WEBVTT

00:00:00.640 --> 00:00:07.120
this is real footage captured outside a micro center computer store in dallas

00:00:05.040 --> 00:00:11.920
texas and you might think that twitch streamer preston a lewis was filming a

00:00:09.120 --> 00:00:15.360
black friday shopping rush or something but no

00:00:13.280 --> 00:00:18.080
no one in this crowd of people was hoping for a discount

00:00:17.039 --> 00:00:22.400
at all the shortage of gpus has gotten so dire

00:00:21.119 --> 00:00:27.519
that this is how desperate people are just to buy

00:00:24.960 --> 00:00:33.040
a graphics card at an msrp that has already been twice inflated since the

00:00:29.840 --> 00:00:35.440
release of NVIDIA's 30 series once by

00:00:33.040 --> 00:00:41.520
component shortages then again by tariffs on u.s imports from china which

00:00:38.320 --> 00:00:42.559
makes NVIDIA's pricing for the new 3070

00:00:41.520 --> 00:00:48.879
ti really confusing because 599 for this

00:00:46.960 --> 00:00:53.199
much raw power that would have been a decent deal even

00:00:50.719 --> 00:00:59.280
before the shortage and what makes it especially unexpected is that last

00:00:55.600 --> 00:01:00.399
week's 3080 ti was also competitively

00:00:59.280 --> 00:01:06.479
priced but only in the context of the current

00:01:02.640 --> 00:01:09.600
landscape where this untested 2080 ti on

00:01:06.479 --> 00:01:12.960
ebay sold for a thousand us dollars is

00:01:09.600 --> 00:01:15.119
NVIDIA trying to start street brawls

00:01:12.960 --> 00:01:19.040
or did they travel to an alternate timeline and bring this thing back for

00:01:16.960 --> 00:01:23.280
the good of all gamers and if they did

00:01:20.799 --> 00:01:27.600
just how good is it and who's our sponsor

00:01:25.200 --> 00:01:31.600
antlion audio their modmic wireless microphone delivers best in class audio

00:01:29.439 --> 00:01:35.920
quality 12 plus hours of battery life and it magnetically mounts to almost any

00:01:33.439 --> 00:01:41.720
pair of headphones get 15 off the mod mic wireless another mod mic product at

00:01:37.759 --> 00:01:41.720
the link in the video description

00:01:48.560 --> 00:01:55.600
599 puts the 3070 ti smack dab in the

00:01:51.920 --> 00:01:57.439
middle of the rtx 3070 and rtx 3080 both

00:01:55.600 --> 00:02:01.840
in terms of price and product positioning which

00:01:58.880 --> 00:02:09.520
in itself makes sense because unlike the rtx 3080 ti which was pretty much a

00:02:04.799 --> 00:02:13.120
watered-down rtx 3090 the rtx 3070 ti is

00:02:09.520 --> 00:02:16.400
more of a beefed up rtx 30 70. it shares

00:02:13.120 --> 00:02:18.640
the same ga104 GPU with about four

00:02:16.400 --> 00:02:23.120
percent more course with the same eight gigs of gddr6 memory on the same 256-bit

00:02:22.239 --> 00:02:29.040
bus pretty unexciting or i'm sorry did i say the same gddr6

00:02:27.360 --> 00:02:35.200
memory what i should have probably mentioned is that it is much faster

00:02:32.239 --> 00:02:38.599
clocking in at a good 2.5 gigahertz higher we are looking at an effective

00:02:37.040 --> 00:02:43.599
bandwidth boost of 160 gigabytes per second which is nearly

00:02:41.920 --> 00:02:49.840
a 24 increase consequently the rated graphics power is

00:02:47.040 --> 00:02:53.920
a hefty 70 watts higher and the cooler is a new ambiguous design that looks

00:02:52.080 --> 00:03:00.560
something like a cross between the rtx 3080 and rtx 3070 with the same totally

00:02:58.080 --> 00:03:04.800
necessary and very industry standard 12-pin connector though at least two

00:03:02.879 --> 00:03:09.360
NVIDIA's credit they were kind enough to angle it instead of having it stick

00:03:06.560 --> 00:03:14.239
straight out like the old 30 70. so the hope then both for NVIDIA and for

00:03:11.760 --> 00:03:20.560
the rest of us is that if the cooler is in between the two cards the rtx 37 etis

00:03:17.440 --> 00:03:23.040
performance will be as well despite it

00:03:20.560 --> 00:03:29.519
having just four percent more course than its namesake and a whopping 29

00:03:26.000 --> 00:03:31.440
fewer cores than the rtx 3080.

00:03:29.519 --> 00:03:35.680
now since the testing is happening so soon after our rtx 3080ti review we're

00:03:33.840 --> 00:03:39.599
going to be using the same bench in the same patch state and we'll even be

00:03:37.680 --> 00:03:44.480
wearing the same shirt and the same underwear from lttstore.com

00:03:42.239 --> 00:03:48.000
freshly washed of course though starting with metro exodus enhanced

00:03:46.080 --> 00:03:53.519
edition the story is already looking pretty interesting for the 30-70 ti it

00:03:50.799 --> 00:03:58.959
managed an average frame rate 12 higher than the 30-70 which is good

00:03:56.080 --> 00:04:03.519
but then that also puts it a good 19 lower than the rtx 3080 which is

00:04:02.239 --> 00:04:08.239
hmm f1 2020 tightens the spread somewhat

00:04:05.680 --> 00:04:14.000
though with a 10 lead over the 30 70 and a 12 loss against the 3080 and forza

00:04:11.760 --> 00:04:18.959
horizon 4 actually continues along this trajectory giving us the best result yet

00:04:16.720 --> 00:04:24.320
so it appears that high frame rates favor the 37 eti's additional memory

00:04:21.199 --> 00:04:26.560
bandwidth while lower frame rates

00:04:24.320 --> 00:04:30.479
really want for more cores and this appears to bear out an assassin's creed

00:04:28.320 --> 00:04:38.320
valhalla where frame rates dip way back down to earth and actually AMD's rx 6700

00:04:34.560 --> 00:04:42.720
xt manages to eclipse not only the 3070

00:04:38.320 --> 00:04:44.720
and 370ti but even the 30 80. cs go is

00:04:42.720 --> 00:04:49.040
another chance for the 3070 ti's lightning fast memory to make up ground

00:04:46.960 --> 00:04:53.840
but while it manages to get close to the middle of the 30 70 and 3080 it ends up

00:04:51.520 --> 00:04:57.680
closer to the former not gonna lie though given that we're

00:04:55.680 --> 00:05:01.600
looking at a handful of extra cores and a hefty memory overclock

00:04:59.759 --> 00:05:04.960
i'm pretty impressed but how much are those going to help if

00:05:03.520 --> 00:05:08.400
you want to do anything other than gaming as it turns out

00:05:07.360 --> 00:05:12.880
not much the rtx 3070 ti doesn't manage much in

00:05:11.039 --> 00:05:19.280
the way of an improvement over the rtx 3070 and loses badly to the rtx 3080 in

00:05:16.720 --> 00:05:23.039
rendering with both blender and v-ray while creative cloud doesn't really show

00:05:21.440 --> 00:05:27.440
enough variation between the cards to warrant picking up one over the other

00:05:25.759 --> 00:05:32.000
octane bench is a little better than blender but we're still looking at a

00:05:29.759 --> 00:05:36.960
much closer result to the 3070 than to the 3080 and this trend continues pretty

00:05:35.120 --> 00:05:42.800
hardcore into spec view perf where the rtx 3080 stomps all over both of its ga

00:05:40.639 --> 00:05:46.479
104 equipped competitors the only thing that could be considered

00:05:44.320 --> 00:05:50.080
a win here is the energy benchmark where the 3070 ti's additional memory

00:05:48.240 --> 00:05:52.960
bandwidth pushes it far enough away from the 3070 to find that middle ground

00:05:52.080 --> 00:05:59.440
again so what have we learned then it turns out

00:05:56.240 --> 00:06:01.759
that fast memory on a GPU can be good

00:05:59.440 --> 00:06:06.960
very good but that it's also very situational and we also learned that

00:06:04.160 --> 00:06:12.000
fast memory with more cores is better than just fast memory in every

00:06:09.120 --> 00:06:16.479
conceivable way okay we didn't really learn either of

00:06:14.319 --> 00:06:20.400
those things but it's always nice to get a refresher isn't it

00:06:18.080 --> 00:06:25.039
and on that subject let's talk cooling the bigger heatsink on the rtx 3070 ti

00:06:22.800 --> 00:06:29.440
founders edition meant that it didn't get all that much hotter than the rtx

00:06:26.880 --> 00:06:34.000
3070 and with a peak of just a little higher than 78 degrees and an average of

00:06:31.680 --> 00:06:38.080
64 degrees across our spec view perf test we're looking at a respectable

00:06:35.919 --> 00:06:44.080
result all things considered power consumption however is less so the

00:06:41.039 --> 00:06:46.240
37 eti ends up sitting much closer to

00:06:44.080 --> 00:06:54.639
the 30 80 than its performance would suggest at 311 watts peak and 151 watts

00:06:51.039 --> 00:06:57.360
average that's about 20 more power draw

00:06:54.639 --> 00:07:01.280
than its non-ti cousin and most of this appears to come from the up-clocked

00:06:59.120 --> 00:07:06.160
memory considering how few additional GPU cores there are and the fact that

00:07:03.520 --> 00:07:13.120
when we look at core clocks the 3070 ti consistently runs about 100 megahertz

00:07:08.800 --> 00:07:14.560
lower than both the 3070 and the 3080.

00:07:13.120 --> 00:07:19.919
at the end of the day though our recommendation like with the rtx 3080ti

00:07:17.599 --> 00:07:24.639
is going to be to immediately buy this card at msrp if you can

00:07:22.639 --> 00:07:29.840
and i know that not everyone agreed with our conclusion last time but the truth

00:07:26.560 --> 00:07:32.160
is i stand behind what i said 100

00:07:29.840 --> 00:07:36.400
because no matter who you are are you a filthy scalper looking to make a quick

00:07:34.000 --> 00:07:39.440
buck or are you a gamer who happens to have that much money to spend on a

00:07:37.840 --> 00:07:43.360
graphics card whichever one you are

00:07:41.039 --> 00:07:47.759
both of the new ti variants at msrp of course are going to be a much

00:07:46.160 --> 00:07:52.800
better deal than anything that you're getting on the secondary market and sure

00:07:50.479 --> 00:07:58.319
you can compare these cards to suggested pricing from a year ago but the reality

00:07:55.759 --> 00:08:01.759
of the situation is that outside of the handful of cards that NVIDIA sells

00:08:00.240 --> 00:08:06.560
through their web store those cards at those prices

00:08:04.479 --> 00:08:10.000
they don't actually exist and that might make you mad but shooting

00:08:08.639 --> 00:08:14.879
the messenger isn't going to get you anywhere and besides i did warn you

00:08:12.879 --> 00:08:18.960
after all so yeah

00:08:16.560 --> 00:08:25.599
it's a great GPU with great NVIDIA features like dlss high quality hardware

00:08:21.759 --> 00:08:27.840
video encoding and more and 5.99

00:08:25.599 --> 00:08:32.800
it's an absolute bargain for this class of GPU in the current conditions

00:08:30.479 --> 00:08:37.839
will the 37 eti still make sense when the dust settles

00:08:34.479 --> 00:08:40.399
honestly my money's on not really but i

00:08:37.839 --> 00:08:46.399
suppose only time can tell for sure and by the look of things

00:08:42.320 --> 00:08:48.399
it's going to be a long time

00:08:46.399 --> 00:08:52.160
but it won't be a long time until i tell you about our sponsor drop's alt

00:08:50.480 --> 00:08:56.000
mechanical keyboard is based on their top selling control keyboard and

00:08:54.160 --> 00:08:59.920
features fully programmable keys so you can assign macros or keybinds to

00:08:58.080 --> 00:09:05.040
whatever you want they removed the function row to reduce it to 67 keys

00:09:02.240 --> 00:09:09.200
instead of 87 and it's created with a solid aluminum frame and a hot swap

00:09:07.040 --> 00:09:13.200
switch plate making it super easy to change out your switches it's both

00:09:11.040 --> 00:09:17.279
portable and rock solid and features a customizable RGB backlight and

00:09:15.200 --> 00:09:20.959
undergliding as well as two usbc connectors with one on each side of the

00:09:19.279 --> 00:09:25.040
keyboard whatever works best for your cable management so pick up your own

00:09:22.959 --> 00:09:30.240
drop alt keyboard at the link down below and you'll get a free set of skylight

00:09:27.120 --> 00:09:32.160
keycaps today using promo code free caps

00:09:30.240 --> 00:09:37.120
at the checkout so thanks for watching guys go check out our recent video where

00:09:34.480 --> 00:09:42.080
we got verified actual gamers to compete against yours truly for a chance to get

00:09:38.959 --> 00:09:44.480
a GPU at that elusive msrp actually some

00:09:42.080 --> 00:09:47.760
of them even ended up with discounts they beat me