RTX 3070 NVENC Video Encoder Speed Test VS GTX 1660 8K 60FPS In Edius X Which Is Faster?

RTX 3070 NVENC Video Encoder Speed Test VS GTX 1660 8K 60FPS In Edius X Which Is Faster?

In this video I'm doing an NVENC video encoding speed test at 8K 60FPS.

This is the Nvidia RTX 3070 VS the Nvidia GTX 1660 and their NVENC encoders, to see which one is the fastest at encoding the 8K 60FPS HEVC H.265 output.

You may think that this is an Ampere VS Turing speed test but it isn't as both the RTX 3070 and the GTX 1660 share the same Nvidia Generation 7 NVENC encoder, basically the Turing NVENC encoder.

So, while the RTX 3070 is the way more powerful GPU as far as its core Ampere GPU architecture is concerned, it's not this GPU technology that does the NVENC HEVC H.265 encoding but the Gen7 Turing based NVENC encoder.

It is also worth noting that the RTX 3080 and the RTX 3090 also have the same NVENC encoder that the GTX 1660 does.

In fact. As and when Nvidia may or may not release any other skews of the RTX 3000 series, be that an RTX 3080 Ti, RTX 3090 Ti, RTX 3070 Ti, RTX 3060 Ti or anything as a Super skew such as an RTX 3070 Super or an RTX 3080 Super. They will all have the same Gen 7 Turing NVENC encoder that the GTX 1660 has.

So, if your only interest in an Nvidia GPU is for NVENC encoding in Edius X, then all you need is the GTX 1660, which is the lowest and cheapest Nvidia GPU that has the Gen 7 Turing NVENC encoder in it.

Video chapters.

00:00 Introduction
01:16 Explanation of test and GTX 1660 setup
05:49 GTX 1660 speed test
07:19 Explanation of RTX 3070 setup
08:29 RTX 3070 speed test
09:48 End summary
13:59 End title

If you liked this video please give it a thumbs up, share it, subscribe and click the bell.

As an Amazon Associate I earn from qualifying purchases. I use this commision for buying review equipment & for my YouTube production studio. Thank you.

My Amazon Shop links.

Amazon global links.

My ethics.
I do not review Amazon products given to me for free in exchange for an Amazon website product review. DO NOT ASK.

I do review products available on Amazon but these are products that I’ve either bought myself or products sent to me for free or on loan to be reviewed specifically for my YouTube channel.

I always mention at the start of any videos if I’ve received any of the products being reviewed for free or on loan.

Other than technical input that may be useful to the viewer, I never accept any input from anyone who sends me products for my reviews.

I always try to be fair and balanced with my reviews but will always make it very clear if I find any issues with any products that I use or review.

Contact me for product reviews, I do not respond to email requests for help or advice.

My web links and video playlists.

I’m David Harry, thank you very much for watching this video, take care and goodbye now.

Cheers,
Dave.

#NVENC#NvidiaRTX3070#SpeedTest

[aussieexpat]

Thank you so much for doing this. It's shocking just how few people actually compare video cards for productive work loads! I've been looking around for days on the simple question, would I get an upgrade by going from a 3060 to a 3080.

[tyson]

It's like seeing a really happy pug. You cheered me up today!

[gnolygnolls]

Thanks, great video. Do you have any idea why the 1660 is faster than the 3070?

[ryekorodynator]

Maybe its problem with video drivers for RTX3070. I have RTX3070 too and it is slowest than my GTX1650 in notebook - NVENC encoding. I tested with FFMPEG and HANDBRAKE.

[TonesAdventuresHD]

I definitely can sleep better tonight knowing I can use either gpu 😀👍🏼

[gianlucapx]

It's an old video, but it's worth to say that today there are programs that use the gpu in different ways when encoding: some use just the encoder, some the cpu and the encoder, some other the cuda cores. And there are some programs that make use also of the tensor cores of the nvidia gpus.
So if someone wants to upgrade the hardware to improve his productivity, he should analyze the way in which is used the gpu by the programs that he uses the most. If they make use of cuda cores, then go for the card that has the higher count of cuda cores.
You can check the usage of cuda cores with the windows task manager: open it, select the performance tab, then gpu and there click on the arrow on the voice 3D in order to open the Cuda monitor.
In this test the 3070 was slower, but that could be caused by different factors, starting from the driver. It's better to use the studio driver when the card is used for productivity and rendering, because they are optimized for those tasks.
If I had to consider to upgrade to it, I'd do it in a blink having the money. 5888 cuda cores, tensor and rt cores, high vram bandwidth. The specs are too good and considering the fact that people use different programs for different tasks, the card can make certain things way faster that what would possibile with an old gen card with the specs of the GTX 1660 (only 1408 cuda cores).

[d]

Very interesting results. Many thanks for sharing. How are the differences between NVENC and Quick-Sync?

[imaner]

Good evening Scouse. I still can't get a 3090, there seems to be.. a supply issue.

[MaxBarwell-sxir]

Would a rtx 3070 with a i9900f, 32gb ram be good enough for streaming, preferably nvenc

[hayabusaTravels]

For rendering though, I use Voukoder free plugin in Adobe Premiere and this gives me at least 1:1 rendering time on very complex timelines with effects and other stuff. For simpler timelines I get far better rendering times. For a 10min simple video I get 5-7min rendering time, depending on the timeline complexity and I'm ok with that.
Sure, 32GB of RAM would be better for editing and stuff, but for now I have no need for more RAM. I'm too lazy to install more RAM as this will imply removing the CPU fan 🤣.
My system is already old but there's no need for upgrading.
If you use Adobe Premiere you will get really fast rendering times with Voukoder and any NVENC enabled GTX cards. Of course, the newer the card, the better the results as Voukoder is a beast on .NVENC cards. It works with AMD cards as well.
Can only imagine what Voukoder will do with a 3090 and the latest Intel processor.

[rookie]

hello, i found your video incredibly useful, i have the 1650 super, i have seen that the 1650 super has the same architecture of the 1660.... so even the 1650 super has Gen 7 Turing NVENC encoder, right?
I think that the real advantage of having a gpu like that is also that the editing in realtime with very heavy file is a lot smoother compared to other low budget gpu's like 1660, do you agree with me?

[ozgur]

Hello, a friend of me says that although Turing and Ampere have the same encedor, they use different alghorhytms and Ampere has a better algo so better qualiity.. Is this true?

And 1650 also is Turing, so the same right?

[andrzejkowalski]

Hi, I have Geforce GTX1650, where from I can download NVENC Encoder exe. installation file for Windos 10, how to get it?

[ysfexsew]

I was so looking forward to Ampere for an improved NVENC, but when I found out that it was the exact same basically, I immediately decided to skip this gen. Let's Lovelace is different.

[DbugII]

Interesting, I was considering upgrading my graphic card to have faster video encoding (in my case it's ffmpeg and shotcut), but it looks like going RTX is not the way to go... sad.

So question: I currently have a Intel i7 4790k (running at 4.6Ghz), 32 gigs of ram, and a GTX 1070ti, everything running on SSD, what do you think would help speed up encoding?

[田田-xy]

Hello, David, forgive me for commenting in unskilled English.For a year, EDIUS has not fulfilled its commitment to support more GPUs. I am so disappointed.
Any NLE user has at least one AMD or NVIDIA GPU, These two SDKs are free, and a large number of users are ignored by them. I didn't expect EDIUS to preferentially adapt to uhd770.

EDIUS export speed is slow because its playback window rasterization consumes a lot of CPU resources(EDIUS use d3dx9a). Only play mp3 audio in 8K resolution, lead to 4 - cores full load, Four cores of computing power wasted. Gvjob export will also go through this process, and EDIUS turns a blind eye to this matter.

To be honest, EDIUS has been surpassed by China's free NLE software Jianying in all aspects. Support all Gpus, voice recognition subtitle accuracy up to 98%, Ai image matting.