WEBVTT

00:00:00.160 --> 00:00:05.480
don't buy that even at msrp that's a

00:00:03.520 --> 00:00:11.840
waste of money let alone during the skullpocalypse but rather than pitch you

00:00:08.080 --> 00:00:14.559
the equally unobtainium radeonrx 6900xt

00:00:11.840 --> 00:00:21.039
i've got something new gamers NVIDIA launched two important things today the

00:00:16.960 --> 00:00:23.359
rtx 3080ti a GPU as powerful as their

00:00:21.039 --> 00:00:25.760
previous king but at a significantly lower price

00:00:24.800 --> 00:00:31.519
and a refreshing change in attitude

00:00:28.640 --> 00:00:36.320
let's talk about both of them and our sponsor aorus the z590 aorus

00:00:34.880 --> 00:00:41.200
master motherboard is compatible with Intel's 11th and 10th gen processors has

00:00:38.879 --> 00:00:46.480
a wide array of i o and built-in Intel wi-fi 6e get yours today at the link in

00:00:43.520 --> 00:00:46.480
the video description

00:00:53.680 --> 00:01:01.920
let's start with the elephant in the room NVIDIA would love for us to compare

00:00:58.079 --> 00:01:04.960
the rtx 3080 ti directly to the rtx 2080

00:01:01.920 --> 00:01:07.200
ti given that it is like the replacement

00:01:04.960 --> 00:01:11.360
but given how few people bought those i doubt it's a burning question for many

00:01:08.880 --> 00:01:16.000
of you besides the more interesting reference point is the rtx 3090 because

00:01:14.479 --> 00:01:20.960
it is so freaking similar

00:01:18.159 --> 00:01:26.560
by comparison the 3080 ti is missing a mere two and a half percent of the rtx

00:01:24.080 --> 00:01:32.560
3090's core count and memory bandwidth and runs at only a two percent slower

00:01:29.040 --> 00:01:35.840
boost clock same gddr6 RAM same total

00:01:32.560 --> 00:01:40.240
graphics power same n bank encoder same

00:01:35.840 --> 00:01:41.439
absolute unit of a GPU except for the

00:01:40.240 --> 00:01:46.240
main difference apart from the smaller cooler on the

00:01:43.439 --> 00:01:50.880
founder's edition cart it only has half of the memory at 12 gigs

00:01:48.960 --> 00:01:56.479
now that's less than AMD's competing 6900 xt which has 16 and it could be a

00:01:54.720 --> 00:02:02.880
problem if you're one of the eight people trying to game at 8k

00:01:59.920 --> 00:02:08.080
but considering how few loads benefit from 24 gigs of RAM it's actually more

00:02:05.520 --> 00:02:12.239
likely to end up being a major win for the 3080 ti since it should help bring

00:02:10.319 --> 00:02:17.440
pricing down more in line with team red's Radeon rx 6900 xt so without

00:02:15.360 --> 00:02:21.440
further ado here's our test bench we've got five gpus for your consideration

00:02:19.360 --> 00:02:25.120
today and i am i'm expecting a lot and my expectations

00:02:24.080 --> 00:02:32.080
are met metro exodus enhanced edition starts things off extremely strong for the 3080

00:02:29.360 --> 00:02:37.280
ti and it manages to keep within the on paper expected 2

00:02:34.400 --> 00:02:42.239
of the rtx 3090 in every parameter but the minimum frame rate and guess what

00:02:40.000 --> 00:02:47.599
that's the very worst news we have for you NVIDIA's new king i guess i have to

00:02:45.280 --> 00:02:54.160
call it that is actually straight up faster than the rtx 3090 in both f1 2020

00:02:51.360 --> 00:02:59.680
and forza horizon 4. perhaps as a direct result of it having less RAM that means

00:02:57.440 --> 00:03:05.360
more power for the GPU itself allowing it to opportunistically turbo more often

00:03:02.800 --> 00:03:10.879
and this same same but different trend continues with assassin's creed valhalla

00:03:07.519 --> 00:03:13.519
and cs go where the 3080ti proves itself

00:03:10.879 --> 00:03:18.480
to be a ridiculously capable gaming card which i mean of course it is it's

00:03:15.920 --> 00:03:22.720
basically an rtx 3090 except for the cut down RAM so what does

00:03:21.120 --> 00:03:26.959
the smaller frame buffer mean for productivity

00:03:24.239 --> 00:03:33.120
well for the most part not much and the rtx 3080 ti continues to show off its 30

00:03:30.480 --> 00:03:37.760
90 like prowess with nearly identical render times in blender in both the cuda

00:03:35.200 --> 00:03:42.720
and optics renderers and the same thing in v-ray where we are still within that

00:03:40.080 --> 00:03:48.080
two percent range in fact even creative cloud gives yet more favorable results

00:03:45.440 --> 00:03:53.040
with one percent being the most common delta between these cards so in these

00:03:50.720 --> 00:03:58.720
use cases unless your large projects require more RAM you're basically giving

00:03:56.000 --> 00:04:03.920
up nothing in davinci resolve however that veneer does begin to crack somewhat

00:04:01.760 --> 00:04:08.400
and we see a larger six percent difference that it's difficult to pin

00:04:06.080 --> 00:04:12.000
down exactly it may be that resolve either wants more memory capacity or it

00:04:10.799 --> 00:04:16.639
could be that it's more bandwidth sensitive but unfortunately accurately

00:04:14.720 --> 00:04:20.639
measuring resolves vram usage isn't really possible at the moment so we're

00:04:19.040 --> 00:04:24.160
not sure what we are sure about is our water

00:04:22.320 --> 00:04:28.560
bottles from ltdstore.com you can measure those 20 or 40 ounces for the

00:04:26.560 --> 00:04:33.040
same price whatever the reason for this result octane bench also sees a larger

00:04:31.199 --> 00:04:35.840
than normal performance delta and spec vue perf continues the trend with

00:04:34.560 --> 00:04:40.320
anywhere from 2 which we expect all the way to 21 in the

00:04:38.639 --> 00:04:46.400
case of the memory intensive energy benchmark which i guess we also expect i

00:04:43.199 --> 00:04:49.120
mean the rtx 3090's claim to fame is its

00:04:46.400 --> 00:04:55.120
massive high bandwidth frame buffer so what the real story is then is just how

00:04:51.360 --> 00:04:58.000
competitive the 3080 ti is without it

00:04:55.120 --> 00:05:02.720
as for thermals NVIDIA ended up reusing the same

00:04:58.960 --> 00:05:04.880
much smaller cooler as the rtx 3080 so

00:05:02.720 --> 00:05:10.400
we weren't really sure what to expect going into this thing i mean there's no

00:05:07.160 --> 00:05:11.759
gddr6x on the back of the pcb now so at

00:05:10.400 --> 00:05:16.400
least we don't have to worry about a repeat of the memory overheating

00:05:13.759 --> 00:05:21.039
problems we saw with the 3090 but we are still looking at roughly the same core

00:05:18.639 --> 00:05:25.600
count and clock speed and power consumption rating

00:05:22.479 --> 00:05:27.440
and yet thermals look spot on never

00:05:25.600 --> 00:05:33.360
breaking past 80 degrees throughout spec view perf the rtx 3090 of course did get

00:05:30.720 --> 00:05:38.960
better thermal results but that card is a triple slot beast and what's most

00:05:36.639 --> 00:05:44.800
impressive is that the 3080ti managed all of this at 11 watts less power draw

00:05:43.120 --> 00:05:49.199
and it didn't even throttle like core clocks were a solid 1.9 gigahertz

00:05:47.440 --> 00:05:53.280
throughout our heavy loaded testing which is just a hundred megahertz lower

00:05:50.960 --> 00:05:55.520
than the rtx 3090 all in all

00:05:54.479 --> 00:06:00.320
it's i don't know if it's a surprising result because we didn't really know what to

00:05:58.160 --> 00:06:06.479
expect but it definitely gives us some insight at least into just how power

00:06:02.800 --> 00:06:09.440
hungry and hot GDDR 6x can be

00:06:06.479 --> 00:06:13.919
which incidentally is now rated for 105 degrees on microns website i guess they

00:06:11.520 --> 00:06:18.960
got tired of all the inquiries about rtx 3090s overheating at 100 degrees of

00:06:16.800 --> 00:06:23.600
course i promise to talk about more than just a new GPU today so let's change

00:06:21.600 --> 00:06:26.800
gears and talk about the new straight talking approach that NVIDIA has taken

00:06:25.280 --> 00:06:32.080
around this launch given that this basically is a 3090 a

00:06:30.080 --> 00:06:36.720
card that they already can't make enough of to meet demand

00:06:34.160 --> 00:06:42.400
we asked NVIDIA what possessed you to release this GPU at a lower price

00:06:39.759 --> 00:06:47.440
and then instead of blathering on about loving gamers so huggy muggy much they

00:06:45.199 --> 00:06:52.319
just confirmed what we already suspected not only is a 24 gig frame buffer

00:06:49.759 --> 00:06:56.080
ludicrous for most gamers today it's also impacting NVIDIA's ability to make

00:06:54.720 --> 00:07:02.160
enough cards surely you didn't think micron was spared from the silicon shortage did you

00:06:59.840 --> 00:07:08.319
so with half the amount of memory compared to the rtx 3090 twice as many

00:07:05.120 --> 00:07:09.919
rtx 30 ti cards can be produced that is

00:07:08.319 --> 00:07:14.720
assuming there are enough gpus to go around not only that but part of the reason

00:07:12.560 --> 00:07:19.280
that NVIDIA was able to use a smaller cooler in the first place is that there

00:07:16.720 --> 00:07:23.280
isn't any memory on the rear of the pcb like the 3090 which also factors into

00:07:21.840 --> 00:07:26.560
its lower cost speaking of lower cost get subscribed

00:07:25.120 --> 00:07:31.039
because we're going to be revisiting our cheap versus expensive vr setups and

00:07:28.800 --> 00:07:34.400
trust me it's getting a lot less obvious than you might think

00:07:32.639 --> 00:07:38.639
NVIDIA was also refreshingly straightforward talking about supply of

00:07:36.160 --> 00:07:42.400
these cards which they admit is not going to meet demand this year as well

00:07:40.800 --> 00:07:48.880
as the difficulties that they've faced making their products available to gamers for msrp

00:07:45.520 --> 00:07:51.360
the target was 9.99 and the idea was to

00:07:48.880 --> 00:07:55.680
embarrass AMD with superior performance and features at the same price the

00:07:54.000 --> 00:07:59.440
features by the way they're not new but it is worth reminding you that NVIDIA's

00:07:57.440 --> 00:08:04.000
hardware video and code functionality and deep learning features like rtx

00:08:01.919 --> 00:08:08.960
voice and deep learning super sampling are very compelling selling points

00:08:06.960 --> 00:08:11.680
but due to the ongoing scarcity of components

00:08:10.160 --> 00:08:16.400
tariffs that are being thrown around willy-nilly these days not to mention

00:08:13.840 --> 00:08:20.879
predatory reselling practices even by some retailers the msrp is going to be

00:08:19.199 --> 00:08:24.800
higher than they would have liked now we're filming this before hearing the

00:08:22.479 --> 00:08:29.840
actual price so i'm gonna do a few different reactions and we'll put it on

00:08:26.639 --> 00:08:29.840
screen with the matching one

00:08:30.319 --> 00:08:36.320
so whatever it said down there you can expect to spend more than that unless

00:08:34.000 --> 00:08:39.599
you end up extremely lucky or you participate in the verified actual gamer

00:08:38.320 --> 00:08:44.080
program finally NVIDIA rather humbly

00:08:41.839 --> 00:08:47.440
acknowledged that you know what we weren't entirely wrong about their

00:08:45.680 --> 00:08:52.880
mining hashrate limiter i know i was shocked they also admitted

00:08:50.320 --> 00:08:56.240
that yeah it'll probably be hacked at some point so

00:08:54.399 --> 00:08:59.519
their new story then is that the main purpose of the limiter is actually

00:08:58.080 --> 00:09:02.480
pretty similar to what we're doing with the verified actual gamer program to

00:09:01.519 --> 00:09:06.880
just slow down the acquisition of these cards

00:09:04.720 --> 00:09:11.200
on mass by cryptominers see NVIDIA that wasn't so hard

00:09:09.839 --> 00:09:15.519
you didn't have to come out and say it was unhackable and split your product

00:09:13.839 --> 00:09:20.720
stack like that and pretend it's some kind of gift to gamers and you know what

00:09:18.720 --> 00:09:24.240
the truth is i still don't agree with the idea that consumers shouldn't be

00:09:22.240 --> 00:09:27.440
allowed to run whatever they want with their own hardware to the best of its

00:09:26.160 --> 00:09:32.720
capabilities but if this admission is NVIDIA

00:09:30.640 --> 00:09:38.080
foreshadowing a willingness to remove the limiter either officially or

00:09:35.360 --> 00:09:42.240
unofficially through a leak once supply catches up

00:09:39.360 --> 00:09:46.320
i'd be a lot more receptive to the idea there's no word yet on whether they

00:09:44.000 --> 00:09:50.320
would do something like that but i think consumers should demand it because it's

00:09:48.160 --> 00:09:55.200
a nice way to subsidize your purchase and NVIDIA has clearly just demonstrated

00:09:52.720 --> 00:10:00.240
a refreshing willingness to listen but let's bring it back to the card should

00:09:57.120 --> 00:10:02.880
you spend your money on an rtx 3080ti

00:10:00.240 --> 00:10:08.880
well if you can get one at msrp it's a no-brainer and if you're comparing it to

00:10:05.279 --> 00:10:11.120
an rtx 3090 it's cheaper is slightly

00:10:08.880 --> 00:10:15.120
less power hungry it has less overall heat output it's smaller in the case of

00:10:13.519 --> 00:10:20.079
the founders edition so compatibility will be better and presumably it will be

00:10:17.279 --> 00:10:23.120
easier to find in stock meaning it's a winner

00:10:21.440 --> 00:10:27.279
and while scalpers are likely to continue to price these things out of

00:10:24.480 --> 00:10:32.079
reach of many gamers the skelpocalypse isn't going to last forever and when the

00:10:29.040 --> 00:10:33.920
dust finally settles the rtx 3080 ti is

00:10:32.079 --> 00:10:38.880
gonna stand at the top of the stack for NVIDIA in competition with the rx 6900xt

00:10:37.680 --> 00:10:44.000
and it's gonna look pretty good doing this

00:10:42.640 --> 00:10:48.399
you know what's not gonna look pretty good our sponsor ting mobile because

00:10:46.720 --> 00:10:52.560
they're all about how good your reception is and how good you sound oh

00:10:50.959 --> 00:10:57.600
wait but also there's video calling so that's a bad segue um don't worry about

00:10:55.440 --> 00:11:01.200
it ting mobile has new rates that make it easier to see how much you can save

00:10:59.360 --> 00:11:05.440
by switching they've got unlimited talk and text for just 10

00:11:03.040 --> 00:11:10.640
data plans starting at 15 a month and their new set 12 plan with 12 gigs of

00:11:08.079 --> 00:11:14.800
data for 35 bucks a month they've even got unlimited data for 45 a month so if

00:11:13.200 --> 00:11:18.720
you're looking for a good deal ting mobile is offering their unlimited plan

00:11:16.480 --> 00:11:22.640
for just 25 a month for the first three months and if monthly plans aren't your

00:11:20.640 --> 00:11:26.640
thing guess what their previous pay for only what you use plans are still there

00:11:24.800 --> 00:11:30.240
they're called ting mobile's flex plans now and they charge just five dollars a

00:11:28.480 --> 00:11:33.839
gig one of the coolest things about flex plans is that if you have a family and

00:11:32.160 --> 00:11:38.160
you have a family plan you can connect more phones and share it and save more

00:11:36.640 --> 00:11:41.440
you'll get the same nationwide coverage in the us an award-winning customer

00:11:39.680 --> 00:11:45.519
service and pretty much any phone will work with ting so don't wait check them

00:11:43.360 --> 00:11:50.240
out at Linus.ting.com and you'll get a 25 credit so thanks for watching guys if

00:11:48.640 --> 00:11:54.160
you enjoyed this video maybe go check out the build we did

00:11:51.839 --> 00:11:59.200
there's a 10 000 machine for playing roblox of all things and it had a 30-90

00:11:57.120 --> 00:12:02.320
in it and boy could it ever have been a lot cheaper with one of these

00:12:02.959 --> 00:12:08.800
sorry guys honestly we didn't know this was coming at least not for sure