WEBVTT

00:00:00.240 --> 00:00:06.160
Intel is back it's not quite the way we all expected

00:00:04.240 --> 00:00:11.040
when ryzen first hit the scene back in 2017 but i don't think we could have

00:00:08.720 --> 00:00:16.000
hoped for a better outcome than this not just for the consumer but for the entire

00:00:13.440 --> 00:00:21.199
industry 12th gen core code named alder lake is like nothing we've ever seen on

00:00:18.480 --> 00:00:25.519
the pc to date and i'm getting some serious ryzen

00:00:22.800 --> 00:00:29.119
or maybe bulldozer vibes from the lateral thinking it took to put it all

00:00:27.039 --> 00:00:35.040
together so let's talk about the who the what the why and how this return to

00:00:31.519 --> 00:00:36.880
competition is so good for everybody

00:00:35.040 --> 00:00:41.200
you know what else is good our sponsor origin pc origin pc laptops can be

00:00:39.360 --> 00:00:45.280
customized with Intel core processors and NVIDIA GeForce rtx 30 series

00:00:43.280 --> 00:00:50.719
graphics with max p design backed by a 24 7 support team you can check out

00:00:46.960 --> 00:00:50.719
origin pc at the link below

00:00:57.680 --> 00:01:04.479
Intel's performance claims for alderlake are mostly comparing against their own

00:01:02.160 --> 00:01:08.880
last gen product but make no mistake they are still coming out swinging

00:01:06.320 --> 00:01:14.320
against AMD in gaming they say alder light gets equal or up to 30

00:01:12.000 --> 00:01:19.520
better performance than ryzen 5000 depending on the game and up to 50

00:01:16.960 --> 00:01:23.840
better than 11th gen core processors this is

00:01:21.280 --> 00:01:28.560
huge although there are some concerns about their methodology Intel's

00:01:26.560 --> 00:01:31.600
graphs mention running Windows 11 for all of their testing and that's a

00:01:30.079 --> 00:01:35.840
problem because if you've been keeping up with the news you'll know that

00:01:33.280 --> 00:01:41.360
Windows 11's new scheduler had a little problem with AMD processors that took a

00:01:38.159 --> 00:01:43.280
big toll on performance it's patched now

00:01:41.360 --> 00:01:46.320
but when they tested it definitely wasn't which means that all of these

00:01:44.960 --> 00:01:50.799
numbers need to be taken with a veritable pillar of salt and that's not

00:01:48.880 --> 00:01:55.439
even considering the potential impacts that this new hybrid architecture is

00:01:52.399 --> 00:01:57.439
going to have on older drm or anti-cheat

00:01:55.439 --> 00:02:01.360
technologies that use CPU detection while providers for these technologies

00:01:59.040 --> 00:02:06.159
have already begun to roll out fixes older games and apps may never receive

00:02:03.439 --> 00:02:11.280
any breaking compatibility but what makes it so different that drm and even

00:02:08.640 --> 00:02:14.720
Windows 11 scheduler might break it's more like apple's m1 processors

00:02:13.440 --> 00:02:19.360
with dedicated performance and efficiency cores p and e cores for short

00:02:17.680 --> 00:02:22.959
those allow for tasks to be intelligently delegated between the

00:02:20.879 --> 00:02:26.560
cores that best suit them this has fueled suspicions that Windows 11's

00:02:24.640 --> 00:02:31.360
problematic new scheduler was developed specifically for alder lake and i can

00:02:28.800 --> 00:02:37.200
see why Intel claims that each cluster of four of the e-cores fit in the same

00:02:33.840 --> 00:02:40.000
space as a single 6th gen skylake core

00:02:37.200 --> 00:02:43.760
but each are more capable per watt and i should point out despite their age

00:02:41.440 --> 00:02:47.920
skylight cores are very capable and usable to this day so that's saying a

00:02:45.680 --> 00:02:52.800
lot still that ultimately means that if all the p cores are full sending a task

00:02:50.959 --> 00:02:56.400
to the simplified e cores might actually be faster than using

00:02:54.879 --> 00:03:01.360
brute force hyper threading to split the p cores on that note hyper threading is

00:02:58.720 --> 00:03:04.720
only available on the p cores so we'll have to see for ourselves what that

00:03:02.800 --> 00:03:09.840
means for heavily threaded tasks like rendering or code compilation my guess

00:03:07.200 --> 00:03:13.280
is each ecore cluster will roughly equal one or two additional p cores if not

00:03:11.680 --> 00:03:18.159
more depending on the task but how do you even use these new cores enter

00:03:15.680 --> 00:03:22.159
thread director which detects what kind of work an app is doing and provides

00:03:20.159 --> 00:03:27.040
feedback to the operating system telling it which course those apps are best

00:03:24.720 --> 00:03:31.760
suited for it works kind of like speed shift or preferred core does except it's

00:03:29.760 --> 00:03:36.640
got the CPU directly talking to the operating system no extra drivers

00:03:33.920 --> 00:03:40.319
required Windows 11 already has support baked in for thread director but this

00:03:38.480 --> 00:03:43.519
should also work with Linux and if microsoft were to grace us with an

00:03:41.680 --> 00:03:47.680
update to Windows 10 it would work there too not that i'm holding my breath on the

00:03:46.159 --> 00:03:52.000
platforms that do support thread director it's going to be a pretty big

00:03:49.200 --> 00:03:56.400
deal background tasks like say a virus scan or a Windows update are going to be

00:03:53.760 --> 00:04:00.799
pushed onto the efficiency course which not only reduces the impact on

00:03:58.400 --> 00:04:05.360
foreground tasks like gaming or content creation but can also reduce power

00:04:03.280 --> 00:04:09.439
consumption and heat output while those tasks are running which on future mobile

00:04:07.680 --> 00:04:13.439
chips will mean better laptop battery life we don't have any info on those

00:04:11.680 --> 00:04:20.320
products right now but what we do know is that the k series desktop i5s i7s and

00:04:16.320 --> 00:04:23.240
i9s will all have a TDP of 125 watts in

00:04:20.320 --> 00:04:30.320
this generation which sounds ridiculously confusing because

00:04:25.919 --> 00:04:32.560
how can a core i5 have the same TDP

00:04:30.320 --> 00:04:36.960
as a core i9 but what Intel is doing here is such a breath of fresh air

00:04:34.479 --> 00:04:42.320
compared to previous releases what used to be called TDP is now processor base

00:04:39.440 --> 00:04:46.960
power or pbp and this is how much power the CPU is assured to have available to

00:04:44.080 --> 00:04:51.840
it in sustained workloads and now Intel is also advertising the maximum turbo

00:04:48.960 --> 00:04:56.720
power or mtp which is how much the CPU can use during boost with that

00:04:54.160 --> 00:05:01.360
information being disclosed now you can finally make informed decisions about

00:04:58.720 --> 00:05:07.440
your cooling setup not only that but k-series cpus are shipping with a

00:05:03.360 --> 00:05:10.400
sustained boost mode where pbp equals

00:05:07.440 --> 00:05:15.680
mtp effectively turning on multi-core enhancement by default and letting the

00:05:12.639 --> 00:05:18.320
CPU boost forever so now there's finally

00:05:15.680 --> 00:05:21.919
a reason to buy a k-series CPU i mean

00:05:19.199 --> 00:05:25.520
aside from overclocking which hey may actually be a thing again

00:05:23.680 --> 00:05:28.960
Intel wouldn't say how much headroom there is but their new extreme tuning

00:05:27.280 --> 00:05:34.080
utility is shipping with a one-click overclock function that adds a fixed 100

00:05:31.680 --> 00:05:39.120
megahertz across all cores so while that's a mild oc it definitely shows a

00:05:37.039 --> 00:05:42.720
lot of confidence in their silicon and they may have good reason to be

00:05:40.400 --> 00:05:47.600
confident not only does 12th gen core have a larger footprint in the updated

00:05:44.400 --> 00:05:49.919
lga 1700 socket it's also built on

00:05:47.600 --> 00:05:54.479
Intel's long-awaited 10 nanometer superfin process now called Intel 7

00:05:52.080 --> 00:05:58.400
because thanks to marketing nanometers have lost all meaning in ship making ice

00:05:56.639 --> 00:06:02.560
lake was the first to use this process but suffered from frequency issues that

00:06:00.319 --> 00:06:05.520
tiger lake largely resolved can i just say that it's

00:06:03.520 --> 00:06:09.440
really fitting that it would be under Intel's new ceo pat gelsinger that the

00:06:07.919 --> 00:06:14.479
new process would finally be ready for the desktop he was the chief technology

00:06:11.680 --> 00:06:19.120
officer back in 2008 when Intel first spoke about 10 nanometer like it's it's

00:06:16.800 --> 00:06:22.960
like they kept the seat warm for him but six generations of 14 nanometer chips

00:06:21.360 --> 00:06:27.039
they could have you know just used a water bottle from ltdstore.com would

00:06:24.720 --> 00:06:31.039
have been a lot cheaper for sure right the new v2 ones even have an Intel

00:06:29.120 --> 00:06:35.440
inspired design there's more to the platform than just

00:06:33.280 --> 00:06:41.280
nanometers and cores though Intel is shipping the first consumer CPU to use

00:06:37.759 --> 00:06:44.479
pci express gen 5 which

00:06:41.280 --> 00:06:46.000
feels really soon after gen 4 doesn't it

00:06:44.479 --> 00:06:51.520
although it was kind of like this with gen 1 through 3 it just took a long time

00:06:49.520 --> 00:06:57.360
for gen 4 to come around to the consumer market the benefits of pci express gen 5

00:06:54.080 --> 00:06:59.759
though aren't super clear just yet but

00:06:57.360 --> 00:07:03.759
as with gen 4 it doubles the bandwidth over its predecessor and once cards are

00:07:02.080 --> 00:07:08.319
available to take advantage of it it could allow for far more flexible lane

00:07:05.919 --> 00:07:14.080
arrangements for things like bifurcation because here's the thing eight lanes of

00:07:10.160 --> 00:07:17.759
gen 5 is equivalent to 16 lanes of gen 4

00:07:14.080 --> 00:07:19.919
or even 32 lanes of gen 3.

00:07:17.759 --> 00:07:23.919
man we are firing on all cylinders now it's kind of like the innovation that we

00:07:21.360 --> 00:07:28.240
missed out on in the 2010s is all just hitting us at once now this is great

00:07:26.560 --> 00:07:32.400
12th gen core cpus are going to ship with a combination of pci express gen 5

00:07:30.479 --> 00:07:36.639
and gen 4. perhaps the most useful improvement for right now though is the

00:07:34.479 --> 00:07:40.880
upgraded gen 4 link to the new 600 series chipsets which is double the

00:07:38.240 --> 00:07:46.639
bandwidth of both the outgoing 500 series and AMD's x570 platform not only

00:07:44.639 --> 00:07:51.759
that but the chipset now supports up to 28 gen 4 lanes taken together

00:07:49.840 --> 00:07:57.039
this should make it a lot easier for z690s new integrated wi-fi 6e and

00:07:54.800 --> 00:08:02.240
standard 2.5 gig Ethernet to coexist with things like 10 gig Ethernet and 20

00:07:59.759 --> 00:08:06.000
gig usb ports additional storage and more also

00:08:03.599 --> 00:08:09.039
it's not common but Intel is curiously supporting a dedicated clock generator

00:08:07.759 --> 00:08:12.479
for more reliable base clock overclocking wouldn't expect to see

00:08:11.039 --> 00:08:18.080
these outside of enthusiast boards though because who actually does that

00:08:15.199 --> 00:08:23.280
alder lake is also the first x86 CPU to ship with support for ddr5 now like many

00:08:21.360 --> 00:08:26.560
transition chips the memory controller on alder lake is compatible with both

00:08:25.360 --> 00:08:30.800
ddr5 and last generation ddr4 but it should

00:08:29.199 --> 00:08:34.800
be noted that you can't run both at the same time ddr5 could end up being one of

00:08:33.039 --> 00:08:39.919
the most interesting things about alder lake because it is a huge departure from

00:08:38.080 --> 00:08:43.599
just about everything we know about computer memory i've got a full deep

00:08:41.760 --> 00:08:47.200
dive on what is different about it it went up yesterday you guys are going to

00:08:45.040 --> 00:08:50.720
want to check that out the tl dr though is that internal bottlenecks have been

00:08:48.720 --> 00:08:55.680
removed the power management has been moved on to each dim and it provides

00:08:53.440 --> 00:09:00.320
much more granular overclocking control Intel's xmp3 memory overclocking spec

00:08:58.480 --> 00:09:05.200
also allows manufacturers to include up to three separate profiles now and

00:09:02.959 --> 00:09:10.480
includes two user-specified profile blocks with descriptive names for things

00:09:07.360 --> 00:09:13.680
like high bandwidth mode or low latency

00:09:10.480 --> 00:09:16.320
mode including per module voltage

00:09:13.680 --> 00:09:19.839
control each of the profiles are crc checksum so if anything should happen to

00:09:18.160 --> 00:09:23.760
corrupt one the system will actually know about it instead of happily moving

00:09:22.080 --> 00:09:28.560
that decimal point and letting up the magic smoke as for the hardware vendors

00:09:26.480 --> 00:09:33.760
self-certification is free of charge which should speed up adoption of xmp3

00:09:31.040 --> 00:09:37.200
perhaps even among lower-end memory kits that's not all Intel's doing with memory

00:09:35.519 --> 00:09:42.080
though where 11th gen core brought the ability to toggle xmp manually after the

00:09:39.680 --> 00:09:46.160
pc has been booted dynamic memory boost is a new feature that can automatically

00:09:43.839 --> 00:09:50.480
toggle it based on load it's something like turbo boost but for RAM this means

00:09:48.399 --> 00:09:54.240
that you can save on power and thermal output while retaining the benefits of

00:09:52.080 --> 00:09:57.600
high speed memory when you need it which might not be a big deal for desktops but

00:09:56.240 --> 00:10:01.040
for laptops this is going to be huge there's no

00:09:59.279 --> 00:10:05.440
software operating system support needed for it either it's entirely handled by

00:10:03.120 --> 00:10:10.399
the CPU and the BIOS so even something esoteric like temple os can use it Intel

00:10:08.000 --> 00:10:13.279
says that it's also usable with ddr4 so you don't even need to buy in on

00:10:11.600 --> 00:10:17.600
expensive ddr5 right away to use it either this will depend on BIOS support

00:10:15.839 --> 00:10:21.040
though and Intel says it'll be rolling out shortly after release

00:10:19.360 --> 00:10:24.480
which will be november 4th when we'll be validating Intel's performance claims

00:10:22.959 --> 00:10:31.440
and checking out the rest of the platform the core 9 12 900k is going to

00:10:27.360 --> 00:10:32.880
be priced at an eye watering 589

00:10:31.440 --> 00:10:38.399
that's a fair bit less than the leak suggested but it is still 50 dollars

00:10:35.360 --> 00:10:42.480
more than the 11900k and 40 dollars more

00:10:38.399 --> 00:10:44.480
than a ryzen 9 5900x a CPU that has no

00:10:42.480 --> 00:10:48.240
efficiency cores just performance course and

00:10:46.560 --> 00:10:51.200
50 more of them there's going to be some spicy

00:10:49.839 --> 00:10:54.720
benchmarks coming your way next week like i've been getting goosebumps

00:10:52.959 --> 00:10:58.480
thinking about all the innovations that AMD and Intel are going to have to make

00:10:56.560 --> 00:11:02.880
over the coming years now that they are at each other's throats again properly

00:11:00.720 --> 00:11:06.399
and we've finally got an engineer back in charge at Intel

00:11:04.480 --> 00:11:11.760
AMD for their part are expecting their next gen Ryzen processors with v cash

00:11:08.640 --> 00:11:14.959
to get up to a 25 gaming improvement and

00:11:11.760 --> 00:11:17.120
a 15 improvement overall if that is true

00:11:14.959 --> 00:11:20.880
Intel and AMD may end up neck and neck in terms of gaming performance by the

00:11:18.640 --> 00:11:25.920
end of the year and it'll be AMD's turn once again to pull off a little magic

00:11:23.040 --> 00:11:31.839
trick don't you love that CPU leapfrog that's my favorite dance

00:11:28.959 --> 00:11:34.720
and i love sponsors like nordpass nordpass wants to help you keep your

00:11:33.200 --> 00:11:39.279
private information safe and their password manager stores passwords notes

00:11:37.200 --> 00:11:44.000
credit cards anything that you want to be protected by a strong secure password

00:11:42.079 --> 00:11:46.880
with two-factor authentication it recognizes your favorite website so

00:11:45.440 --> 00:11:51.120
it'll automatically fill in your login details and allows you to create new

00:11:49.120 --> 00:11:55.120
complex and secure passwords with their built-in password generator nordpass

00:11:53.360 --> 00:11:58.639
premium starts at only two dollars and fifty cents a month and gives you

00:11:56.800 --> 00:12:03.600
additional features like data breach alerts password health reports and up to

00:12:00.880 --> 00:12:07.120
six active devices for nord passes back to school sale for a limited time you

00:12:05.360 --> 00:12:11.680
can get fifty percent off a nord pass premium plan with an extra four months

00:12:09.440 --> 00:12:16.079
for free so protect your passwords today at nordpass.com

00:12:13.519 --> 00:12:19.600
Linus and then use code minus go check out our recent first look at ddr5 for a

00:12:18.000 --> 00:12:25.120
more technical deep dive on what that means for the industry and your next pc

00:12:21.600 --> 00:12:27.519
going forward thanks for watching

00:12:25.120 --> 00:12:27.519
indeed