaded the deb package and extracted it
to get libnpps.so.
Hi,
On a system which I can not install package, I have built yasm from source
as below:
mahmood@fury0:/home/mahmood/yasm-1.3.0$ ls bin/
vsyasm yasm ytasm
mahmood@fury0:/home/mahmood/yasm-1.3.0$ ls lib
libyasm.a
mahmood@fury0:/home/mahmood/yasm-1.3.0$ ls include/
libyasm libyasm.h libyasm-stdi
Hi,
If you look at the apt output, there is no dev package.
Do you know exactly what is that library?
The command I used is mentioned at https://developer.nvidia.com/ffmpeg
Regards,
Mahmood
On Tue, Aug 6, 2019 at 9:19 PM Reindl Harald wrote:
>
>
> you need the dev packages to compile and li
Hi
When I configure ffmpeg with the following command
$ ./configure --enable-cuda --enable-cuvid --enable-nvenc --enable-nonfree
--enable-libnpp
--extra-cflags=-I/usr/local/cuda/include,../nv-codec-headers/include
--extra-ldflags=-L/usr/local/cuda/lib64
I get this error
ERROR: libnpp not found
>Correction:
>
>./ffmpeg -init_hw_device cuda=0 -filter_hw_device 0 -i ../4k_normal.mp4 -vf
>format=nv12,hwupload,scale_npp=1280:720 -c:v h264_nvenc 720p1.mp4
OK that is good
$ time ./ffmpeg -init_hw_device cuda=0 -filter_hw_device 0 -i
../4k_normal.mp4 -vf format=nv12,hwupload,scale_npp=1280
>Using your sample above:
>
>./ffmpeg -init_hw_device cuda=0 -i ../4k_normal.mp4 -vf
>format=nv12,hwupload,scale_npp=1280:720 -c:v h264_nvenc 720p1.mp4
>
>Try that and report back.
>
>
It fails
$ time ./ffmpeg -init_hw_device cuda=0 -i ../4k_normal.mp4 -vf
format=nv12,hwupload,scale_npp=1280:720
>Note that hardware accelerated
>decode is targeted for real time PLAYBACK without using any CPU resources,
>and does not necessarily imply to be faster than a software based decode.
If I got the point, you are saying that GPU version is somehow used to free
CPU cores, so that CPU cores are ready
>Then don't use the cuda filter.
So, my time measurements show that
CPU: ./ffmpeg -i ../4k_normal.mp4 -vf scale=1280:720 720p1.mp4
real0m23.748s
GPU: ./ffmpeg -hwaccel cuvid -c:v h264_cuvid -i ../4k_normal.mp4 -vf
scale_npp=1280:720 -c:v h264_nvenc 720p1.mp4
real0m20.889s
Do you thi
>I believe if you want to use the cuda scaling filter you should also
>use cuda decoding.
The original cuda compatible run is
./ffmpeg -hwaccel cuvid -c:v h264_cuvid -i ../4k_normal.mp4 -vf
scale_npp=1280:720 -c:v h264_nvenc 720p1.mp4
I want to do that with multicore CPU. That command basically
Hi,
I want to run a multicore CPU job with ffmpeg and I think the command
should be
./ffmpeg -i ../4k_normal.mp4 -vf scale_npp=1280:720 720p1.mp4
but it fails
./ffmpeg -i ../4k_normal.mp4 -vf scale_npp=1280:720 720p1.mp4
ffmpeg version N-93005-gd92f06eb66 Copyright (c) 2000-2019 the FFmpeg
dev
>Your question is very difficult to understand:
>Are you surprised that the values for the utilization are different?
>Or do you find them too high? Or too low?
Different and low.
The example is given in the Nvidia website.
Regards,
Mahmood
___
ffmpeg-u
Hi,
For this command
an input with this specification
./ffmpeg -vsync 0 -c:v h264_cuvid -i in.mp4 -f rawvideo out.mp4
and two input files
Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p,
1920x1080 [SAR 1:1 DAR 16:9], 8239 kb/s, 29.97 fps, 29.97 tbr, 30k tbn,
59.94 tbc (default)
xing overhead: 0.307457%
[aac @ 0x55e602553980] Qavg: 135.844
That utilizes GPU for about 20%. Is that all? Or there are some options for
better using GPU?
[1] https://developer.nvidia.com/ffmpeg
Regards,
Mahmood
On Thu, Jan 24, 2019 at 10:25 AM Mahmood Naderan
wrote:
> >Yo
>You're only using the GPU for decoding. (The "dec" in "nvdec" hints at
>this.) The GPU probably idling, while the CPU encoding is the bottleneck.
Even for encoding, I see 10% GPU utilization
./ffmpeg -i ../4k_normal.mp4 -c:v h264_nvenc 4k.mp4
Regards,
Mahmood
__
Hi,
I have noted that the following command
./ffmpeg -hwaccel nvdec -i ../4k_normal.mp4 -preset medium -movflags
+faststart -c:a aac ../4k_normal_conv.mp4
utilizes the GPU for about 20% according to nvidia-smi. That is pretty low
while 500MB out of 4GB of device memory is occupied.
Is that norma
obal headers:0kB
muxing overhead: 0.169521%
[aac @ 0x561bd68d7d40] Qavg: 65536.000
Regards,
Mahmood
On Wed, Jan 23, 2019 at 10:13 PM Mahmood Naderan
wrote:
> So, I think the command should be similar to this
>
> ./ffmpeg -i ../fast_video.mp4 h264_nvenc -preset medium -movflags
&
So, I think the command should be similar to this
./ffmpeg -i ../fast_video.mp4 h264_nvenc -preset medium -movflags
+faststart -c:a aac ../fast_video_conv.mp4
but that fails
[NULL @ 0x558bd7fcadc0] Unable to find a suitable output format for
'h264_nvenc'
h264_nvenc: Invalid argument
Regards,
M
0.169521%
[aac @ 0x55c859e9b880] Qavg: 65536.000
Regards,
Mahmood
On Wed, Jan 23, 2019 at 9:37 PM Carl Zwanzig wrote:
> On 1/23/2019 9:48 AM, Mahmood Naderan wrote:
> > Any idea?
>
> Post the full output of the command you're trying? (it
Carl,
Please see the make output at https://pastebin.com/BzpEHvjV
Regards,
Mahmood
On Wed, Jan 23, 2019 at 9:44 PM Mahmood Naderan
wrote:
>
> Thanks for the quick reply. The config log can be seen at
> https://pastebin.com/WeYbXiWW
> Also, the full output of the command I
Hi,
I have built FFMPEG from source on an Ubuntu 18.04 with the following
commands:
git clone https://git.videolan.org/git/ffmpeg/nv-codec-headers.git
cd nv-codec-headers/
make
sudo make install
cd ../
./configure --enable-cuda --enable-cuvid --enable-nvenc --enable-nonfree
--enable-libnpp --extra
28 matches
Mail list logo