WEBVTT

00:00:00.240 --> 00:00:05.120
we all know by now that the rtx launch

00:00:03.600 --> 00:00:10.400
sucked hard hard enough to almost smell like an

00:00:07.919 --> 00:00:16.800
unabashed cash grab i mean this is nvidia's most expensive generation ever

00:00:13.440 --> 00:00:19.520
and what do we get even hang on when did

00:00:16.800 --> 00:00:19.520
this thing launch

00:00:21.840 --> 00:00:28.880
it's been eight months and ray tracing is still only supported by three games

00:00:26.640 --> 00:00:34.559
and two tech demos but hold on a minute

00:00:30.960 --> 00:00:37.120
now dxr or directx ray tracing is also

00:00:34.559 --> 00:00:40.160
available on gtx cards so

00:00:37.840 --> 00:00:44.800
the question we have to ask is was it all just a scam to get us to

00:00:41.840 --> 00:00:49.920
spend more on their shiny new cards let's take a closer look shall we

00:00:47.520 --> 00:00:54.640
amp by cued coders is the easy way to create and manage your game servers it

00:00:52.239 --> 00:00:58.879
supports dozens of applications runs on both windows and linux and is simple to

00:00:57.120 --> 00:01:01.520
install check it out at the link in the video description

00:01:08.880 --> 00:01:15.920
the fact that only a few games support rtx is important because each game's

00:01:13.760 --> 00:01:19.360
implementation uses raytracing in a different somewhat limited way more on

00:01:18.159 --> 00:01:24.080
that a bit that said we can get a good idea for how

00:01:21.360 --> 00:01:27.759
well ray tracing on gtx cards will work and we can finally give you guys some

00:01:25.920 --> 00:01:32.479
tangible numbers and show you exactly what kind of value rtx brings to the

00:01:29.520 --> 00:01:38.000
table but first the obvious question did nvidia scam early adopters by hiding

00:01:36.079 --> 00:01:41.520
pascal's ray tracing support behind a software lock

00:01:39.360 --> 00:01:45.680
the answer boils down to the dxr fallback layer the microsoft baked into

00:01:43.280 --> 00:01:50.000
directx 12. it allows ray tracing calculations to take place on a standard

00:01:47.439 --> 00:01:55.119
gpu shader cores rather than requiring dedicated hardware like the rt cores

00:01:52.159 --> 00:01:59.360
that rtx cards have that means that gpus with the necessary spec can merely have

00:01:57.520 --> 00:02:04.640
dxr fallback support exposed in the driver so

00:02:01.840 --> 00:02:08.800
yes nvidia did technically artificially lock ray tracing out of the pascal cards

00:02:07.040 --> 00:02:13.120
but the flip side to that is that this fallback layer requires some significant

00:02:11.680 --> 00:02:18.319
optimization to gain acceptable performance so nvidia would have had to put in some

00:02:16.480 --> 00:02:21.920
significant work to the driver level just beyond just enabling it and i'll

00:02:20.480 --> 00:02:28.000
leave you guys to debate whether they should have waited on that one in the comments section because today we have

00:02:25.520 --> 00:02:32.640
some rtx list ray tracing the test so we grabbed our gpu test bench along

00:02:30.080 --> 00:02:39.120
with four cards repping gtx we've got the pascal based gtx 1080 ti and touring

00:02:35.599 --> 00:02:42.959
based gtx 1660 ti going up against the

00:02:39.120 --> 00:02:44.879
rtx 2080 and rtx 2060. first let's take

00:02:42.959 --> 00:02:48.480
a closer look at ray tracing and see for ourselves whether or not it actually

00:02:46.480 --> 00:02:51.440
makes a huge difference in our games or if traditional tricks are still good

00:02:50.480 --> 00:02:55.680
enough as i alluded to earlier each of the

00:02:53.680 --> 00:03:00.000
three titles available does something different so let's start off with shadow

00:02:57.680 --> 00:03:05.519
of the tomb raider where rays are only traced to appropriately enough give us

00:03:02.959 --> 00:03:08.959
dynamic shadows and not for reflections or global illumination

00:03:07.280 --> 00:03:12.480
this is perhaps the simplest use of the technology as it relies only on the

00:03:10.879 --> 00:03:15.200
position and number of dynamic light sources for calculation

00:03:14.400 --> 00:03:19.840
but simple is a very deceptive term here

00:03:18.080 --> 00:03:24.239
because they take into account every light source candles spotlights area

00:03:22.640 --> 00:03:28.800
lights and even sunlight as a result shadows are natural looking

00:03:26.720 --> 00:03:33.040
and cast realistically across surfaces in ways that elude even the most

00:03:31.120 --> 00:03:37.440
sophisticated shadow maps or shadow volumes techniques that most games have

00:03:35.360 --> 00:03:43.120
used up until now moving on to battlefield v our rays are

00:03:39.840 --> 00:03:44.879
used not for shadows but for reflections

00:03:43.120 --> 00:03:48.560
because rays need to be traced for everything in the scene to accurately

00:03:46.720 --> 00:03:52.799
portray a reflection this can be an intensive operation

00:03:50.480 --> 00:03:57.519
depending on the number of and size of reflective surfaces visible as evidenced

00:03:55.360 --> 00:04:01.360
by the initial release of the game too many surfaces were incorrectly being

00:03:59.439 --> 00:04:05.760
processed as though they were reflective and this caused a major performance hit

00:04:03.760 --> 00:04:09.840
that has since been fixed so what's the image quality like

00:04:07.760 --> 00:04:15.439
there are reflections without dxr of course but these non-traced reflections

00:04:12.239 --> 00:04:17.359
rely on the screen space so if an object

00:04:15.439 --> 00:04:21.040
isn't on your screen it's not being reflected

00:04:18.479 --> 00:04:24.960
with txr enabled however you get a more accurate and more useful reflection

00:04:23.280 --> 00:04:28.320
potentially showing you things that are outside of your field of view

00:04:26.880 --> 00:04:31.759
usually this isn't a big deal for gameplay but being able to see enemies

00:04:30.400 --> 00:04:35.199
on the reflection of a window for example has probably saved at least a

00:04:33.680 --> 00:04:39.919
few digital butts from getting hurt online

00:04:37.199 --> 00:04:45.520
finally the most intensive scenario metro exodus this game uses ray tracing

00:04:42.720 --> 00:04:49.360
for global illumination which means all light sources and resulting shadows are

00:04:47.440 --> 00:04:52.479
calculated via dxr which in turn means there's a

00:04:50.720 --> 00:04:55.440
significant amount of number crunching required in order to actually run a game

00:04:54.240 --> 00:04:59.840
like this that's not to say that the standard rendering engine looks bad it's far from

00:04:59.199 --> 00:05:05.520
it in fact the effect that dxr adds to metro exodus can at times be quite

00:05:03.680 --> 00:05:08.800
subtle and difficult to discern especially underground

00:05:07.199 --> 00:05:13.199
the biggest advantage that raytracing lends to metro therefore

00:05:10.479 --> 00:05:16.479
is lighting accuracy or how light reflects and refracts around the scene

00:05:15.199 --> 00:05:22.240
to create a realistically lit environment particularly above ground with games day

00:05:19.840 --> 00:05:25.600
and night cycle ambient occlusion which can often otherwise be somewhat jittery

00:05:24.000 --> 00:05:31.039
and inaccurate is taken care of essentially the same way as shadow of the tomb raider's shadows creating a

00:05:28.639 --> 00:05:35.120
much more immersive feel of course nothing is without trade-offs and when

00:05:32.960 --> 00:05:39.919
we look at our performance results well as expected enabling dxr and shadow

00:05:38.080 --> 00:05:43.520
of the tomb raider resulted in a larger performance impact on our gtx cards than

00:05:42.160 --> 00:05:46.240
on rtx but overall the frame rates are quite

00:05:45.199 --> 00:05:50.880
playable when we move on to battlefield 5 again

00:05:48.800 --> 00:05:55.520
we're still playable but our gtx 1080 ti by comparison to

00:05:53.440 --> 00:06:01.120
traditional rendering tanks its frame rate so hard that it's basically halved

00:05:58.639 --> 00:06:07.280
but we ain't seen nothing yet it's when we try our gtx cards on metro exodus

00:06:04.080 --> 00:06:09.440
where things get messy even the gtx 1080

00:06:07.280 --> 00:06:14.000
ti struggled to give us cinematic frame rates at 1080p

00:06:11.039 --> 00:06:18.800
while our rtx 2060 doubled that and our rdx 2080 averaged over 100 fps

00:06:17.280 --> 00:06:23.520
all of which means if you're planning to stick with pascal and want to use dxr

00:06:22.080 --> 00:06:28.319
you should pay close attention to what exactly a game's rays are being used for

00:06:26.000 --> 00:06:34.080
shadows and reflections probably fine global elimination to quote hardware

00:06:30.400 --> 00:06:37.199
canucks don't bother so is it worth it

00:06:34.080 --> 00:06:38.960
well while gtx owners can run dxr the

00:06:37.199 --> 00:06:43.120
computationally expensive global elimination method is the direction more

00:06:40.880 --> 00:06:47.039
impressive games will try to take and as a result rtx cards will gradually claim

00:06:45.680 --> 00:06:51.440
their place in the market for the enthusiasts and eventually may even

00:06:49.280 --> 00:06:56.160
become a necessity whether your upgrade path is team green or team red when navi

00:06:53.600 --> 00:06:59.360
based radions launch with dxr support chances are you'll be doing so sooner

00:06:57.840 --> 00:07:04.880
than later if you want the best graphics in your games and these games are coming we spoke with

00:07:03.039 --> 00:07:08.880
epic games engineering director marcus wasper and while the complexity of

00:07:07.280 --> 00:07:13.599
unreal engine meant that it took both his team and nvidia the better part of a

00:07:11.280 --> 00:07:18.240
year to get it implemented from here on out game developers using ue4 are

00:07:16.160 --> 00:07:21.840
finally going to have that it just works experience nvidia's ceo jensen wong was

00:07:20.720 --> 00:07:27.520
talking about without interfering with traditional baked light mapping techniques

00:07:25.440 --> 00:07:31.039
taken together support like this in unreal engine and other popular game

00:07:29.360 --> 00:07:36.080
engines means that there's no significant increase in workload for

00:07:32.800 --> 00:07:38.319
developers so content creators can more

00:07:36.080 --> 00:07:41.520
easily create dxr enabled games for us consumers

00:07:39.759 --> 00:07:46.080
whatever their reasoning rather than transform the landscape overnight in my

00:07:44.080 --> 00:07:50.000
opinion what nvidia did with rtx was race out ahead of the competition and

00:07:47.599 --> 00:07:54.160
kickstart the ray tracing conversation how many people actually cared when

00:07:51.759 --> 00:07:57.680
microsoft announced dxr back in march of 2018

00:07:55.520 --> 00:08:02.240
did anyone think that hardware would be able to implement it so soon i sure

00:08:00.080 --> 00:08:06.800
didn't and it certainly caught amd off guard if radion7 is any indication

00:08:04.879 --> 00:08:10.879
and now that the next generation consoles are coming even they are

00:08:08.560 --> 00:08:13.039
rumored to have ray tracing capabilities so

00:08:11.680 --> 00:08:20.560
while this might not have been the ray tracing generation it's truly set us up for the next one

00:08:17.680 --> 00:08:26.960
the volta xl charging usb cable uses reversible n52 grade neodymium magnets

00:08:23.680 --> 00:08:28.960
with usbc tips it provides instant power

00:08:26.960 --> 00:08:33.599
and it's got pvv braiding for flexibility and strength they feature

00:08:31.440 --> 00:08:37.760
intelligent strain relief collars and can handle up to 87 watts of charging so

00:08:36.080 --> 00:08:43.599
that means you can charge a macbook pro 15-inch nintendo switch a pixel 3 xl and

00:08:41.599 --> 00:08:47.600
more you can also use it for on-the-go charging using one smartphone to charge

00:08:45.839 --> 00:08:51.680
another it comes in three different lengths half one and one and a half

00:08:49.839 --> 00:08:55.200
meters so check it out at the link in the video description

00:08:53.440 --> 00:08:58.560
thanks for watching guys if this video sucked you know what to do but it was

00:08:56.720 --> 00:09:01.519
awesome get subscribed hit that like button or check out the link to where to

00:09:00.080 --> 00:09:06.480
buy some of the stuff we featured in the video description also link down there

00:09:03.760 --> 00:09:12.000
is our merch store which has cool shirts like this one and our community forum

00:09:08.480 --> 00:09:12.000
which you should totally join