how to use the video memory in discrete card

According to section "Working with hardware acceleration" in sdk reference manual, we can use d3d interface to handle the video memory, and then to accelerate the speed of process. But actually in the sdk samples (sample_encode), we add parameter "-d3d" or "-d3d11" , it just use the video memory of intel HD graphics, and those memory is just the system memory shared to intel HD graphics. So, if I connect a discrete card such as Nvidia graphics, how can I used that video memory?

INDE MSDK cannot dynamically load

 The hardware:E31285V3

The syestem:win7

The INDE Version:w_inde_2015.1.021.exe

The driver :4223(20140522)

INDE MSDK  dynamically load as  follow:

hmMSDK  = LoadLibrary("libmfxhw32.dll");
 int n  = GetLastError();
 if (NULL ==  hMSDK)

The results is thar "hmMSDK = 00000;n = 126"

The mean of 126 is that can't find the specified module。


My problem:

The Inde MSDK can't  load dynamically? My methods are wrong?How to load ?


Sample_encode_x11 run twice, the same command first successful, the second error, error contents -7


 I installed the intel media sdk R6 version on ubuntu, ubuntu is 14.04.3

uname -a

Linux suirui 3.19.0-25-generic # 26 ~ 14.04.1-Ubuntu SMP Fri Jul 24 21:16:20 UTC 2015 x86_64 x86_64 x86_64 GNU / Linux

I installed documentation generic way to install and compile the sample MediaSamples_Linux_6.0.16043175.175, produced sample_encode_x11 bin file, and then I use the command ./sample_encode_x11 h264 -i test_stream_176x96.yuv -o 3.264 -w 176 -h 96 -vaapi

First run properly, resulting in a h264 file

Can't get SDK to work - Windows or Linux on Broadwell or Skylake


I'm trying to get the SDK working but on the following platforms it installs but when run the SDK samples or "ffmpeg vcodec h264_qsv" it hangs on the first frame.I've tried a variety of presets and profiles as well as async settings to see if the default params are broken, no luck.

I'm installing Intel Media SDK 2015R6.

I've tried:

CentOS 7.1
Fedora 21 (this is supported by the Intel Driver for Linux)
Windows 10


Problem when creating multiple sessions for compressing jpegs of differing sizes

I've successfully used the media SDK to produce a RTSP application that streams mjpeg, where the image size is fixed. However, I have a problem when trying to create jpegs for varying size images, which are patches extracted from the raw images and saved to disk.

The code below illustrates the problem, with the call to MFXInit returning MFX_ERR_UNSUPPORTED after a number of calls (usually 66 on my machine).

运行sample_encode_x11两次, 同样的命令第一次成功, 第二次报错, 报错内容为-7


  我安装intel media sdk R6版本在ubuntu 上, ubuntu是14.04.3

uname -a

Linux suirui 3.19.0-25-generic #26~14.04.1-Ubuntu SMP Fri Jul 24 21:16:20 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux

我安装文档的generic 的方式安装的, 然后编译sample MediaSamples_Linux_6.0.16043175.175, 产生了sample_encode_x11 bin文件, 然后我使用命令./sample_encode_x11 h264 -i test_stream_176x96.yuv -o 3.264 -w 176 -h 96  -vaapi

第一次运行正常, 产生了h264文件

CRF for Intel H.264 encoder ?

Hi, I was wondering if there is a setting in the Intel encoder that is equivalent to CRF (Constant Rate Factor) in x264. I did some research, and now I have more questions:

1. The Intel Media SDK 2014 Developer's Guide (I believe it's the latest publicly available version as of today) talks about the parameters MFX_RATECONTROL_LA_CRF and mfxInfoMFX::CRFQuality. Judging by the param names, I guess they are probably something similar to the CRF setting in x264. Am I guessing it right ?

Assine o Gráficos