WEBVTT

00:00:00.160 --> 00:00:05.839
the rtx 3080 is the most powerful GPU

00:00:03.280 --> 00:00:09.200
we've ever seen at this price point but

00:00:06.799 --> 00:00:13.679
we've said that before and in fact we say it every time there's a new graphics

00:00:10.960 --> 00:00:17.600
card generation so what makes this one so darn special i mean the hype

00:00:15.519 --> 00:00:22.560
surrounding it was so huge that even we got caught up in it i mean look how

00:00:19.520 --> 00:00:24.080
excited this guy is it's infectious but

00:00:22.560 --> 00:00:30.320
as we all know hype disappoints and absolute hype

00:00:26.880 --> 00:00:33.040
disappoints absolutely

00:00:30.320 --> 00:00:36.719
unlike our sponsor glassware glasswire is the tool that shows you which apps

00:00:34.719 --> 00:00:40.320
are slowing down your connection in real time and it's used by security pros to

00:00:38.480 --> 00:00:46.960
monitor for malware block bandwidth wasters and detect suspicious activity

00:00:42.399 --> 00:00:46.960
get 25 off today at the link below

00:00:54.079 --> 00:01:01.600
you hear that it's rattling weird um while you read over our test

00:00:59.120 --> 00:01:05.760
platform let me get this off of my chest NVIDIA man you guys really went out of

00:01:04.479 --> 00:01:10.799
your way to set yourselves up here didn't you like i don't

00:01:08.000 --> 00:01:15.520
get it spoiler alert you're shipping the fastest graphics card on the market at

00:01:13.119 --> 00:01:20.560
half the price of the previous king why don't you just leave it there why

00:01:18.000 --> 00:01:26.520
push your luck with claims like twice as fast as rtx 2080 because it's not

00:01:30.640 --> 00:01:37.280
now to be very clear it's not like i'm

00:01:33.680 --> 00:01:38.880
saying you know burns 3080 biggest

00:01:37.280 --> 00:01:43.439
disappointment ever because right out of the gate at 1440p

00:01:41.600 --> 00:01:49.200
we're looking at a performance lead over not just the price comparable last gen

00:01:45.680 --> 00:01:52.640
2080 but even the 2080 ti

00:01:49.200 --> 00:01:55.840
and even with rtx disabled i mean this

00:01:52.640 --> 00:01:57.200
is a freaking awesome leap in raw

00:01:55.840 --> 00:02:04.079
horsepower overall we saw anywhere from a 20 to 75

00:02:01.920 --> 00:02:09.039
uplift over the 2080 with the biggest benefits coming in directx 12 and vulcan

00:02:06.960 --> 00:02:15.120
titles so then the only reason i'm disappointed at all

00:02:12.480 --> 00:02:20.640
is because NVIDIA told me to expect double the performance of the 2080 which

00:02:18.080 --> 00:02:26.080
as it turned out was anywhere from a small embellishment to a fish this big

00:02:23.120 --> 00:02:30.879
whopper but wait okay because perhaps there's more to this

00:02:27.920 --> 00:02:35.360
could our lackluster cs go performance numbers hold the key

00:02:32.800 --> 00:02:41.120
look how CPU-bound this game ends up even at 1440p so let's kick it up a

00:02:38.400 --> 00:02:47.200
notch to 4k where we're gonna leave out our rx 5700 xt so we can test out dlss

00:02:45.120 --> 00:02:53.920
NVIDIA's deep learning up sampling technology and ooh is that

00:02:51.280 --> 00:03:00.879
41 frames per second in microsoft flight simulator at 4k ultra no less holy sh

00:02:58.239 --> 00:03:05.200
i mean not only that but the minimum of 34 FPS

00:03:02.720 --> 00:03:10.480
that is perfectly playable in a game that's more focused on visuals versus

00:03:07.840 --> 00:03:14.879
fast movement which means that we can definitively say yes it can run flight

00:03:13.360 --> 00:03:21.280
sim 2020 like damn now in general we're still not seeing

00:03:18.800 --> 00:03:26.800
that 2x performance bump with results closer to 75

00:03:23.599 --> 00:03:29.440
but we're also still outpacing our rtx

00:03:26.800 --> 00:03:35.360
2080 ti by anywhere from 10 to 30 percent and at 4k cs go really lets the

00:03:32.959 --> 00:03:41.360
3080 flex its big gains in memory bandwidth thanks to its 10 gigs of

00:03:37.720 --> 00:03:44.159
gddr6x memory that is a very small drop

00:03:41.360 --> 00:03:49.440
in performance compared to 1440p so alright then while it may not be exactly

00:03:46.879 --> 00:03:54.239
what NVIDIA promised us the 3080 is still a gaming monster

00:03:52.560 --> 00:03:56.879
maybe productivity will give us a clearer understanding of what they meant

00:03:55.680 --> 00:04:01.840
by that whole double the performance thing

00:03:58.959 --> 00:04:03.840
and yeah okay that's double

00:04:03.120 --> 00:04:10.640
and well ahead of the rtx 2080 ti in fact in

00:04:06.799 --> 00:04:12.480
blender the rtx 3080's cuda score is so

00:04:10.640 --> 00:04:18.959
good that it rivals excuse me scratch that beats the rtx

00:04:15.760 --> 00:04:20.000
2080s rtx optimized optics render time

00:04:18.959 --> 00:04:24.560
like excuse me pardon and don't even get me

00:04:22.400 --> 00:04:28.560
started on the 3080s optics times i mean this

00:04:25.600 --> 00:04:33.440
this is game-changing for students and prosumers and even professionals looking

00:04:30.960 --> 00:04:37.280
to build affordable and powerful 3d rendering stations spec view perf does

00:04:35.680 --> 00:04:40.479
bring us back down to earth somewhat with performance that's more in line

00:04:38.880 --> 00:04:44.880
with our gaming numbers which is a bummer but at least it gives us a hint

00:04:43.040 --> 00:04:48.400
as to where NVIDIA may have been focusing when they came up with that

00:04:46.880 --> 00:04:53.280
double figure interestingly though team red pulled off

00:04:50.800 --> 00:04:59.040
a couple of wins here too so while the 3080 tops the charts in most cases it's

00:04:56.320 --> 00:05:03.520
not all powerful and AMD's big navi announcement next month could extend

00:05:01.520 --> 00:05:07.919
some of those leads now part of the rationale for choosing a

00:05:05.759 --> 00:05:13.360
ryzen-based GPU test bench was so that we could test pci express gen 3 versus

00:05:10.720 --> 00:05:17.840
gen 4 starting with our review of the rx 5700 xt

00:05:15.199 --> 00:05:22.240
unfortunately we had some issues to exercise so that'll have to

00:05:20.240 --> 00:05:27.039
wait for another day get subscribed so you don't miss our deeper dive on pci

00:05:24.720 --> 00:05:30.560
express gen 4. for now let's look at power draw

00:05:28.160 --> 00:05:34.720
using NVIDIA's pcat tool we captured this data while running spec view perf

00:05:32.400 --> 00:05:40.240
on our NVIDIA cards here and we can see that the rtx 3080 is sitting roughly 100

00:05:37.600 --> 00:05:44.120
watts higher than the rtx 2080 ti impressively hitting a lofty maximum

00:05:43.120 --> 00:05:49.199
above 350 watts

00:05:46.560 --> 00:05:54.320
that is a lot of juice ladies and gentlemen and if the trend continues it

00:05:51.919 --> 00:05:58.880
definitely explains why NVIDIA felt they needed this new 12-pin connector this

00:05:56.720 --> 00:06:02.080
power draw translates into a significant increase in heat output as you might

00:06:00.720 --> 00:06:07.840
expect as well but it turns out the new founders edition

00:06:05.120 --> 00:06:11.840
cooler design is nothing short of incredible

00:06:09.120 --> 00:06:16.720
in spite of it drawing a hundred watts more our 3080 maintained lower

00:06:14.560 --> 00:06:20.639
temperatures in a closed chassis than our 2080 ti

00:06:18.319 --> 00:06:23.919
and boosted to roughly two gigahertz under full load throughout our testing i

00:06:23.120 --> 00:06:29.199
mean hey it's unconventional you know blowing

00:06:27.360 --> 00:06:34.400
air up here and sucking air up through there but gosh darn it it works at least

00:06:32.319 --> 00:06:39.360
for the GPU itself one small detail NVIDIA glossed over in

00:06:36.560 --> 00:06:45.840
their marketing is that all of this heat has to go somewhere and when we dug a

00:06:42.160 --> 00:06:48.720
little deeper well by jove we found it

00:06:45.840 --> 00:06:54.080
as it turns out dumping heat directly into your CPU and RAM makes them run

00:06:51.919 --> 00:06:58.080
hotter we even made this little graph where you can see system thermals

00:06:55.600 --> 00:07:02.000
climbing by 10 degrees in a gentle arc across the board indicating that it's

00:07:00.160 --> 00:07:06.880
not just spec view perf putting a load on these components or we'd be seeing a

00:07:04.080 --> 00:07:10.160
spiky graph similar to our CPU and GPU temperatures

00:07:08.400 --> 00:07:15.440
there is good news though when we ran the same test with a founder's edition

00:07:12.240 --> 00:07:17.440
rtx 2080 ti we actually got even worse

00:07:15.440 --> 00:07:20.880
results indicating that the more efficient

00:07:18.720 --> 00:07:25.599
system airflow that NVIDIA boasted about for their cooler actually manages to

00:07:23.199 --> 00:07:28.479
extend to the rest of the system as well so

00:07:26.639 --> 00:07:33.599
good job guys and there's more to the story here than

00:07:30.720 --> 00:07:37.520
performance and power alone NVIDIA has become

00:07:34.560 --> 00:07:42.000
laser focused on system latency as a way to market their products even going as

00:07:39.520 --> 00:07:47.440
far as to seed press with the internal tools that they use to measure it and as

00:07:45.120 --> 00:07:53.280
it turns out that's for good reason now more frames per second does result in

00:07:50.160 --> 00:07:56.720
smoother animations but since most

00:07:53.280 --> 00:07:59.520
gamers would probably say 100 FPS looks

00:07:56.720 --> 00:08:04.000
butter smooth it's clear that this smooth animations benefit has already

00:08:01.680 --> 00:08:08.720
hit a wall of diminishing returns what doesn't seem to have a limit yet

00:08:06.240 --> 00:08:14.160
however is the effect of higher frame rates on a gamer's actual performance

00:08:11.599 --> 00:08:19.680
last year we demonstrated a benefit in competitive play all the way up to 240

00:08:17.199 --> 00:08:24.479
hertz in this video featuring shroud mr grimms corey and paul's hardware so then

00:08:22.639 --> 00:08:29.520
if the reason for this improvement in gaming acumen isn't

00:08:26.800 --> 00:08:34.000
smoother animations well what is it it's the responsiveness of the system

00:08:31.919 --> 00:08:37.680
the more often the image updates on screen the less delay a gamer is going

00:08:36.080 --> 00:08:42.479
to feel between the movement of their mouse and the corresponding crosshair

00:08:39.839 --> 00:08:46.880
making it easier to track down opponents well NVIDIA reflex which is launching

00:08:44.720 --> 00:08:51.120
today as well takes this research NVIDIA's done and goes a step further

00:08:49.360 --> 00:08:55.920
than even the ultra low latency mode that's been enabled in both NVIDIA and

00:08:53.040 --> 00:09:00.880
AMD's drivers since last year and NVIDIA has worked with developers to improve

00:08:58.160 --> 00:09:06.240
rendering latency directly in the game itself with a CPU-bound scenario it's

00:09:03.760 --> 00:09:10.320
not exactly a big deal you're getting a very slight improvement in response time

00:09:08.240 --> 00:09:14.480
with a bit more consistency we're not sure what that plus boost option is all

00:09:12.240 --> 00:09:18.959
about but in fortnite anyway it doesn't seem to do much at least at high frame

00:09:16.399 --> 00:09:23.760
rates other games implementations may be better NVIDIA says that reflex works

00:09:21.519 --> 00:09:29.920
best in GPU bound loads anyway though and in that scenario latency goes from

00:09:26.080 --> 00:09:33.519
90 to 120 milliseconds to around 45 to

00:09:29.920 --> 00:09:35.360
80 with reflex enabled that's huge one

00:09:33.519 --> 00:09:40.080
of the biggest problems with a low frame rate is low responsiveness so the small

00:09:37.839 --> 00:09:44.080
sacrifice of one to two FPS ends up being a really good trade the best part

00:09:42.320 --> 00:09:50.720
is that this feature works all the way back to 900 series maxwell gpus

00:09:46.800 --> 00:09:52.399
great stuff also revolutionary is rtxio

00:09:50.720 --> 00:09:57.440
which uses the same direct storage technology that microsoft is integrating

00:09:54.320 --> 00:09:59.920
into the xbox series x and series x to

00:09:57.440 --> 00:10:05.839
allow the GPU to stream and decompress data directly from an NVMe SSD using the

00:10:03.040 --> 00:10:10.000
cuda cores from an rtx 2000 series card or newer now we can't test this feature

00:10:08.399 --> 00:10:14.560
yet because there are no games on the market that support it yet but we would

00:10:12.080 --> 00:10:18.160
expect gamers to see similar benefits to what sony showed off with their

00:10:16.000 --> 00:10:22.880
analogous technology with improvements to everything from texture resolution in

00:10:20.480 --> 00:10:26.800
game to even level design eliminating things like unnecessary elevator rides

00:10:25.040 --> 00:10:31.440
while the assets load in but wait there's more

00:10:28.720 --> 00:10:37.200
what we also can't test is the reports that all ampere-based cards support

00:10:34.240 --> 00:10:43.440
sriov at a hardware level for the uninitiated sriov allows a single GPU to

00:10:41.040 --> 00:10:48.720
be shared between multiple virtual machines never mind two gamers one GPU

00:10:46.480 --> 00:10:53.600
with performance like this you could run a whole freaking lan center off a single

00:10:51.040 --> 00:10:57.760
threadripper 3990x and like four of these i mean this would be a

00:10:55.680 --> 00:11:01.120
monumental shift from NVIDIA considering that they've locked down g-force and

00:10:59.360 --> 00:11:04.399
virtual machines since basically forever or at least it would

00:11:03.600 --> 00:11:10.720
be NVIDIA clarified with us that the reports of their change of heart were

00:11:08.160 --> 00:11:15.360
sadly overstated and as it turns out just because they can support the

00:11:12.640 --> 00:11:20.320
feature doesn't mean that it's enabled and sure enough when we fired up our

00:11:17.279 --> 00:11:23.040
3080 and Linux it didn't report any sri

00:11:20.320 --> 00:11:27.040
ov capabilities NVIDIA does say though that the final decision on whether to

00:11:24.880 --> 00:11:30.720
enable it or not has yet to be made so it is still possible they'll enable it

00:11:29.200 --> 00:11:35.600
maybe if enough people tweeted them jensen will get the hint wink wink not

00:11:33.200 --> 00:11:39.839
just please in summary then just like last time we have no way of

00:11:37.760 --> 00:11:43.920
testing some of the headline features and no time to test others like NVIDIA

00:11:42.480 --> 00:11:48.480
broadcast which we'll be sure to do a follow-up video on but unlike last time

00:11:46.399 --> 00:11:52.240
the raw performance and pricing are compelling enough story on their own

00:11:50.480 --> 00:11:56.560
even if NVIDIA overplayed their hand a little

00:11:53.519 --> 00:11:57.279
and unlike last time the big feature you

00:11:56.560 --> 00:12:02.560
know the one that's right in the name the hardware support for real-time ray

00:12:00.560 --> 00:12:06.320
tracing has had a couple of years to mature with a much longer list of

00:12:04.399 --> 00:12:10.800
supported games to go along with graphics cards that are now actually

00:12:08.800 --> 00:12:15.600
powerful enough to turn the feature on without tanking the frame rate to be

00:12:13.040 --> 00:12:20.959
clear it's still not the kind of thing that i would personally buy a new card

00:12:17.519 --> 00:12:23.040
over but as far as added bonuses go

00:12:20.959 --> 00:12:26.959
raytraced minecraft is pretty sweet

00:12:25.200 --> 00:12:31.120
and if you're looking for a segue to our sponsor this is it

00:12:28.959 --> 00:12:35.200
do you think making a website is hard well it doesn't have to be use

00:12:33.120 --> 00:12:39.200
squarespace and you'll have your website up and running in a matter of hours

00:12:37.279 --> 00:12:42.160
maybe even faster they've got award-winning templates that'll help

00:12:40.560 --> 00:12:45.680
make your website stand out instead of looking like it's from the 90s and if

00:12:44.000 --> 00:12:48.880
you're interested in how your website's performing they've got built-in tools to

00:12:47.600 --> 00:12:53.120
help you find out what you're doing right and what you're doing wrong both

00:12:50.959 --> 00:12:56.480
our Linus media group and ltx expo websites were built quickly using

00:12:54.560 --> 00:13:00.560
squarespace and if you get stuck making your site they have a 247 support team

00:12:58.639 --> 00:13:05.600
that's ready to help you out just go to squarespace.com LTT and you can get 10

00:13:04.000 --> 00:13:08.720
off today we're going to have that linked below

00:13:07.040 --> 00:13:13.600
so thanks for watching guys if you're looking for more ampere go check out our

00:13:11.360 --> 00:13:17.760
post announcement explainer on ampere for a little bit more background info

00:13:15.360 --> 00:13:20.760
and a little more context for all the hype

00:13:23.120 --> 00:13:30.959
hey i didn't mention lttstore.com go to lttstore.com buy stuff

00:13:29.200 --> 00:13:35.839
we got water bottles and graphic card shirt

00:13:33.200 --> 00:13:35.839
oh yeah that too