QuickSync vs NVEnc, which one is more reliable and has better response time?

[ivan001 2]

I'm setting up a new server for Emby which has an Intel i7-7700 CPU and a Nvidia GTX 1050 ti.

I'm trying to figure out which hardware encoder would be a better option.

I've read around the forum about quality and energy efficiency of Intel's QuickSync, but I couldn't find anything on reliability (how often either will crash) or response time when seeking through the video.

It has happened to me many times that when I'm watching a video which is being transcoded and try to skip ahead for let's say 5 minutes, the playback will hang and either take a long time to restart or not restart at all. I would like to avoid that. Did anyone test both options to see which one performs better?

 

 

[Jdiesel 1431]

They are pretty similar but I think the pascal chip in the 1050 support 12bit hvec decode whereas the Kaby Lake igpu does not. I don't know if hardware accelerated tone mapping is supported on either...maybe someone else can comment.

 

https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new

https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video#Hardware_decoding_and_encoding

 

Knowing you OS might be helpful too. Also be aware that the Nvidia GPU has an artificial stream limit. This can be removed with software. The Intel GPU has no such restriction.

 

It comes down to whether you need specific features, want to give up a pci-e slot, and want the extra 20w of idle power draw from the Nvidia GPU. Only you can answer that.

Edited September 5, 2025 by Jdiesel

[RanmaCanada 494]

On 05/09/2025 at 02:55, ivan001 said:

I'm setting up a new server for Emby which has an Intel i7-7700 CPU and a Nvidia GTX 1050 ti.

I'm trying to figure out which hardware encoder would be a better option.

I've read around the forum about quality and energy efficiency of Intel's QuickSync, but I couldn't find anything on reliability (how often either will crash) or response time when seeking through the video.

It has happened to me many times that when I'm watching a video which is being transcoded and try to skip ahead for let's say 5 minutes, the playback will hang and either take a long time to restart or not restart at all. I would like to avoid that. Did anyone test both options to see which one performs better?

 

 

Both of those are ancient, and horrible. If you are going to use hardware encoding, seriously upgrade to newer equipment. If this is just going to be an encode server with the movies and content stored on a NAS, you can easily buy a refurbished i3-12100 for under $200 USD. If your motherboard supports resizable bar, get an Arc A310 or A380 instead of the ancient 1050Ti. Your system will have problems with 4k content, specially with tone mapping as it won't have enough umpf.

[SikSlayer 249]

On 9/7/2025 at 2:14 AM, RanmaCanada said:

Both of those are ancient, and horrible. If you are going to use hardware encoding, seriously upgrade to newer equipment. If this is just going to be an encode server with the movies and content stored on a NAS, you can easily buy a refurbished i3-12100 for under $200 USD. If your motherboard supports resizable bar, get an Arc A310 or A380 instead of the ancient 1050Ti. Your system will have problems with 4k content, specially with tone mapping as it won't have enough umpf.

Even the cheaper N150/N300/N355 are better for this. I agree, get something newer. Better performance/features, and less power draw.

[ivan001 2]

On 9/5/2025 at 11:00 PM, Jdiesel said:

They are pretty similar but I think the pascal chip in the 1050 support 12bit hvec decode whereas the Kaby Lake igpu does not. I don't know if hardware accelerated tone mapping is supported on either...maybe someone else can comment.

 

https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new

https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video#Hardware_decoding_and_encoding

 

Knowing you OS might be helpful too. Also be aware that the Nvidia GPU has an artificial stream limit. This can be removed with software. The Intel GPU has no such restriction.

 

It comes down to whether you need specific features, want to give up a pci-e slot, and want the extra 20w of idle power draw from the Nvidia GPU. Only you can answer that.

Thanks. I'm running Debian 13 (Trixie). I'm only streaming to my family of 4, so it's unlikely that there will be more than 2 or 3 streams running at the same time. My TV supports pretty much anything I throw at it, so there's hardly ever any conversion going on there.

I haven't measured the GPU consumption when idle yet, but I doubt it comes close to 20W. Before installing the GPU the computer was idling at around 16W with the HDD spinning so that will be easy to check.