WEBVTT

00:00:00.760 --> 00:00:08.840
x264 quicksync video nvnc

00:00:05.000 --> 00:00:11.719
vce there is no shortage of ways to

00:00:08.840 --> 00:00:18.920
transform your screen into a video stream visible to potentially millions

00:00:14.559 --> 00:00:23.000
of people around the world for the Mad

00:00:18.920 --> 00:00:26.720
Ballers Among Us a second PC dedicated

00:00:23.000 --> 00:00:30.439
to that task is the Undisputed king of

00:00:26.720 --> 00:00:33.559
performance and quality but how do the

00:00:30.439 --> 00:00:38.320
rest of us get our streams going without

00:00:33.559 --> 00:00:39.800
Crossing them up AMD claims to have the

00:00:38.320 --> 00:00:48.480
best solution but do they welcome to a new

00:00:43.559 --> 00:00:53.450
episode of Manufacturers

00:00:55.960 --> 00:01:02.519
say browse privately and securely with

00:00:59.519 --> 00:01:04.839
tunnel bear the simple VPN app try

00:01:02.519 --> 00:01:08.799
tunnel bear for free at the link in the video

00:01:05.920 --> 00:01:15.680
description one of AMD's top marketing points is that the 16 threads in their

00:01:11.759 --> 00:01:19.240
ryzen 7 CPUs allow Gamers to dedicate

00:01:15.680 --> 00:01:21.840
ample CPU resources to both gaming and

00:01:19.240 --> 00:01:28.960
video streaming coding without impacting the performance of either since games

00:01:25.680 --> 00:01:32.079
won't make use of all 16 threads and

00:01:28.960 --> 00:01:34.799
neither will an coder meanwhile one of

00:01:32.079 --> 00:01:41.200
Intel selling points for their consumer level chips is their integrated graphics

00:01:37.960 --> 00:01:44.240
processor and its ability to use

00:01:41.200 --> 00:01:46.759
quicksync video technology to encode a

00:01:44.240 --> 00:01:52.920
video stream without impacting the performance of their

00:01:48.960 --> 00:01:55.799
CPU well that all sounds great but with

00:01:52.920 --> 00:02:01.240
both of them claiming to achieve the same objective by very different means

00:01:59.360 --> 00:02:08.840
well surely one of them's got to be better right well predictably enough AMD

00:02:04.759 --> 00:02:12.000
says that their way software encoding is

00:02:08.840 --> 00:02:14.280
superior and the traditional wisdom

00:02:12.000 --> 00:02:19.160
would back them up on this one the software-based

00:02:15.480 --> 00:02:21.680
x264 encoder is well known to offer

00:02:19.160 --> 00:02:27.879
higher image quality and more flexibility in its setup compared to

00:02:24.200 --> 00:02:30.200
other Solutions but until now if you

00:02:27.879 --> 00:02:34.640
cared about performance you'd almost never use it while gaming because its

00:02:32.519 --> 00:02:38.680
consumption of system resources was pretty much guaranteed to impact your

00:02:36.720 --> 00:02:43.760
gameplay unless you were running an extreme Edition or something like that

00:02:41.560 --> 00:02:50.720
but with more processing cores at ryzen 7's disposal it could be that AMD is on

00:02:47.400 --> 00:02:53.280
to something meanwhile what are the

00:02:50.720 --> 00:03:00.239
other guys doing different here Intel's Quicks sync video and for that matter

00:02:55.760 --> 00:03:02.599
NVIDIA's nvnc and rtgs radon video codec

00:03:00.239 --> 00:03:09.159
engine are all what are known as dedicated hardware-based encoders which

00:03:06.319 --> 00:03:16.319
present an interesting conundrum in theory you actually stand to lose

00:03:12.239 --> 00:03:19.400
virtually no performance which is cool

00:03:16.319 --> 00:03:21.440
but you're forced to use the optimized

00:03:19.400 --> 00:03:26.840
settings that are baked into the hardware meaning that you can't really

00:03:23.879 --> 00:03:32.879
tune anything beyond their presets besides simply increasing the raw bit

00:03:29.439 --> 00:03:35.599
rate to improve quality so finally back

00:03:32.879 --> 00:03:40.640
to the big questions how much does the encoder matter for quality is twitch's

00:03:38.200 --> 00:03:45.280
new maximum of 6 megabits per second enough that it won't really matter

00:03:42.280 --> 00:03:47.799
anymore are any of these Solutions

00:03:45.280 --> 00:03:51.840
really able to untether you from the second PC and let you stream without

00:03:50.640 --> 00:03:57.720
impacting performance well we touched on this from

00:03:54.239 --> 00:03:59.480
AMD's perspective in our ryzen 7 review

00:03:57.720 --> 00:04:03.840
which you can check out here and we asked you guys at that time if you

00:04:01.280 --> 00:04:09.720
wanted something more in depth and since you did ask so nicely your wish is our

00:04:07.480 --> 00:04:14.879
Command we're going to focus on two games for our testing first ashes of the

00:04:12.040 --> 00:04:20.280
singularities CPU Benchmark not because it's a hot game on Twitch right now but

00:04:17.799 --> 00:04:25.960
for technical reasons it stresses the CPU and has a good mix of slow moving

00:04:23.160 --> 00:04:30.560
terrain with lots of small fast-moving projectiles that are great at

00:04:27.759 --> 00:04:35.639
introducing blocky artifact during encoding the second will be the for

00:04:33.199 --> 00:04:42.639
Honor Benchmark with its sweeping cameras and loads of special effects we

00:04:39.160 --> 00:04:45.120
know again most people aren't streaming

00:04:42.639 --> 00:04:49.759
these games but canned benches are much more repeatable across our platforms and

00:04:47.800 --> 00:04:55.000
both of these titles represent a realistic gaming load that is more

00:04:52.400 --> 00:05:00.080
demanding than today's popular Esports titles for our encoders we've used

00:04:57.919 --> 00:05:07.120
common Twitch streaming bit rates and resolutions for all of our comparisons

00:05:02.960 --> 00:05:09.680
you'll note that 1080p30 is technically

00:05:07.120 --> 00:05:16.840
more data to cram into our lower bit rates than 720p60 but the higher frame

00:05:13.759 --> 00:05:18.919
rate of 720p60 should better highlight

00:05:16.840 --> 00:05:26.800
motion artifacts in our encoders so any CPU even one with

00:05:22.319 --> 00:05:29.360
onboard Graphics can use x264 encoding

00:05:26.800 --> 00:05:35.440
but we observed in our testing that not all are created equal in this regard we

00:05:32.479 --> 00:05:39.919
found that AMD's ryzen 7 managed to oh excuse

00:05:36.360 --> 00:05:42.080
me just snuck up on me managed to avoid

00:05:39.919 --> 00:05:48.840
dropping frames throughout our testing at a higher quality preset than either

00:05:45.479 --> 00:05:50.600
of Team Blues Solutions in some cases

00:05:48.840 --> 00:05:56.199
even some of our hardware encoders needed lower presets so to find out how

00:05:53.840 --> 00:06:02.039
these differences in encoding settings affect real world image quality we

00:05:59.000 --> 00:06:04.600
captured a lossless reference recording

00:06:02.039 --> 00:06:10.680
with our Blackmagic intensity Pro and used it for comparison so we can see

00:06:07.919 --> 00:06:16.479
some subtle differences here ryzen manages to retain smaller details that

00:06:13.520 --> 00:06:22.680
get smeared out on team blue at their very fast and super fast settings

00:06:19.280 --> 00:06:24.800
compared to ryzen's faster this not only

00:06:22.680 --> 00:06:30.919
translates into better picture quality overall but also contributes to better

00:06:27.639 --> 00:06:33.800
quality when twitch then re-encodes it

00:06:30.919 --> 00:06:41.000
for its various quality options moving on here we see our Hardware encoders at

00:06:36.759 --> 00:06:45.199
1080 30p at 2500 kilobit per second

00:06:41.000 --> 00:06:48.039
alongside our ryzen x264 footage of the

00:06:45.199 --> 00:06:54.840
four x264 on ryzen appears to be the least likely to lose small projectiles

00:06:51.080 --> 00:06:57.479
in ashes of The Benchmark nvnc and

00:06:54.840 --> 00:07:02.639
quicksync don't seem half bad here to be clear but they do block up more often

00:07:00.479 --> 00:07:11.199
and lose finer details when A lot's going on at once rtgs AMF encoder on the

00:07:06.000 --> 00:07:14.240
other hand sorry radon fans but

00:07:11.199 --> 00:07:16.319
uh things get more interesting again at

00:07:14.240 --> 00:07:21.680
720p60 x264 holds up best once more although

00:07:19.680 --> 00:07:27.120
the shimmering of compression artifacts on the cobblestones as the camera flies

00:07:24.240 --> 00:07:30.639
by is pretty noticeable but not nearly as noticeable as it is with the rest of

00:07:29.000 --> 00:07:36.080
our contest nvnc is again at the head of the

00:07:33.000 --> 00:07:38.960
hardware pack looking decent but our

00:07:36.080 --> 00:07:44.319
quick sync and AMF footage are both abysmal at times blocking up so badly as

00:07:42.440 --> 00:07:50.360
to appear as though they'd been recorded at 240p instead of

00:07:47.639 --> 00:07:56.720
720p we'll keep our visual comparisons rolling while we move on to Performance

00:07:53.280 --> 00:07:59.919
with x264 we see a respectable showing

00:07:56.720 --> 00:08:03.240
from our ryzen 7 1800x at averaging

00:07:59.919 --> 00:08:06.680
about a 5 to 10% performance penalty on

00:08:03.240 --> 00:08:09.960
the faster and very fast presets over

00:08:06.680 --> 00:08:12.879
our Baseline reading on the 6900k

00:08:09.960 --> 00:08:18.680
meanwhile we get an average of a roughly 3 to

00:08:14.759 --> 00:08:22.360
23% Performance drop clearly a much more

00:08:18.680 --> 00:08:23.879
varied result while our 7700k could get

00:08:22.360 --> 00:08:30.919
as high as a 36% hit both of these ones on the super

00:08:27.440 --> 00:08:33.599
fast preset AMD high-speed Infinity

00:08:30.919 --> 00:08:39.440
fabric interconnect secret sauce for ryzen's video encoding while gaming

00:08:36.159 --> 00:08:42.640
performance looks like it's working so

00:08:39.440 --> 00:08:45.120
far our Hardware tests meanwhile show

00:08:42.640 --> 00:08:52.279
ryzen actually taking a wider hit with nvenc at roughly 5 to 15% on average

00:08:48.959 --> 00:08:56.560
while our 6900k managed a 3 to 10%

00:08:52.279 --> 00:09:00.720
variance and our 7700k ranged from 2 to

00:08:56.560 --> 00:09:04.279
18% rg's AMF encoder seemed to hit all

00:09:00.720 --> 00:09:06.839
of our chips less with ryzen taking a

00:09:04.279 --> 00:09:13.680
roughly 7 to 8% performance hit and sitting in between the other two though

00:09:11.200 --> 00:09:20.200
as we've seen the AMF encoder is also far from the highest quality for

00:09:16.519 --> 00:09:23.079
streaming but we have one final encoder

00:09:20.200 --> 00:09:29.519
one that we can't directly compare with any other CPU and that's Intel's

00:09:25.760 --> 00:09:32.120
quicksync video compared to the 7700k

00:09:29.519 --> 00:09:37.279
Baseline it caused a 3 to 12% performance drop about in line with the

00:09:34.320 --> 00:09:44.880
rest of our Hardware encoders so finally then lining all our encoders up next to

00:09:40.480 --> 00:09:46.760
each other we can see that actually

00:09:44.880 --> 00:09:54.720
ryzen running x264 blows the doors off every other CPU

00:09:51.279 --> 00:09:56.920
encoder combination it has a tight

00:09:54.720 --> 00:10:02.880
variation in its performance impact over our Baseline numbers and it has a higher

00:10:00.160 --> 00:10:09.839
quality CPU preset meaning that when twitch re-encodes you lose less detail

00:10:06.120 --> 00:10:13.040
overall so taken together it looks like

00:10:09.839 --> 00:10:14.640
AMD's marketing department wasn't asleep

00:10:13.040 --> 00:10:20.600
at the wheel on this one Squarespace takes the hassle of

00:10:18.360 --> 00:10:26.320
building a website for your small business for your portfolio for the for

00:10:23.720 --> 00:10:31.760
the sports team that you help run and it throws it all in the garbage because

00:10:28.000 --> 00:10:33.959
Squarespace is super super easy to use

00:10:31.760 --> 00:10:39.040
it's only 12 bucks a month to start you get a free domain if you buy Squarespace

00:10:35.880 --> 00:10:41.920
for a year they offer 247 support via

00:10:39.040 --> 00:10:45.720
live chat and email and all you pretty much got to do you might not even need

00:10:43.600 --> 00:10:51.839
it is pick one of their gorgeous templates put in all your own text copy

00:10:48.480 --> 00:10:54.279
and all your own images and boom you've

00:10:51.839 --> 00:10:59.320
got a beautiful functional website that you can spend your time updating rather

00:10:57.040 --> 00:11:04.079
than spending your time fighting with compatibility of so check it out at the

00:11:02.160 --> 00:11:10.519
link in the video description and start a free trial with no credit card

00:11:05.959 --> 00:11:12.959
required then use offer code LT to save

00:11:10.519 --> 00:11:16.800
10% so thanks for watching guys dislike if you disliked leave a like though if

00:11:14.720 --> 00:11:20.800
you liked it and get subscribed and maybe check out where to buy the stuff

00:11:18.560 --> 00:11:24.760
we featured at the link in the video description also down there is our merch

00:11:22.480 --> 00:11:29.560
store as well as our community Forum which you should totally join