WEBVTT

00:00:00.080 --> 00:00:06.799
i just heard NVIDIA's got a sick new GPU

00:00:03.760 --> 00:00:06.799
for us to play with

00:00:07.040 --> 00:00:14.000
well what part of that is wrong did we not get it or is there no new GPU

00:00:12.799 --> 00:00:20.400
yes wait oh this is the rtx 2060 relaunch isn't

00:00:18.560 --> 00:00:23.359
it yep but

00:00:21.840 --> 00:00:26.320
but why think if i know you want to talk about

00:00:24.880 --> 00:00:31.760
it yes yes i do

00:00:28.960 --> 00:00:36.480
and i want to talk about our sponsor MSI's new meg z690 UniFi motherboard

00:00:34.800 --> 00:00:40.559
supports the latest 12 gen Intel core cpus up to

00:00:38.040 --> 00:00:46.559
6666 mega transfer per second ddr5 memory and a stealthy black aesthetic

00:00:43.200 --> 00:00:46.559
check it out at the link below

00:00:54.480 --> 00:01:02.160
NVIDIA's new old stock GPU uses the same core layout as the rtx 2060 super but

00:01:00.239 --> 00:01:06.400
instead of that GPU's eight Gigabyte memory configuration they're bumping it

00:01:03.920 --> 00:01:10.479
up to 12 gigabytes of gddr6 at first glance that might sound like a

00:01:08.400 --> 00:01:13.840
decent upgrade but there's more to the story than just the size of the frame

00:01:12.159 --> 00:01:20.080
buffer alone the available memory bandwidth is still being stifled by the

00:01:16.000 --> 00:01:22.400
original rtx 2060's 192-bit bus so in

00:01:20.080 --> 00:01:27.680
that respect it's actually a step back from the rtx 2060 super that means this

00:01:25.360 --> 00:01:32.400
card should perform somewhere between the og 2060 and the super variant rather

00:01:30.400 --> 00:01:36.479
than surpassing them unless you have a specific need for the extra video memory

00:01:34.320 --> 00:01:40.640
and you just might to account for newer titles coming out with bigger and more

00:01:38.159 --> 00:01:45.680
varied assets NVIDIA would have seen fit to give the rtx 2060 some extra juice in

00:01:43.520 --> 00:01:50.399
the caboose not to mention the marketing benefits 12 gigs puts them on par with

00:01:48.240 --> 00:01:54.640
AMD who has for some time been criticizing NVIDIA's lower end cards for

00:01:52.799 --> 00:01:58.560
their smaller frame buffers if nothing else at least it helps to

00:01:56.640 --> 00:02:03.439
ensure that anyone who's buying one of these rtx 2060s isn't stuck with all of

00:02:01.520 --> 00:02:06.079
the limitations of a three-year-old design

00:02:04.640 --> 00:02:09.920
obviously you're not going to be cranking the visuals on these cards but

00:02:08.319 --> 00:02:15.599
you're also not going to be scrambling to upgrade again in a year or so

00:02:12.800 --> 00:02:19.680
maybe we'd love to verify that but we certainly can't today

00:02:17.599 --> 00:02:22.640
not only has NVIDIA canned the founders edition that they briefly listed on

00:02:21.599 --> 00:02:27.120
their website but we didn't get a card to review and they

00:02:25.599 --> 00:02:34.160
haven't been talking to the press either us included as far as we could tell this card was

00:02:30.800 --> 00:02:37.360
still a maybe on the horizon

00:02:34.160 --> 00:02:39.120
rumor as recently as last week there's

00:02:37.360 --> 00:02:44.160
two ways you could look at this first obviously we already know pretty

00:02:41.440 --> 00:02:48.160
much how the rtx 2060 performs so reviews aren't likely to reveal anything

00:02:46.160 --> 00:02:51.120
mind-blowing maybe they felt it wasn't worth bothering

00:02:49.440 --> 00:02:54.080
the second is more troubling

00:02:52.640 --> 00:02:58.400
as of writing i have no idea how much it costs

00:02:56.879 --> 00:03:03.360
each board partner is going to have their own skus with their own pricing

00:03:00.720 --> 00:03:07.280
and NVIDIA hasn't announced a baseline msrp like they usually would instead

00:03:05.599 --> 00:03:11.920
opting for the cryptic it's a premium product so it'll be

00:03:10.159 --> 00:03:16.000
priced accordingly line it could be that this card will be

00:03:14.239 --> 00:03:19.920
very expensive and they're trying to get out in front of a negative news cycle

00:03:17.599 --> 00:03:25.200
full of inevitable comparisons to the 30 series launch msrps and to be fair to

00:03:22.560 --> 00:03:29.920
NVIDIA there's no real benefit to tech press and subsequently consumers yelling

00:03:27.920 --> 00:03:34.239
at them about how their msrp doesn't exist when it's inevitably higher for

00:03:31.840 --> 00:03:39.519
most partner boards now we've talked at length about why msrps have been rising

00:03:36.720 --> 00:03:44.159
and why the msrps that many gpus originally launched at are functionally

00:03:41.840 --> 00:03:48.319
extinct but the tldr is that it's a combination of the ongoing chip shortage

00:03:46.319 --> 00:03:53.120
raw materials skyrocketing in price thanks to coveted recovery and

00:03:50.319 --> 00:03:57.200
speculation and just plain and simple supply and demand

00:03:54.879 --> 00:04:01.680
it's becoming more expensive just to get the chips let alone put it all together

00:03:59.040 --> 00:04:04.879
and this cost isn't going away anytime soon

00:04:02.560 --> 00:04:09.200
NVIDIA sells gpus to everyone else so that they can eat that cost then but

00:04:07.200 --> 00:04:13.200
partners don't have the luxury of selling loss leaders oh this is

00:04:11.280 --> 00:04:17.840
interesting i just got a response from ASUS apparently they don't know what the

00:04:15.439 --> 00:04:23.280
msrp is yet i'd love to say that's surprising but i just don't think it is

00:04:20.799 --> 00:04:27.440
that NVIDIA won't commit to an msrp could be seen as a small mercy

00:04:26.080 --> 00:04:32.479
or another way of messing with their partners whatever the case partners

00:04:30.960 --> 00:04:35.360
won't have to do much to both piss everyone off and sell all their stock at

00:04:34.240 --> 00:04:42.720
once the fact of the matter is that both the rtx 2060s launched at 349 and the rtx

00:04:40.639 --> 00:04:46.560
3060 that's supposed to replaced it launched at 3.29

00:04:44.639 --> 00:04:50.720
if the price ends up being anything higher than that our slash hardware is

00:04:48.400 --> 00:04:54.400
going to lose their minds whether it's justified or not

00:04:52.400 --> 00:04:57.840
it feels like NVIDIA is hanging partners out to dry again while

00:04:56.320 --> 00:05:02.400
they continue to enjoy the positive perception of reasonable msrp on their

00:05:00.240 --> 00:05:06.240
cards that board partners have no hope of matching with this release or any

00:05:04.160 --> 00:05:08.720
other either in terms of price or performance

00:05:07.520 --> 00:05:13.520
we'll have them link down below once they're available but i'm not hopeful

00:05:10.720 --> 00:05:19.120
it'll be a good deal though as unpopular as this take might be it'll still be a

00:05:16.160 --> 00:05:24.400
bargain at least at off-the-shelf prices compared to scalper rtx 2060 prices on

00:05:21.840 --> 00:05:29.759
ebay i mean look at this they are regularly selling for more than 500 to

00:05:26.960 --> 00:05:35.039
600 depending on whether it's the super variant of the card and whoever bought

00:05:31.759 --> 00:05:36.720
this one for over 800 us dollars must

00:05:35.039 --> 00:05:40.080
have been pretty desperate the people buying these scalped

00:05:38.320 --> 00:05:43.759
secondary market cards are the people who will benefit the most they've

00:05:42.000 --> 00:05:47.440
clearly already accepted that they're going to have to shell out flagship

00:05:45.440 --> 00:05:51.680
money for a less than flagship experience and more cards on the market

00:05:50.080 --> 00:05:55.680
means that there's less pressure for these second-hand gpus

00:05:53.759 --> 00:06:00.479
i mean it's really a win-win for consumers as far as i'm concerned just

00:05:57.840 --> 00:06:05.600
like our new swag kit on lttstore.com it's heavier than a sweater and lighter

00:06:01.919 --> 00:06:07.520
than a jacket everyone wins except

00:06:05.600 --> 00:06:11.759
not really if NVIDIA has been holding on to these

00:06:09.919 --> 00:06:16.400
gpus because they couldn't sell them before now then this is more of a

00:06:13.600 --> 00:06:20.560
positive for them than for anyone else and it means that these gpus have a

00:06:18.319 --> 00:06:24.560
finite quantity but i don't think that's true NVIDIA would have sold remaining

00:06:22.639 --> 00:06:29.840
stock for cheap or to a system integrator for pre-built or what have

00:06:26.240 --> 00:06:32.800
you before now remember the 7900 gto

00:06:29.840 --> 00:06:39.120
those were basically 7 900 gtx chips that were perfectly fine but deprecated

00:06:35.600 --> 00:06:40.800
by the outstanding 8800 series besides

00:06:39.120 --> 00:06:45.280
there's no way that they'd have enough surplus to launch a whole new sku at

00:06:43.360 --> 00:06:51.520
this point across their entire range of partners 7900 gto as you might recall

00:06:48.319 --> 00:06:53.280
was MSI and evga exclusively due to its

00:06:51.520 --> 00:06:59.680
limited quantities now turing rtx 20 series was produced on

00:06:56.319 --> 00:07:01.759
tsmc's 12 nanometer node and more likely

00:06:59.680 --> 00:07:06.400
what's happening is NVIDIA is utilizing otherwise idle production because you

00:07:04.080 --> 00:07:10.560
got to remember guys even though tsmc offers seven nanometer and five

00:07:08.560 --> 00:07:15.039
nanometer for cutting edge products and is working on ramping three nanometer

00:07:12.880 --> 00:07:18.880
they haven't necessarily just taken their last gen old equipment and hucked

00:07:17.680 --> 00:07:22.800
it in a bin due to the large upfront investment that

00:07:20.880 --> 00:07:28.880
kind of stuff ends up getting phased out very gradually i mean even AMD's

00:07:25.360 --> 00:07:31.360
flagship x570 chipset is still using 14

00:07:28.880 --> 00:07:35.599
nanometer because it's economical so if no other gpus are being produced

00:07:33.759 --> 00:07:41.840
on 12 nanometer right now any new capacity that NVIDIA gets out of

00:07:38.639 --> 00:07:44.720
those fabs is new capacity as far as the

00:07:41.840 --> 00:07:49.120
GPU market is concerned and i think they made the right call going with the rtx

00:07:46.639 --> 00:07:53.599
2060 super cores as well because assuming that they're going to be using

00:07:50.479 --> 00:07:55.759
the relatively small tu-106 die they

00:07:53.599 --> 00:08:00.160
should be able to get more good dies per wafer than they would with something

00:07:57.039 --> 00:08:02.160
like a 2070 super redux or something

00:08:00.160 --> 00:08:06.560
along those lines okay this is starting to make sense

00:08:04.080 --> 00:08:13.039
except for the fact that NVIDIA is bundling these gpus with 12 gigabytes of

00:08:10.080 --> 00:08:17.680
gddr6 gdgara 6 was a large part of the production bottleneck that played both

00:08:14.639 --> 00:08:19.440
NVIDIA and AMD through the 2020 and 2021

00:08:17.680 --> 00:08:23.199
phases of this shortage and i haven't found any evidence to

00:08:21.440 --> 00:08:28.560
suggest that things are so much better now that a relief skew like this can

00:08:26.000 --> 00:08:34.240
reliably use so much memory 12 gigs at 192 bit means that we're looking at six

00:08:31.120 --> 00:08:36.560
16 gigabit dies the exact same layout of

00:08:34.240 --> 00:08:41.200
the rtx 3060 well this might have been done so partners could use the same or

00:08:38.320 --> 00:08:46.399
similar pcbs it means that this memory is not going into rtx 3060 production or

00:08:43.919 --> 00:08:50.800
rtx 3050 production which is expected sometime in early 2022 get subscribed so

00:08:49.200 --> 00:08:54.320
you don't miss our coverage of that card when it launches

00:08:52.640 --> 00:08:58.160
i can only assume that they have their reasons clearly NVIDIA knows more about

00:08:56.320 --> 00:09:02.800
supply chain management than we do at this point or indeed than they did when

00:09:00.720 --> 00:09:09.120
they launched the 8800 series all those years ago it did seem that around summer

00:09:06.080 --> 00:09:11.760
of this year demand for gddr6 had fallen

00:09:09.120 --> 00:09:16.320
in comparison with supply so who knows and on the subject of lessons

00:09:14.399 --> 00:09:20.480
learned it does say a lot about the mature way that NVIDIA is handling this

00:09:18.399 --> 00:09:26.720
by straight up calling this card what it is 20 60 12 gig remember how NVIDIA

00:09:24.399 --> 00:09:32.480
rebranded fermi and then later kepler into two generations each i mean we got

00:09:29.440 --> 00:09:34.560
new gpus with new names but they were

00:09:32.480 --> 00:09:39.440
the same bloody thing it's stupid that they have to get a gold

00:09:37.360 --> 00:09:43.680
star for calling a spade a spade in this industry but

00:09:41.120 --> 00:09:47.920
here we are and as long as the logistics do work out this is going to be a net

00:09:45.760 --> 00:09:51.839
benefit to the overall availability of graphics cards for

00:09:49.440 --> 00:09:56.080
at least a little while which is good for everyone just as long as we don't go

00:09:54.160 --> 00:10:00.560
back much further than this i had a dream the other night that NVIDIA

00:09:57.920 --> 00:10:05.600
re-released the gtx 780 and everyone was happy about it i mean 28 nanometer is an

00:10:03.440 --> 00:10:08.800
even less utilized node these days right so

00:10:07.279 --> 00:10:14.079
i don't think that dream's coming true anytime soon re-releasing an older GPU

00:10:11.680 --> 00:10:17.920
is only remotely feasible because we are in such

00:10:15.120 --> 00:10:21.680
desperate circumstances and because the process node for that GPU is still in

00:10:20.240 --> 00:10:26.000
production and maybe not fully tapped by other industries it

00:10:24.800 --> 00:10:30.160
certainly carries with it some challenges like the fact that NVIDIA

00:10:28.240 --> 00:10:34.079
might have to continue to support 20 series cards for much longer than they

00:10:32.160 --> 00:10:37.200
might have originally planned in order to maintain their reputation for

00:10:35.600 --> 00:10:41.200
industry-leading long-term driver support this move then requires some

00:10:39.360 --> 00:10:46.399
lateral thinking on the part of both NVIDIA and prospective buyers

00:10:43.920 --> 00:10:51.040
hey it's not going to be for everyone but we live in interesting times and the

00:10:49.040 --> 00:10:56.240
market is clearly there for this class of GPU i just hope that the power

00:10:53.600 --> 00:11:01.040
consumption per hash rate is not appealing enough to make it appealing

00:10:58.560 --> 00:11:05.680
because i doubt they respun it for a light hash rate version

00:11:03.839 --> 00:11:09.839
just like i respond telling you about our sponsor jackery makes solar ready

00:11:08.240 --> 00:11:14.160
portable power stations designed to improve outdoor life by providing power

00:11:11.600 --> 00:11:18.160
outdoors whenever and wherever you need it their explorer 1500 portable power

00:11:16.720 --> 00:11:23.519
station has enough juice to keep all your devices powered and connected it's

00:11:20.160 --> 00:11:25.120
huge 1500 watt hour capacity and 1800

00:11:23.519 --> 00:11:29.519
watt output rate allows up to seven devices plugged in simultaneously it

00:11:27.600 --> 00:11:33.519
takes only four hours to recharge up from zero to eighty percent and you can

00:11:31.600 --> 00:11:37.760
check out the explorer 1500 at the link down below and get ten percent off with

00:11:35.440 --> 00:11:42.720
code Linus tech tips thanks for watching guys go check out our video on how i was

00:11:40.480 --> 00:11:49.360
right that you should have built a pc back in 2019 but for all the wrong

00:11:46.000 --> 00:11:51.519
reasons yep we all lost the game but

00:11:49.360 --> 00:11:57.519
there's more info about the shortage and prices in there that might help to make

00:11:53.040 --> 00:11:57.519
some sense of these troubled times