WEBVTT

00:00:00.160 --> 00:00:07.279
most computer enthusiasts know Intel for their famous x86 CPU architecture that

00:00:05.279 --> 00:00:10.639
powers most of today's desktops and laptops

00:00:08.480 --> 00:00:15.679
but did you know that not so long ago team blue actually tried to kill off x86

00:00:14.080 --> 00:00:20.880
entirely if you're not sure what i'm talking about maybe the name itanium

00:00:18.480 --> 00:00:25.199
will ring a bell huh probably not for a lot of you this was

00:00:22.720 --> 00:00:29.760
an all-new architecture and CPU brand from Intel that made its debut back in

00:00:27.119 --> 00:00:34.079
2001 and was billed as being so efficient that it would just naturally

00:00:32.000 --> 00:00:39.680
come to dominate the market at some point but these days you can barely even

00:00:36.640 --> 00:00:41.920
buy an itanium CPU anymore so

00:00:39.680 --> 00:00:46.399
what was the big idea let's start as we always do with a little bit of history

00:00:44.239 --> 00:00:50.079
back in the 1990s hewlett-packard was looking at ways to make cpus more

00:00:48.079 --> 00:00:54.320
parallel but instead of sticking more cores onto a CPU or using hyper

00:00:52.000 --> 00:00:58.559
threading hp wanted to simplify the architecture by having software

00:00:56.480 --> 00:01:03.199
determine what order to execute instructions in instead of having the

00:01:00.399 --> 00:01:06.960
CPU do the heavy lifting and hardware the hope was that this would free up

00:01:05.040 --> 00:01:10.880
processing resources to execute more instructions in parallel ultimately

00:01:08.880 --> 00:01:15.360
making for a more powerful energy efficient CPU hp took their new

00:01:13.280 --> 00:01:20.560
architecture which was called epic no not that one to Intel which then helped

00:01:17.759 --> 00:01:25.520
develop it further into ia64 the architecture used in titanium titanium

00:01:23.600 --> 00:01:31.680
was also Intel's first attempt at a 64-bit processor and actually hit the

00:01:27.920 --> 00:01:34.799
market before AMD's first 64-bit CPU

00:01:31.680 --> 00:01:36.960
opteron which released in 2003. hp and

00:01:34.799 --> 00:01:40.320
Intel felt confident that titanium would become popular in the server and

00:01:38.479 --> 00:01:45.600
workstation settings and go on to overtake x86 and home pcs as well

00:01:43.840 --> 00:01:49.840
there was even a version of Windows specifically written for titanium's

00:01:47.360 --> 00:01:53.439
architecture so why did it flop so badly we'll tell you right after this message

00:01:51.360 --> 00:01:56.720
from our sponsor i fix it i fix ifixit wants to help you repair or upgrade

00:01:55.040 --> 00:02:01.520
everything from your cameras to your game consoles their pro tech kit has 64

00:01:59.280 --> 00:02:06.479
bits and ifixit's most popular opening tools all rolled up into one package the

00:02:04.159 --> 00:02:10.879
kit also includes suction cups tweezers anti-static wrist straps and more and

00:02:08.879 --> 00:02:15.920
everything is covered by ifix's lifetime warranty go to ifixit.com techwiki and

00:02:13.599 --> 00:02:19.840
find your perfect toolkit today so it turns out that trying to determine what

00:02:17.680 --> 00:02:25.120
instructions the CPU should run in software is a lot trickier than hp

00:02:22.640 --> 00:02:28.560
and Intel thought it would be typically the CPU has a specialized group of

00:02:26.959 --> 00:02:33.920
circuits called a scheduler that determines what order to run instructions the scheduler can make

00:02:31.520 --> 00:02:37.760
these decisions much more effectively than software can since the scheduler

00:02:35.680 --> 00:02:42.239
can make adjustments based on when exactly the CPU is trying to access the

00:02:39.920 --> 00:02:46.319
memory this can't be done effectively in software since you can't figure out when

00:02:44.319 --> 00:02:50.800
the CPU is accessing memory ahead of time and because the software would

00:02:48.080 --> 00:02:54.800
pre-schedule instructions titanium cpus suffer from stalls quite frequently

00:02:53.040 --> 00:02:57.680
slowing them down there was also the issue that even though titanium's

00:02:56.080 --> 00:03:01.760
architecture was intended to be the way of the future

00:02:59.200 --> 00:03:05.760
Intel still had to do business in the present and the vast majority of

00:03:03.680 --> 00:03:11.760
software for servers and workstations was written for x86 or other non-itanium

00:03:09.519 --> 00:03:15.599
architectures running these programs on titanium meant that they had to be

00:03:13.360 --> 00:03:19.840
emulated which introduced a huge amount of performance overhead and to make

00:03:17.680 --> 00:03:24.720
matters even worse for titanium the aforementioned opteron from AMD was

00:03:22.080 --> 00:03:30.799
released two years later although optron was also 64-bit it used the

00:03:27.800 --> 00:03:34.799
x-8664 instruction set the same one that powers the chips that's probably sitting

00:03:32.959 --> 00:03:39.519
in your home pc right now and as the name implies it's natively compatible

00:03:37.120 --> 00:03:44.560
with x86 since it was essentially built on top of it so the folks running x86

00:03:42.400 --> 00:03:50.640
based servers and workstations had a far easier time upgrading to operon instead

00:03:46.879 --> 00:03:52.239
of titanium for their 64-bit fix making

00:03:50.640 --> 00:03:56.640
team red's offering much more popular Intel quickly

00:03:54.480 --> 00:04:02.080
realizing that titanium wasn't going to catch on instead implemented x8664 into

00:04:00.239 --> 00:04:06.560
their own xeon lineup of server processors which went on to become very

00:04:04.000 --> 00:04:10.720
successful so Intel actually let xeon cannibalize the titanium lineup to a

00:04:08.720 --> 00:04:15.599
large extent and titanium became relegated to niche markets only Intel

00:04:13.680 --> 00:04:20.239
did actually keep developing new models of titanium chips until 2017 but despite

00:04:18.160 --> 00:04:24.320
the fact it hung on for quite a while it never made up more than a small fraction

00:04:22.320 --> 00:04:27.280
of Intel's CPU sales so what's the lesson here if you're

00:04:25.919 --> 00:04:33.120
trying to reinvent the wheel make sure you have a car to stick that wheel on or

00:04:29.919 --> 00:04:34.880
at least like a tricycle

00:04:33.120 --> 00:04:38.400
we're not really good at metaphors anyways thanks for watching guys if you

00:04:36.479 --> 00:04:41.520
like this video hit like hit subscribe and hit us up in the comment section

00:04:39.919 --> 00:04:44.960
with your suggestions for topics that we should cover in the future