Comparison between H.264 and H.265 with Citrix XD 7.16

I’m currently in the process to prepare a session for the upcoming GTC in San Jose. So I thought it is a good time to share some first results from my testing during the last weeks regarding the topic above.

As you might have seen, Citrix released H.265 support with latest XenDesktop 7.16:

H.265 Encoding Now Available for XenDesktop Using NVIDIA GPUs!

And I already got different questions from customers around this topic regarding our Tesla boards. Not all Tesla boardsĀ  support H.265 encoding. So have a look on our matrix to check the boards accordingly. Link here

Be aware that implementing H.265 with Citrix requires different requirements on server and client side as well.

On the sever side:

  • Make sure you have a Maxwell or Pascal board that supports H.265 encoding
  • Having the correct Citrix licensing (Platinum) and set the correct Citrix Policies (Optimize for 3D workloads, Enable hardware encoding)

On the client side:

  • Make sure your endpoint supports H.265 decoding (check with DXVA checker for example)
  • Set the right policy with the Citrix receiver admx template on your endpoint (Enable H.265 decoding)

 

But from my perspective what is even more important why customers really want to use H.265? Just as it is something new? There are different dependencies for get it up and running and I want to show you my results and my conclusion where it would make sense to use H.265.

As you can read in different blogs and sites there is a promise that H.265 saves bandwidth compared to H.264 at a better quality. So I started to compare a Win10 VDI session with H.264 YUV420 and H.265 YUV420 in regards of bandwidth usage when playing a FullHD video in a 1080p FullScreen session.

 

So what I did here was playing a 2:30min video and you can see on the RDAnalyzer output that the bandwidth consumption is almost identical between both protocols. 392MB (H.264) to 393MB (H.265)

So at least for this test there is no reduction in bandwidth consumption but what we can see there is a real increase in image quality as we can see with this comparison:

 

 

 

 

 

This is a direct comparison between the two different codecs with H.264 on the left and H.265 on the right and we see way less chroma subsampling effects.

 

Conclusion:

H.265 might become an interesting choice for specific workloads that needs better image quality on a decent bandwidth consumption.
In the past I always recommended to use H.264 YUV444 for such kind of use cases but sometimes the highly increased bandwidth consumption with H.264 YUV444 might not be acceptable or possible and here comes H.265 into play.

Tags :

About the Author

sschaber

GRID Solution Architect and owner of this Blog site.

Leave a Reply

Your email address will not be published. Required fields are marked *