WEBVTT

00:00:00.080 --> 00:00:07.919
this is a tough video to make because the popular cool thing to do right now

00:00:04.240 --> 00:00:10.080
is to praise AMD and dunk on Intel and

00:00:07.919 --> 00:00:14.160
for good reason i mean it does appear to be that thanks to AMD we have

00:00:12.000 --> 00:00:19.039
experienced more excitement in the last three years than the prior decade but i

00:00:16.720 --> 00:00:22.960
need to make it anyway not because i'm trying to influence your next purchasing

00:00:20.720 --> 00:00:28.480
decision you guys should always buy the product not the cool brand but rather

00:00:25.680 --> 00:00:31.519
because i'm worried about Intel not until the company they're doing fine

00:00:30.000 --> 00:00:36.160
they're making billions of dollars but rather the people at Intel and i need

00:00:34.160 --> 00:00:39.600
them to hear this after this message from our sponsor

00:00:37.680 --> 00:00:43.600
ridge wallet ridge wallet wants to redefine the wallet with its compact

00:00:41.360 --> 00:00:47.440
frame and rfid blocking plates check out how they can help keep wallet bulge down

00:00:45.280 --> 00:00:51.399
and use offer code Linus to save 10 and get free worldwide shipping

00:00:58.239 --> 00:01:05.920
it can be really easy to forget that branding aside a company isn't some kind

00:01:03.120 --> 00:01:11.119
of hive mind entirely aligned to one purpose and that it's easy for a handful

00:01:08.960 --> 00:01:15.680
of people who are too disconnected from the customer or too focused on bean

00:01:13.680 --> 00:01:21.280
counting to steer the ship in the wrong direction in spite of the hard work and

00:01:18.880 --> 00:01:27.040
best intentions of the vast majority and the thing is while Intel does a lot of

00:01:24.080 --> 00:01:33.040
stuff wrong as someone who has actually met many individuals over there over the

00:01:30.240 --> 00:01:36.079
years i can say confidently that Intel employees

00:01:34.320 --> 00:01:42.320
well they might not have cute nicknames like googlers or yahoos but

00:01:39.439 --> 00:01:46.960
that company is chock full of really awesome people who care a lot

00:01:44.960 --> 00:01:51.280
like imagine the most hardcore overclocking enthusiast you know

00:01:49.680 --> 00:01:55.280
now imagine that he's got like an awesome beard and he's jacked and he's

00:01:53.040 --> 00:02:00.240
like carrying a surfboard i basically met that guy at an opt-in event one of

00:01:57.600 --> 00:02:04.880
my other favorites is a gentleman who is always inviting me to see the secret

00:02:02.719 --> 00:02:10.879
room at trade shows all you guys have ever seen of it is some laptop b-roll

00:02:07.840 --> 00:02:13.680
but even though there's literally no roi

00:02:10.879 --> 00:02:18.640
for the company or really for me since i can't report on it he's always sneaking

00:02:16.319 --> 00:02:22.879
me away from the laptops to get a look at what his team has been working on

00:02:20.560 --> 00:02:26.239
behind the scenes for no reason other than that he knows that i'm going to

00:02:24.160 --> 00:02:30.239
share his appreciation for it even ex-im tellers seem to take a

00:02:28.560 --> 00:02:34.480
disproportionate amount of pride in their work i've had it happen on

00:02:32.000 --> 00:02:39.840
multiple times that i do a retrospective on an old piece of enthusiast gear from

00:02:37.120 --> 00:02:43.519
Intel only to have an ex-engineer reach out and be like you know what hey i

00:02:41.840 --> 00:02:47.599
poured my soul into that thing and i'm just really glad that you liked it

00:02:45.519 --> 00:02:52.400
knowing these people i can say with confidence that many if not most of

00:02:50.400 --> 00:02:56.080
Intel's mistakes happen for reasons that are closer to

00:02:53.680 --> 00:03:00.080
human flaws than they are to pure evil

00:02:57.440 --> 00:03:04.319
and we all make mistakes let's take their most recent

00:03:01.519 --> 00:03:08.080
self-inflicted wound it's very clear from the revenue growth and steady

00:03:06.000 --> 00:03:13.040
profits over at team blue that the business folks are in charge and doing a

00:03:10.560 --> 00:03:17.440
very good job of delivering maximum shareholder value or whatever it is that

00:03:15.280 --> 00:03:21.680
they do but it's also clear that this strategy wasn't one that rewarded

00:03:19.760 --> 00:03:25.519
delivering exciting products to enthusiasts and gamers Intel spent the

00:03:23.840 --> 00:03:29.920
better part of a decade with no competition in the CPU market and used

00:03:27.920 --> 00:03:34.959
that opportunity to do what monopolies do best get very

00:03:32.400 --> 00:03:38.879
very comfortable delivering incremental generational improvements targeting the

00:03:37.200 --> 00:03:43.120
widening upgrade cycles of their corporate and small and medium business

00:03:40.959 --> 00:03:48.239
customers even something as simple as using proper thermal interface material

00:03:45.200 --> 00:03:50.480
on their cpus didn't happen until AMD

00:03:48.239 --> 00:03:53.840
forced their hand it's also clear that they were focused on growing markets

00:03:52.720 --> 00:03:59.200
rather than mature ones with r d so

00:03:56.720 --> 00:04:04.319
laser focused on mobile and data center products that talking points about

00:04:01.599 --> 00:04:08.640
performance per watt always seem to make their way into the product sheets and

00:04:05.760 --> 00:04:13.200
reviewer guides for like overclockable k series desktop cpus where young Linus

00:04:11.280 --> 00:04:17.600
was looking at them going sorry what performance per watt who cares how fast

00:04:15.519 --> 00:04:22.880
does it go how much does it overclock why is there a GPU taking up valuable

00:04:20.079 --> 00:04:27.280
silicon on my CPU where is my mainstream six core well the sleeping giant

00:04:25.199 --> 00:04:31.280
awakened two years ago and don't kid yourself they have awakened and they

00:04:29.360 --> 00:04:35.680
will come roaring back i've seen this movie a couple of times now but some of

00:04:33.360 --> 00:04:38.960
the rumors coming out have me worried that until they're ready to make a

00:04:37.280 --> 00:04:42.960
legitimate play for the performance crown again Intel might do some

00:04:41.120 --> 00:04:48.160
desperate things that will damage the things about them that i still love even

00:04:46.000 --> 00:04:53.520
if i'm not a big fan of their price to performance right now so

00:04:50.160 --> 00:04:55.440
Intel employees intellers blue badges in

00:04:53.520 --> 00:04:59.600
no particular order here are some things that i really don't want to see go away good

00:04:58.320 --> 00:05:05.199
engineering in my experience with our AMD based NVMe

00:05:02.400 --> 00:05:10.240
storage server i learned that AMD has a whole server slash data center platform

00:05:07.680 --> 00:05:15.280
out there that as far as i can tell no one ever thought to go hey what happens

00:05:13.039 --> 00:05:20.240
if i light up all these PCIe lanes with data and then actually load up the CPU

00:05:17.840 --> 00:05:23.919
at the same time turns out that the data gets retrieved so fast from that big

00:05:22.160 --> 00:05:26.880
flash array that by the time the CPU comes back around to get it

00:05:25.759 --> 00:05:32.080
it's gone rarrow wendell made an offhand comment to me

00:05:30.560 --> 00:05:35.840
that really stood out while we were troubleshooting the issue i said to him

00:05:34.400 --> 00:05:40.320
you know man i never saw any of these kinds of stupid issues when we rolled

00:05:37.520 --> 00:05:45.840
our Intel all NVMe storage server and he goes yeah Intel's got fewer PCIe lanes

00:05:43.680 --> 00:05:49.280
and they're a generation behind but they put some really good engineering into

00:05:47.440 --> 00:05:54.880
them i've always kind of taken for granted that a PCIe lane is a PCIe lane

00:05:52.800 --> 00:05:59.360
and it made me hesitate to fully transition the backbone of my company to

00:05:57.440 --> 00:06:04.080
pneumonic server until these wrinkles are sorted out either by AMD or by the

00:06:02.080 --> 00:06:08.319
Linux community through kernel patches and as someone who has forgotten more

00:06:06.000 --> 00:06:12.479
about both software and hardware than i will ever know about either of them put

00:06:10.000 --> 00:06:16.080
together when those words carry a lot of weight with me go sub to him by the way

00:06:14.240 --> 00:06:22.639
level one tags great guy then there's the talent pool and team size Intel is a

00:06:18.960 --> 00:06:25.120
much much bigger company than AMD and

00:06:22.639 --> 00:06:29.440
that can be a double-edged sword their massive r d resources are how they can

00:06:27.199 --> 00:06:35.440
be simultaneously competing head-to-head with AMD also a multi-billion dollar

00:06:32.240 --> 00:06:37.759
company on cpus samsung on data center

00:06:35.440 --> 00:06:43.039
storage broadcom on network connectivity solutions and soon to be NVIDIA on gpus

00:06:41.120 --> 00:06:47.520
and deep learning all multi-billion dollar companies but guys it's the

00:06:44.960 --> 00:06:52.560
attention to detail that makes customers happy back to my AMD server again i was

00:06:50.319 --> 00:06:58.400
blown away that there is no official Windows 10 support on epic or an onboard

00:06:56.000 --> 00:07:03.680
raid controller on the epic platform there is basically no material cost to

00:07:01.199 --> 00:07:08.639
adding either these things so as far as i can tell it just came down to r d

00:07:06.240 --> 00:07:12.800
priorities and not having enough to do everything i mean Intel might try and

00:07:10.560 --> 00:07:16.400
nickel and dime you on a license to unlock motherboard capabilities like a

00:07:14.960 --> 00:07:20.400
special raid mode but i wouldn't even consider the

00:07:18.080 --> 00:07:25.120
possibility that it wouldn't be there the next big one is partnerships Intel

00:07:22.800 --> 00:07:30.160
is always out there nurturing new and emerging technologies whether it's vr or

00:07:27.840 --> 00:07:33.759
wearables or global connectivity solutions

00:07:31.280 --> 00:07:38.639
even if those initiatives don't use any Intel silicon yet and the time and

00:07:36.639 --> 00:07:43.199
effort that they put into working with other industry leaders is a big part of

00:07:40.560 --> 00:07:46.639
why they end up at the heart of so many products i mean do you guys think it's

00:07:44.560 --> 00:07:52.319
an accident that Intel design wins outnumber AMD ones and laptops

00:07:49.039 --> 00:07:55.039
by so much even at a time when AMD

00:07:52.319 --> 00:07:59.919
genuinely has a competitive laptop product like their new 4th gen mobile

00:07:57.199 --> 00:08:04.960
ryzen finish goods manufacturers rely on silicon makers like Intel AMD and NVIDIA

00:08:02.879 --> 00:08:09.280
for a lot of guidance and engineering support something that takes time and

00:08:06.879 --> 00:08:12.400
resources if you want to do it right industry consortiums are another prime

00:08:11.199 --> 00:08:17.759
example of this name one and Intel is probably a member

00:08:15.199 --> 00:08:21.120
and recently they even gave Thunderbolt 3 to the usb implementation forum to

00:08:19.919 --> 00:08:27.599
make its features available cross-platform as usb4 something

00:08:24.560 --> 00:08:28.720
everything in my list had in common is

00:08:27.599 --> 00:08:32.399
trust and over the last little while the

00:08:30.720 --> 00:08:38.560
cracks in that trust have been starting to show meltdown spectre and since we're on the

00:08:36.080 --> 00:08:43.279
subject a Thunderbolt 3 security flaw that microsoft has evidently been wary

00:08:40.399 --> 00:08:48.240
of for years but while those problems have catchy names and they do make for

00:08:45.040 --> 00:08:49.680
good headlines they're honestly unlikely

00:08:48.240 --> 00:08:54.560
to cause real everyday problems and damage to Intel's

00:08:52.160 --> 00:08:58.800
long-term reputation with the average joe my real concern there comes from

00:08:56.880 --> 00:09:03.120
rumors about the upcoming 10th gen consumer chips early reports seem to

00:09:00.880 --> 00:09:06.880
indicate that Intel's geriatric 14 nanometer process with however many

00:09:05.200 --> 00:09:11.920
pluses we're adding to the end of it by now is going to be used to build 10 core

00:09:09.519 --> 00:09:15.600
chips that will run at 80 plus degrees underwater

00:09:13.040 --> 00:09:20.320
out of the box without any overclocking there is a reason that in the absence of

00:09:17.600 --> 00:09:25.200
competition Intel wasn't doing that obviously so

00:09:22.720 --> 00:09:29.920
that gives me some worry that a compromise has been made somewhere like

00:09:27.279 --> 00:09:35.760
i said i've seen this movie last time Intel fell behind their pentium d

00:09:32.000 --> 00:09:38.080
processors ran hot and loud and while to

00:09:35.760 --> 00:09:41.760
my knowledge the cpus didn't have a higher than acceptable failure rate the

00:09:40.399 --> 00:09:47.279
motherboards oh man that was hard that was hard to look

00:09:44.720 --> 00:09:52.240
at we would pull those boards out even just two or three years down the road

00:09:49.600 --> 00:09:55.600
from the system and the backs the backs of the boards where the vrms were

00:09:53.760 --> 00:10:00.480
soldered on on the front they would be dark and sometimes like warped like bowed

00:09:58.480 --> 00:10:05.519
from the heat even if they were somehow still running i'm not saying that's

00:10:02.640 --> 00:10:09.680
going to happen uh validation techniques and CPU power delivery technology have

00:10:07.920 --> 00:10:12.880
certainly come a long way since the early 2000s

00:10:11.279 --> 00:10:16.720
i'm just saying that pushing the envelope in this way is a dangerous path

00:10:15.279 --> 00:10:22.079
with the potential for unforeseen consequences because here's the thing

00:10:19.360 --> 00:10:27.279
when an enthusiast like like you asks me for a recommendation these days you're

00:10:24.240 --> 00:10:29.360
going to hear a lot of AMD AMD AMD

00:10:27.279 --> 00:10:32.640
can i say they're rocking it right now but when someone who might come back to

00:10:31.440 --> 00:10:36.880
me for help asks for a recommendation

00:10:34.560 --> 00:10:41.920
they are more likely to hear Intel i want to know for sure that everything

00:10:38.720 --> 00:10:44.079
from the CPU to memory compatibility to

00:10:41.920 --> 00:10:48.160
the usb controller on the motherboard to wi-fi

00:10:45.279 --> 00:10:51.680
is all just gonna basically work i'm not saying the upcoming 10th gen consumer

00:10:50.000 --> 00:10:55.920
chips are bad or that they're going to fail prematurely not by a long shot i've

00:10:54.240 --> 00:11:01.839
still got that trust i'm just saying please tread carefully guys because the

00:10:59.279 --> 00:11:06.880
better long-term play may be to just take it on the chin for a couple years

00:11:03.920 --> 00:11:10.640
learn from it especially you executives more r d to enthusiast stuff please

00:11:09.200 --> 00:11:14.720
there's not a lot of us but we're really noisy and then come back with your

00:11:12.640 --> 00:11:18.640
performance leadership and your reliability intact

00:11:16.560 --> 00:11:22.079
because it's not even my word on the subject that matters really it's the

00:11:20.399 --> 00:11:26.480
tier one system integrators that are pumping out millions of machines a year

00:11:24.320 --> 00:11:32.160
and i'm sure those guys are watching all of this much more closely than i am

00:11:30.560 --> 00:11:35.120
just like you guys are watching closely to see how he's going to segue to that

00:11:33.920 --> 00:11:39.760
sponsor micro center is open to supply all your

00:11:38.079 --> 00:11:44.160
work from home or learn from home technology needs the main gear element

00:11:41.680 --> 00:11:47.279
laptop is available at 25 micro central locations as well as on amazon for

00:11:46.079 --> 00:11:53.040
anyone who's not near one of their stores it's got an Intel core i7 9750h

00:11:50.320 --> 00:12:00.000
processor NVIDIA GeForce rtx 2070 graphics 32 whopping gigs of ddr4 memory

00:11:56.720 --> 00:12:02.880
two terabytes of NVMe storage and a 15.6

00:12:00.000 --> 00:12:07.600
inch full hd narrow bezel ips display check out it and other micro center

00:12:04.720 --> 00:12:11.760
specials at the links below if you guys enjoyed this video uh maybe

00:12:09.600 --> 00:12:15.519
you'll enjoy a more critical letter to Intel from back at computex

00:12:14.240 --> 00:12:20.720
2018 i'm not sure if anyone over there really you know took it to heart but

00:12:19.200 --> 00:12:24.639
hey i'm gonna i'm gonna keep doing these every time i feel like they're necessary