1
00:00:00,280 --> 00:00:02,960
Speaker 1: Brought to you by the reinvented two thousand twelve Camray.

2
00:00:03,160 --> 00:00:08,920
Speaker 1: It's ready. Are you get in touch with technology? With

3
00:00:09,039 --> 00:00:18,040
Speaker 1: tech Stuff from how stuff works dot com. Hello again, everyone,

4
00:00:18,079 --> 00:00:20,560
Speaker 1: and welcome to tech stuff. My name is Chris Polette

5
00:00:20,600 --> 00:00:23,160
Speaker 1: and I am an editor at how stuff works dot com.

6
00:00:23,160 --> 00:00:25,520
Speaker 1: Sitting across from me as always a senior writer Jonathan

7
00:00:25,720 --> 00:00:29,160
Speaker 1: stir Rickland. Wake up in the morning feeling like p didd.

8
00:00:29,480 --> 00:00:32,080
Speaker 1: I grabbed my glasses. I'm out the door. I'm going

9
00:00:32,240 --> 00:00:37,040
Speaker 1: to hit this city. I don't want to hit something too.

10
00:00:37,080 --> 00:00:40,479
Speaker 1: But that's not what I'm thinking about. Yeah, I I

11
00:00:40,560 --> 00:00:44,600
Speaker 1: despise the source of that. But but at the same time,

12
00:00:44,640 --> 00:00:47,400
Speaker 1: it was too it was too apropos. I could not

13
00:00:47,479 --> 00:00:50,480
Speaker 1: pass it up with all the the TikTok related things

14
00:00:50,520 --> 00:00:54,960
Speaker 1: that you could think of. It was TikTok spoiler. So, guys,

15
00:00:55,040 --> 00:00:59,080
Speaker 1: we're gonna talk today about a an interesting strategy developed

16
00:00:59,080 --> 00:01:04,240
Speaker 1: by into l uh in their micro processor micro architecture

17
00:01:04,440 --> 00:01:09,160
Speaker 1: design work. Right, Yeah, this is this is something that

18
00:01:10,120 --> 00:01:12,240
Speaker 1: chip heads would you would you call them chip heads

19
00:01:12,240 --> 00:01:15,280
Speaker 1: people who really care about the processor speed of computers.

20
00:01:15,280 --> 00:01:19,199
Speaker 1: I call them processor freaks, but with a pH Okay,

21
00:01:20,160 --> 00:01:22,319
Speaker 1: this is something that they would pay attention to, but

22
00:01:22,520 --> 00:01:25,280
Speaker 1: probably the general public doesn't know a whole lot about

23
00:01:25,440 --> 00:01:29,080
Speaker 1: because it's something that goes on somewhat behind the scenes.

24
00:01:29,080 --> 00:01:32,039
Speaker 1: Although Intel doesn't really make a big secret of of

25
00:01:32,080 --> 00:01:34,280
Speaker 1: doing it this way. No, No, they've got they've got

26
00:01:34,280 --> 00:01:37,080
Speaker 1: plenty of information on their own website, completely open to

27
00:01:37,120 --> 00:01:39,400
Speaker 1: the public that explains this, because I mean, really, it's

28
00:01:39,400 --> 00:01:42,000
Speaker 1: just it's showing a process, right. It's not giving any

29
00:01:42,040 --> 00:01:45,080
Speaker 1: proprietary information away, not in the least. It's kind of

30
00:01:45,120 --> 00:01:48,280
Speaker 1: like saying that you know, a car manufacturing plant uses

31
00:01:48,360 --> 00:01:51,000
Speaker 1: an assembly line of some sort that that tells you

32
00:01:51,120 --> 00:01:53,320
Speaker 1: the process or using, but doesn't give you a need detail.

33
00:01:54,520 --> 00:01:59,560
Speaker 1: True enough, So the uh, it's appropriately named the TikTok

34
00:01:59,680 --> 00:02:02,880
Speaker 1: strate g because it's sort of like a pendulum. Uh.

35
00:02:02,880 --> 00:02:04,880
Speaker 1: It moves one way and then the other and then

36
00:02:04,920 --> 00:02:09,240
Speaker 1: back the other way. Um, but that's not really the

37
00:02:09,440 --> 00:02:12,519
Speaker 1: pendulum is not really completely a good analogy because it's

38
00:02:12,560 --> 00:02:15,079
Speaker 1: not just moving back and forth that the process is

39
00:02:15,120 --> 00:02:18,799
Speaker 1: actually moving forward during this time. Yeah, and we can.

40
00:02:19,000 --> 00:02:21,600
Speaker 1: Let's let's before we get into the actual details of

41
00:02:21,600 --> 00:02:24,720
Speaker 1: what TikTok is all about, let's talk about the reason

42
00:02:24,800 --> 00:02:28,240
Speaker 1: TikTok even needs to exist in the first place, and

43
00:02:28,280 --> 00:02:32,880
Speaker 1: that would be from Intel's co founder Gordon Moore. Yeah. So,

44
00:02:32,880 --> 00:02:36,160
Speaker 1: so Gordy, you may remember we've talked about Gordy in

45
00:02:36,200 --> 00:02:38,640
Speaker 1: the past, and uh, you know somebody from Intel is

46
00:02:38,639 --> 00:02:42,480
Speaker 1: listening to that. You know. Yeah, we've had we've had

47
00:02:42,600 --> 00:02:45,320
Speaker 1: some important people email us in the past. Like I'm

48
00:02:45,320 --> 00:02:49,000
Speaker 1: thinking of Vinton Surf who sent me an email and

49
00:02:49,040 --> 00:02:51,840
Speaker 1: that just it's hard for me to remember that that

50
00:02:52,200 --> 00:02:57,639
Speaker 1: important really like people I really legitimately admire and respect

51
00:02:57,680 --> 00:03:00,240
Speaker 1: and who have had an enormous impact upon in the

52
00:03:00,400 --> 00:03:04,640
Speaker 1: technology industries. Um, I can hear what I have to say.

53
00:03:04,960 --> 00:03:07,040
Speaker 1: That just blows my mind because normally I'm just having

54
00:03:07,040 --> 00:03:08,760
Speaker 1: these conversations in a room and we don't have a

55
00:03:08,800 --> 00:03:12,560
Speaker 1: microphone in front of me. But anyway, Gordy, uh night.

56
00:03:12,600 --> 00:03:17,440
Speaker 1: Back in NINETI wrote this paper about cramming more components

57
00:03:17,919 --> 00:03:20,959
Speaker 1: into a chip. I can't remember. There's something along those

58
00:03:20,960 --> 00:03:24,360
Speaker 1: lines as the title I'm paraphrasing, but I can. I

59
00:03:24,360 --> 00:03:26,280
Speaker 1: can look that up for you if you want. You

60
00:03:26,360 --> 00:03:28,079
Speaker 1: go ahead and look that up while I talk about

61
00:03:28,160 --> 00:03:30,480
Speaker 1: about this. So we're talking about Moore's law here. So

62
00:03:30,880 --> 00:03:34,560
Speaker 1: Gordon came up with this observation, and it was an observation.

63
00:03:34,560 --> 00:03:36,760
Speaker 1: You've got to remember that Moore's law was not some

64
00:03:36,800 --> 00:03:39,800
Speaker 1: sort of fundamental law of the universe. The observation was

65
00:03:39,880 --> 00:03:43,080
Speaker 1: that over a certain period of time, and I believe

66
00:03:43,360 --> 00:03:47,120
Speaker 1: initially it might have been between twelve and eighteen months,

67
00:03:47,160 --> 00:03:50,040
Speaker 1: it's now closer to twenty four months, but over a

68
00:03:50,080 --> 00:03:53,880
Speaker 1: certain amount of time, More observed that the number of

69
00:03:53,920 --> 00:03:57,560
Speaker 1: transistors that one that a company could fit on a

70
00:03:57,640 --> 00:04:02,680
Speaker 1: single square inch of so con wafer material doubled, so

71
00:04:02,960 --> 00:04:06,080
Speaker 1: you could fit twice as many transistors on a chip

72
00:04:06,640 --> 00:04:11,120
Speaker 1: UH within a year or two years um. And that

73
00:04:11,320 --> 00:04:13,720
Speaker 1: he his observation was that this was a trend that

74
00:04:13,760 --> 00:04:17,560
Speaker 1: would continue indefinitely until we started to reach some fundamental

75
00:04:17,600 --> 00:04:21,680
Speaker 1: limits of how small we could make transistors. And at

76
00:04:21,680 --> 00:04:24,160
Speaker 1: the time, no one was really sure you know how

77
00:04:24,200 --> 00:04:27,120
Speaker 1: long that would be. And it's the crazy thing is

78
00:04:27,160 --> 00:04:29,320
Speaker 1: it's held true even today. And you've got to remember,

79
00:04:29,360 --> 00:04:34,320
Speaker 1: this is an exponential uh pattern, right, I mean it's

80
00:04:34,320 --> 00:04:37,960
Speaker 1: it's decreasing, the sizes decreasing by half, the number is

81
00:04:38,080 --> 00:04:42,960
Speaker 1: increasing bye bye, you know, twice as much each time frame,

82
00:04:43,520 --> 00:04:46,880
Speaker 1: So before too long you get an incredible number of

83
00:04:46,880 --> 00:04:51,359
Speaker 1: transistors on a silicon chip. The name of the paper. Yeah,

84
00:04:51,400 --> 00:04:53,880
Speaker 1: And as a matter of fact, Jonathan cited this for

85
00:04:54,000 --> 00:04:58,880
Speaker 1: his really awesome article on War's Law. Um a couple

86
00:04:58,920 --> 00:05:01,880
Speaker 1: of years ago. Wow, that that along, are you uh called?

87
00:05:01,920 --> 00:05:05,600
Speaker 1: It's actually called cramming more components onto integrated circuits. So yeah,

88
00:05:05,640 --> 00:05:07,640
Speaker 1: that's pretty much right. Yeah, it was pretty close. But

89
00:05:07,680 --> 00:05:11,039
Speaker 1: you can find that in Electronics. That's the name of

90
00:05:11,080 --> 00:05:16,120
Speaker 1: the journal. It was April volume thirty eight, number eight.

91
00:05:16,160 --> 00:05:18,120
Speaker 1: If you want to read it, and it's actually it's

92
00:05:18,160 --> 00:05:20,760
Speaker 1: not a dry read. You can actually find a link

93
00:05:20,800 --> 00:05:23,000
Speaker 1: to it on Intel. If you go to Intel and

94
00:05:23,040 --> 00:05:25,720
Speaker 1: you start looking at Moore's law, there is a link

95
00:05:25,760 --> 00:05:28,159
Speaker 1: to a PDF of this paper, and I do recommend

96
00:05:28,160 --> 00:05:31,159
Speaker 1: you read it if you're interested in the original observation.

97
00:05:31,720 --> 00:05:33,800
Speaker 1: One thing that I thought was really interesting was that

98
00:05:33,880 --> 00:05:37,880
Speaker 1: Moore was pointing out this isn't just a technological issue.

99
00:05:37,920 --> 00:05:40,600
Speaker 1: In fact, that was not the main thrust of his paper.

100
00:05:41,200 --> 00:05:44,440
Speaker 1: It's it's a financial issue because one, you have to

101
00:05:44,440 --> 00:05:46,880
Speaker 1: find the technology to be able to decrease the size

102
00:05:46,880 --> 00:05:49,719
Speaker 1: of these transistors, but too you have to make that

103
00:05:49,760 --> 00:05:53,480
Speaker 1: technology affordable enough to use for it to make sense

104
00:05:54,160 --> 00:05:57,320
Speaker 1: to use it right if you if you develop the

105
00:05:57,360 --> 00:06:02,120
Speaker 1: technology to create a up that has UH an incredible

106
00:06:02,200 --> 00:06:05,800
Speaker 1: number of transistors on it, but it's slow, inefficient, it

107
00:06:06,240 --> 00:06:09,880
Speaker 1: costs a lot of money to create each chip. You're

108
00:06:09,920 --> 00:06:11,680
Speaker 1: not you don't have a good business plan because you

109
00:06:11,720 --> 00:06:15,480
Speaker 1: can't sell those chips to consumers. They would be too expensive.

110
00:06:15,560 --> 00:06:18,400
Speaker 1: You would never recapture those costs. So you have to

111
00:06:18,400 --> 00:06:20,800
Speaker 1: be able to one develop the technology and to develop

112
00:06:20,880 --> 00:06:24,600
Speaker 1: the right procedure so that the technology is efficient and

113
00:06:24,720 --> 00:06:28,839
Speaker 1: you can actually make money off of it. Speaking of which, um,

114
00:06:28,880 --> 00:06:32,720
Speaker 1: some people have predicted the demise of Moore's law, of

115
00:06:32,760 --> 00:06:36,599
Speaker 1: course due to the physical limitations UH for years, but

116
00:06:36,880 --> 00:06:40,800
Speaker 1: also due to the recent UH financial troubles the world over.

117
00:06:41,200 --> 00:06:44,000
Speaker 1: People have said, well, you know, there won't be really

118
00:06:44,000 --> 00:06:48,400
Speaker 1: a need to have faster, faster, faster, faster computers because

119
00:06:48,600 --> 00:06:52,719
Speaker 1: people can't afford to go by them. So UM, but

120
00:06:52,800 --> 00:06:55,120
Speaker 1: I haven't really seen it slow down as much as

121
00:06:55,320 --> 00:06:58,920
Speaker 1: become sort of sort of zag a little bit. And

122
00:06:59,000 --> 00:07:02,480
Speaker 1: that when and the past years we've had single core

123
00:07:02,560 --> 00:07:05,760
Speaker 1: processors and you would see a speed a substantial speed

124
00:07:05,760 --> 00:07:08,280
Speaker 1: boost over the period of a year. Um, you know

125
00:07:08,320 --> 00:07:11,000
Speaker 1: it would go from one gigga hurts to one point

126
00:07:11,040 --> 00:07:13,480
Speaker 1: six gigga hurts, and you know then to two point

127
00:07:13,520 --> 00:07:16,680
Speaker 1: to giga hurts. Well, now we have multi core processors,

128
00:07:16,880 --> 00:07:20,280
Speaker 1: and they don't seem to move as fast because you know,

129
00:07:20,320 --> 00:07:23,160
Speaker 1: we're going from uh, you know, three point two to

130
00:07:23,280 --> 00:07:26,640
Speaker 1: three point six gigga hurts. But then we're also doubling

131
00:07:26,640 --> 00:07:29,840
Speaker 1: the number of cours on that chip, so they are

132
00:07:29,880 --> 00:07:33,520
Speaker 1: getting faster. Just doesn't appear to move as fast because

133
00:07:33,600 --> 00:07:36,040
Speaker 1: of the way it doesn't. At least that's that's my

134
00:07:36,120 --> 00:07:39,440
Speaker 1: personal observer. Yeah, the the jumps, the jumps, an actual

135
00:07:39,640 --> 00:07:44,880
Speaker 1: number of cycles per second that a processor is capable

136
00:07:45,040 --> 00:07:48,360
Speaker 1: of of executing doesn't seem to be jumping at the

137
00:07:48,400 --> 00:07:51,920
Speaker 1: same rate as it was, you know, ten years ago, right,

138
00:07:52,360 --> 00:07:55,680
Speaker 1: But the the multi core approach has created a more

139
00:07:55,720 --> 00:08:00,720
Speaker 1: efficient way to deal with computational problems, and thus ultimately

140
00:08:00,760 --> 00:08:05,000
Speaker 1: you're computing power has has increased, even though the number

141
00:08:05,040 --> 00:08:07,640
Speaker 1: of cycles themselves may not have jumped as high as

142
00:08:07,640 --> 00:08:09,800
Speaker 1: you would have expected. And that's that's part of the

143
00:08:09,800 --> 00:08:12,880
Speaker 1: whole TikTok philosophy actually, which I guess we can kind

144
00:08:12,880 --> 00:08:16,440
Speaker 1: of segue into UH. Until really started to adopt this

145
00:08:16,560 --> 00:08:21,200
Speaker 1: around two thousand seven. Really, was it that long ago? Yeah? Yeah,

146
00:08:21,280 --> 00:08:23,920
Speaker 1: it was. Um it was right around then when they

147
00:08:23,920 --> 00:08:27,520
Speaker 1: had started to develop the core technology. That was when

148
00:08:27,600 --> 00:08:31,440
Speaker 1: the core technology was first starting to be uh UM introduced.

149
00:08:31,440 --> 00:08:35,600
Speaker 1: And at that time, the the number of the the

150
00:08:35,760 --> 00:08:42,240
Speaker 1: discreet elements on a chip, we're around the sixty nanometer scale.

151
00:08:43,080 --> 00:08:45,640
Speaker 1: So we're already talking about on the nano scale, which

152
00:08:45,679 --> 00:08:49,040
Speaker 1: is yeah, remember a nanometer, And I got a lot

153
00:08:49,080 --> 00:08:51,080
Speaker 1: of people yelling at me for calling it a nanometer.

154
00:08:51,520 --> 00:08:53,520
Speaker 1: I'm sorry, I get people yelling at me, no matter what,

155
00:08:53,559 --> 00:08:55,400
Speaker 1: we're gonna go nanometer this time, and then all the

156
00:08:55,400 --> 00:08:57,120
Speaker 1: other people can yell at me because I'm sure they

157
00:08:57,120 --> 00:09:01,360
Speaker 1: were tired of being left out. So a nanometer is

158
00:09:01,440 --> 00:09:05,360
Speaker 1: one billionth of a meter, so this is incredibly tiny.

159
00:09:05,360 --> 00:09:07,240
Speaker 1: We're talking about on a scale where even a moat

160
00:09:07,360 --> 00:09:11,360
Speaker 1: of dust on a silicon wafer will ruin an entire

161
00:09:11,440 --> 00:09:15,280
Speaker 1: chip because the mode of dust dwarfs the elements that

162
00:09:15,320 --> 00:09:19,040
Speaker 1: would be printed on that chip. UM. So, sixty five

163
00:09:19,120 --> 00:09:22,199
Speaker 1: nanometer scale for the core technology, now that was technically

164
00:09:22,640 --> 00:09:26,600
Speaker 1: a talk that meant that Intel had already developed a

165
00:09:26,679 --> 00:09:31,000
Speaker 1: chip that could be at sixty five nanometers that would

166
00:09:31,000 --> 00:09:34,319
Speaker 1: be the size of the transistors essentially on that chip.

167
00:09:35,240 --> 00:09:38,439
Speaker 1: So the core technology was a new way of arranging

168
00:09:38,480 --> 00:09:42,320
Speaker 1: those sixty five nanometer transistors in such a way that

169
00:09:42,360 --> 00:09:47,280
Speaker 1: they were more efficiently used, so that they consumed less power,

170
00:09:47,720 --> 00:09:50,880
Speaker 1: they had better output, there was a more streamlined flow,

171
00:09:51,880 --> 00:09:54,640
Speaker 1: so that they were, in a sense an essence, a

172
00:09:54,760 --> 00:09:57,319
Speaker 1: more powerful chip. Because one thing we also need to

173
00:09:57,400 --> 00:09:59,800
Speaker 1: remember about Moore's law is today a lot of people

174
00:09:59,800 --> 00:10:02,640
Speaker 1: in herpret it not as there are twice as many

175
00:10:02,679 --> 00:10:06,440
Speaker 1: transistors now as there were two years ago, but rather

176
00:10:06,679 --> 00:10:09,920
Speaker 1: the chip itself is twice as powerful as it was

177
00:10:10,120 --> 00:10:13,240
Speaker 1: two years ago. This is not exactly the same thing.

178
00:10:13,240 --> 00:10:16,080
Speaker 1: They're related, but you can get more power out of

179
00:10:16,080 --> 00:10:19,440
Speaker 1: a chip just by realigning certain elements and making a

180
00:10:19,480 --> 00:10:24,160
Speaker 1: more efficient workflow. You know, when you start talking about

181
00:10:24,200 --> 00:10:26,959
Speaker 1: more power, I'm starting to get James doing in the

182
00:10:27,040 --> 00:10:30,080
Speaker 1: back of my head. I'm giving her all. She's gone.

183
00:10:30,679 --> 00:10:33,440
Speaker 1: Um so coming to the Jeffreys tubes in your ear.

184
00:10:33,640 --> 00:10:38,760
Speaker 1: Nice Jeffreys too in your PC, trying to a squeak

185
00:10:38,800 --> 00:10:40,640
Speaker 1: out that extra Now, if you guys want to know,

186
00:10:40,920 --> 00:10:42,200
Speaker 1: if you guys want to know more about that, you

187
00:10:42,240 --> 00:10:44,760
Speaker 1: need to read how warp speed works. We actually have

188
00:10:44,800 --> 00:10:49,200
Speaker 1: it on the site. Um the so, yeah, the whole

189
00:10:49,200 --> 00:10:52,000
Speaker 1: basis of tiktalk, right, we have. We've kind of danced

190
00:10:52,000 --> 00:10:55,480
Speaker 1: around it. But Intel's TikTok strategy is that the TICK

191
00:10:56,240 --> 00:10:58,800
Speaker 1: is figuring out a way to reduce the size of

192
00:10:58,800 --> 00:11:02,560
Speaker 1: those transistors, but you bace it upon the previous UH

193
00:11:02,880 --> 00:11:07,160
Speaker 1: chips micro architecture. All right. So so it's like you've

194
00:11:07,160 --> 00:11:10,040
Speaker 1: got the plans for a house. Now you're going to

195
00:11:10,080 --> 00:11:12,319
Speaker 1: build that same house, but you're gonna build it at

196
00:11:12,400 --> 00:11:15,679
Speaker 1: half that scale. I got it, I got it. So

197
00:11:15,760 --> 00:11:18,560
Speaker 1: the TALK strategy is finding out the best way to

198
00:11:18,760 --> 00:11:22,400
Speaker 1: use those smaller components so that it is the most

199
00:11:22,440 --> 00:11:28,280
Speaker 1: efficient effective transistor arrangement. So in that case, what you

200
00:11:28,360 --> 00:11:31,080
Speaker 1: do is you look at the plans for that house

201
00:11:31,640 --> 00:11:34,120
Speaker 1: that you built at half the size of the previous one,

202
00:11:34,320 --> 00:11:36,680
Speaker 1: and you say, all right, I'm going to rearrange this

203
00:11:36,760 --> 00:11:39,480
Speaker 1: now so that this house makes sense at this scale.

204
00:11:39,600 --> 00:11:41,560
Speaker 1: It's I'm not going to change the scale. I'm just

205
00:11:41,559 --> 00:11:44,320
Speaker 1: going to change the layout of the house. So what

206
00:11:44,360 --> 00:11:48,000
Speaker 1: you're saying is they they create a blueprint for a

207
00:11:48,080 --> 00:11:52,640
Speaker 1: chip the next UH and then the next cycle they

208
00:11:52,640 --> 00:11:59,400
Speaker 1: work on making the components using that blueprint uh more efficient, right,

209
00:11:59,679 --> 00:12:02,000
Speaker 1: and then and they take the efficiency that they learn

210
00:12:02,080 --> 00:12:05,160
Speaker 1: on that cycle and build a new architecture on it

211
00:12:05,240 --> 00:12:07,880
Speaker 1: for the next cycle. And so on one side they're

212
00:12:07,920 --> 00:12:12,440
Speaker 1: working on making everything more uh you know, reducing the

213
00:12:12,480 --> 00:12:14,920
Speaker 1: size of it, and the next cycle is all about

214
00:12:14,960 --> 00:12:19,880
Speaker 1: the actual design of it. Right, So exactly goes size

215
00:12:19,920 --> 00:12:22,400
Speaker 1: design size design. And this is, by the way, a

216
00:12:22,520 --> 00:12:26,600
Speaker 1: never ending research process. It's not like you've got people

217
00:12:26,679 --> 00:12:29,960
Speaker 1: researching how to reduce the size and then they stop,

218
00:12:30,040 --> 00:12:32,040
Speaker 1: and then they switch to figure out how to make

219
00:12:32,120 --> 00:12:34,640
Speaker 1: it more efficient. You've got teams working on both of

220
00:12:34,679 --> 00:12:39,320
Speaker 1: these things simultaneously. And so um, it's interesting. You know

221
00:12:39,400 --> 00:12:41,680
Speaker 1: we look back and the core technology being the talk

222
00:12:41,679 --> 00:12:46,600
Speaker 1: at SS Well, the next one was the pen Ryn chip.

223
00:12:47,160 --> 00:12:49,400
Speaker 1: That was that was the tick. Yeah. Now that one

224
00:12:49,600 --> 00:12:53,160
Speaker 1: used the same micro architecture as the core processors, but

225
00:12:53,240 --> 00:12:56,520
Speaker 1: this time they had reduced the size of the elements

226
00:12:56,559 --> 00:13:01,480
Speaker 1: to the forty five nanometer scale. Now, after Pentrin came

227
00:13:01,520 --> 00:13:05,600
Speaker 1: one of the chips that I wrote about, Yes, the Haleem,

228
00:13:05,679 --> 00:13:10,360
Speaker 1: yes Nehalem microprocessor micro architecture. That was the next talk,

229
00:13:10,800 --> 00:13:15,160
Speaker 1: and the Haleem had introduced a lot of interesting features

230
00:13:15,200 --> 00:13:19,800
Speaker 1: like multi threading and and uh arranging memory in such

231
00:13:19,840 --> 00:13:22,360
Speaker 1: a way so that it would uh that the various

232
00:13:22,400 --> 00:13:25,600
Speaker 1: cores could share memory certain parts of memory very quickly

233
00:13:25,640 --> 00:13:28,080
Speaker 1: to make it more efficient. That's what we were talking

234
00:13:28,120 --> 00:13:31,120
Speaker 1: about here. They're actually rearranging the elements that are on

235
00:13:31,160 --> 00:13:34,880
Speaker 1: the microprocessor chip in such a way that that the

236
00:13:35,000 --> 00:13:39,360
Speaker 1: data flow is faster just because it makes more sense, right,

237
00:13:39,840 --> 00:13:42,600
Speaker 1: And it doesn't necessarily it's it's it's an arrangement that

238
00:13:42,640 --> 00:13:45,320
Speaker 1: would not have necessarily worked at the larger scale because

239
00:13:45,320 --> 00:13:50,200
Speaker 1: you could not physically find that same configuration with the

240
00:13:50,280 --> 00:13:52,960
Speaker 1: elements being larger. They had to be that size for

241
00:13:53,000 --> 00:13:55,200
Speaker 1: you to be able to kind of shift them around.

242
00:13:55,240 --> 00:13:57,400
Speaker 1: It's almost like one of those puzzles where you have

243
00:13:57,520 --> 00:14:00,440
Speaker 1: one piece missing and you have to slow the other

244
00:14:00,480 --> 00:14:03,240
Speaker 1: pieces around until you make the right picture. It's kind

245
00:14:03,240 --> 00:14:05,439
Speaker 1: of like that. You know you've got you've got this

246
00:14:05,679 --> 00:14:08,520
Speaker 1: certain amount of space that you are allowed to use,

247
00:14:09,000 --> 00:14:11,400
Speaker 1: and you have elements of a certain size, and you

248
00:14:11,440 --> 00:14:13,160
Speaker 1: have to find the right way to fit all those

249
00:14:13,160 --> 00:14:15,839
Speaker 1: elements together so that it's the most efficient possible. Well,

250
00:14:15,880 --> 00:14:18,319
Speaker 1: with the larger ones, you just don't have as many configurations.

251
00:14:18,320 --> 00:14:21,520
Speaker 1: You don't have as much freedom because the elements themselves

252
00:14:21,520 --> 00:14:23,560
Speaker 1: are bigger. You don't, you can't, you know, there's only

253
00:14:23,600 --> 00:14:28,280
Speaker 1: so many configurations you can use. So after Nehalem came Westmere,

254
00:14:28,840 --> 00:14:31,400
Speaker 1: and Westmere was another tick. So it was using the

255
00:14:31,480 --> 00:14:35,720
Speaker 1: same micro architecture as Nehalem. But now we have gone

256
00:14:35,760 --> 00:14:40,640
Speaker 1: down to the thirty two nanometer size, right, And here's

257
00:14:40,680 --> 00:14:44,240
Speaker 1: where another challenge comes in, because when you get down

258
00:14:44,240 --> 00:14:47,920
Speaker 1: to this size, this this nanometer scale, you're starting to

259
00:14:48,000 --> 00:14:54,000
Speaker 1: encounter some pretty funky quantum physics problems, quantum mechanics problems.

260
00:14:55,000 --> 00:14:58,360
Speaker 1: You're talking about electron tunneling. That would problem, that would

261
00:14:58,360 --> 00:15:00,400
Speaker 1: be a big one. Yeah, because I mean, of course,

262
00:15:00,440 --> 00:15:02,520
Speaker 1: and we've talked about electron tunneling before as well, so

263
00:15:02,680 --> 00:15:04,960
Speaker 1: long time listeners will know what we're talking about here.

264
00:15:05,320 --> 00:15:09,920
Speaker 1: Electrons have this wacky little way of apparently defying the

265
00:15:10,000 --> 00:15:13,000
Speaker 1: laws of physics as we understand them. Uh. Actually it's

266
00:15:13,000 --> 00:15:16,640
Speaker 1: not entirely true quantum physics. It makes perfect sense. Classic physics,

267
00:15:16,680 --> 00:15:19,000
Speaker 1: it makes no sense at all. So on my scale

268
00:15:19,120 --> 00:15:21,480
Speaker 1: where I you know, if I drop something, it falls

269
00:15:21,560 --> 00:15:24,320
Speaker 1: and then it hits something and it stops. That makes

270
00:15:24,320 --> 00:15:27,280
Speaker 1: sense to me, Right, that's the world I grew up in.

271
00:15:28,160 --> 00:15:30,560
Speaker 1: If I were to drop something and it would pass

272
00:15:30,720 --> 00:15:34,000
Speaker 1: through the floor beneath me without making a hole and

273
00:15:34,040 --> 00:15:38,520
Speaker 1: then continue to go down, I'd say, huh, that's strange.

274
00:15:39,000 --> 00:15:42,120
Speaker 1: But on the quantum world, not so much. Those wacky

275
00:15:42,120 --> 00:15:45,280
Speaker 1: electrons Tuesdays at eight. Yes, so electrons, if if a

276
00:15:45,320 --> 00:15:48,480
Speaker 1: barrier is thin enough, and we're talking about a couple

277
00:15:48,520 --> 00:15:52,320
Speaker 1: of nanometers wide or thick, I should say if if

278
00:15:52,760 --> 00:15:56,840
Speaker 1: if it's thin enough and electron can tunnel through, and

279
00:15:56,920 --> 00:16:00,400
Speaker 1: that it passes through that barrier as if the barrier

280
00:16:00,440 --> 00:16:03,280
Speaker 1: we're not there. It doesn't make a hole. And in

281
00:16:03,320 --> 00:16:05,840
Speaker 1: a way, it almost is like it's on one side

282
00:16:05,840 --> 00:16:07,600
Speaker 1: of the barrier at one moment and on the other

283
00:16:07,640 --> 00:16:10,720
Speaker 1: side of the barrier the next um. But that's one

284
00:16:10,760 --> 00:16:13,360
Speaker 1: of the problems. And then what Intel has found is

285
00:16:13,360 --> 00:16:16,400
Speaker 1: they found that by using different materials, by by switching

286
00:16:16,800 --> 00:16:21,360
Speaker 1: their transistory gates to other kinds of elements, that they

287
00:16:21,360 --> 00:16:25,200
Speaker 1: were more resistant to electron tunneling. And you don't want

288
00:16:25,200 --> 00:16:28,840
Speaker 1: electron tunneling, by the way, because leaking electrons means that

289
00:16:28,920 --> 00:16:33,600
Speaker 1: transistor is pretty much useless. Transistors are all about governing

290
00:16:33,800 --> 00:16:36,480
Speaker 1: the flow of electrons and if you can't stop them,

291
00:16:36,600 --> 00:16:42,720
Speaker 1: then the transistor is always open essentially. Sorry. I think

292
00:16:42,720 --> 00:16:46,040
Speaker 1: two on semiconductors, which we've discussed in the past. Two

293
00:16:46,040 --> 00:16:51,520
Speaker 1: because semiconductors are uh you know, essential to running pretty

294
00:16:51,600 --> 00:16:55,520
Speaker 1: much all kinds of electronics, including computers, um and it's

295
00:16:55,560 --> 00:16:59,400
Speaker 1: all about controlling using certain materials to control the flow

296
00:16:59,400 --> 00:17:01,240
Speaker 1: of electron So you knew, in order to have a

297
00:17:02,080 --> 00:17:05,320
Speaker 1: computer processor to function as it needs to, it also

298
00:17:05,480 --> 00:17:07,959
Speaker 1: it needs to be able to control when and where

299
00:17:07,960 --> 00:17:14,720
Speaker 1: the electrons and the uh inside the actual transistors are going. Yes,

300
00:17:15,000 --> 00:17:17,240
Speaker 1: So I mean that's it's it's crucial. And if you

301
00:17:17,359 --> 00:17:21,439
Speaker 1: start having electrons going willy nilly and put it, your

302
00:17:21,480 --> 00:17:24,000
Speaker 1: processor is just gonna have errors. It's not going to

303
00:17:24,040 --> 00:17:27,880
Speaker 1: be able to compute things because it can't. You know,

304
00:17:27,960 --> 00:17:30,720
Speaker 1: the data that's working from the operations that's performing are

305
00:17:30,800 --> 00:17:34,720
Speaker 1: all going to be affected by that electron leakage. So

306
00:17:35,080 --> 00:17:36,679
Speaker 1: you have to find a way to reduce that and

307
00:17:36,760 --> 00:17:39,639
Speaker 1: Intel has been doing that by experimenting with different materials.

308
00:17:39,760 --> 00:17:43,240
Speaker 1: So we left off at Westmere, which was the tick

309
00:17:43,359 --> 00:17:46,480
Speaker 1: we talked about going down to thirty tick we talked about.

310
00:17:46,640 --> 00:17:49,679
Speaker 1: It was the tick we talked about right and the

311
00:17:49,760 --> 00:17:52,960
Speaker 1: next one you actually wrote about yourself again. Yeah, it

312
00:17:53,320 --> 00:17:55,399
Speaker 1: turns out I write a lot about the talks. I

313
00:17:55,400 --> 00:17:58,520
Speaker 1: haven't written about the ticks. But the next talk is,

314
00:17:58,520 --> 00:18:01,520
Speaker 1: of course, as a time we're recording this podcast, the

315
00:18:01,560 --> 00:18:07,600
Speaker 1: most recent Intel processor the sandy Bridge processor, and sandy

316
00:18:07,640 --> 00:18:10,480
Speaker 1: Bridge is again it's at that thirty two nanometer scale,

317
00:18:10,480 --> 00:18:12,359
Speaker 1: because remember it's going to be on the same scale

318
00:18:12,359 --> 00:18:15,520
Speaker 1: as the Tick before it, but it's got a different layout.

319
00:18:15,600 --> 00:18:18,760
Speaker 1: It's no longer based on the new halam Mark micro architecture.

320
00:18:18,800 --> 00:18:22,160
Speaker 1: It's got its own micro architecture, which includes a section

321
00:18:22,359 --> 00:18:26,439
Speaker 1: on the chip specifically dedicated to graphics processing, which that

322
00:18:26,560 --> 00:18:29,440
Speaker 1: was the big thing that set it apart from its predecessors.

323
00:18:29,760 --> 00:18:34,600
Speaker 1: It also has a very small bowling alley. Alright, I

324
00:18:34,640 --> 00:18:36,360
Speaker 1: don't even know where you're going with that joke. I'm

325
00:18:36,359 --> 00:18:39,159
Speaker 1: just gonna I was just imagining I was going with

326
00:18:39,200 --> 00:18:43,560
Speaker 1: your metaphor earlier in thinking about the building. You know,

327
00:18:43,600 --> 00:18:45,960
Speaker 1: it's got its own graphics processor and a bowling alley. Okay,

328
00:18:45,960 --> 00:18:48,240
Speaker 1: I'm gonna I'm gonna call that on a gutter ball

329
00:18:48,359 --> 00:18:53,959
Speaker 1: right now. Yes, the sandy Bridge micro architecture is is

330
00:18:54,240 --> 00:18:56,640
Speaker 1: brand new as of the time we're recording this. Yes,

331
00:18:57,000 --> 00:19:00,760
Speaker 1: so new that there are problems with other chips associated

332
00:19:00,800 --> 00:19:03,880
Speaker 1: with sandy Bridge, not sandy Bridge itself, we should point out,

333
00:19:04,240 --> 00:19:05,960
Speaker 1: but we can talk a little bit about that. It's

334
00:19:06,000 --> 00:19:09,160
Speaker 1: only it's only vaguely related to what we're chatting about

335
00:19:09,200 --> 00:19:12,480
Speaker 1: in this in this podcast. So the neat thing about

336
00:19:12,480 --> 00:19:15,800
Speaker 1: the graphics processing, of course, is that this means that

337
00:19:15,840 --> 00:19:18,760
Speaker 1: if you have a computer with a sandy Bridge processor,

338
00:19:18,840 --> 00:19:20,400
Speaker 1: especially if you have one of the faster ones, because

339
00:19:20,400 --> 00:19:23,119
Speaker 1: they'd come in different flavors. You know, there's some that

340
00:19:23,200 --> 00:19:27,040
Speaker 1: can have there's some that have two cores and can

341
00:19:27,119 --> 00:19:30,320
Speaker 1: run up to four threads of data. And then there

342
00:19:30,440 --> 00:19:32,200
Speaker 1: it goes all the way up to I think four

343
00:19:32,240 --> 00:19:35,399
Speaker 1: cores that can run eight threads of data. Um, it

344
00:19:35,440 --> 00:19:37,040
Speaker 1: may even go higher than that. I have to look

345
00:19:37,040 --> 00:19:40,880
Speaker 1: it up again. But the the you know, you get

346
00:19:40,920 --> 00:19:43,520
Speaker 1: one of the faster ones and you get this graphics

347
00:19:43,520 --> 00:19:46,480
Speaker 1: processing built onto the chip, it means that you may

348
00:19:46,520 --> 00:19:51,320
Speaker 1: not need a dedicated graphics processing unit to add to

349
00:19:51,320 --> 00:19:53,040
Speaker 1: your computer in order to play some of the more

350
00:19:53,160 --> 00:19:58,400
Speaker 1: advanced video games or to do things like video editing. Um,

351
00:19:58,440 --> 00:20:03,080
Speaker 1: you you might not need more additional power because you've

352
00:20:03,119 --> 00:20:05,960
Speaker 1: got everything you need on that one micro chip, which

353
00:20:06,000 --> 00:20:08,720
Speaker 1: is pretty fascinating. I mean, that chip is tiny, and

354
00:20:08,760 --> 00:20:11,760
Speaker 1: to think that it does the the equivalent of a

355
00:20:11,800 --> 00:20:17,320
Speaker 1: dedicated graphics processing card is a phenomenal. Now, granted, I'm

356
00:20:17,359 --> 00:20:20,000
Speaker 1: sure there are going to be games out there that

357
00:20:20,040 --> 00:20:22,080
Speaker 1: if you crank them to the highest setting, you're still

358
00:20:22,160 --> 00:20:25,120
Speaker 1: gonna want your own graphics processing card because it's not

359
00:20:25,800 --> 00:20:28,399
Speaker 1: it's not able to you know, take over the entire load.

360
00:20:29,680 --> 00:20:32,560
Speaker 1: I had a feeling when you said that that someone

361
00:20:32,680 --> 00:20:36,680
Speaker 1: will write in to to tell you that that, uh,

362
00:20:36,760 --> 00:20:38,320
Speaker 1: you know, that is not the case that you're if

363
00:20:38,320 --> 00:20:41,080
Speaker 1: you're going to be running a high frame rate first

364
00:20:41,119 --> 00:20:43,199
Speaker 1: person shooter or you know, something with a lot of

365
00:20:43,200 --> 00:20:46,399
Speaker 1: detailed games like that, uh, that you're not going to

366
00:20:46,480 --> 00:20:49,040
Speaker 1: want that. And yes, we're we're aware of that. Yeah.

367
00:20:49,040 --> 00:20:52,760
Speaker 1: But if you're playing you know, Crush the Castle, yeah,

368
00:20:53,440 --> 00:20:56,359
Speaker 1: uh well, and and it's funny too. I was thinking

369
00:20:56,920 --> 00:21:00,720
Speaker 1: about the most recent release of the Mac os ten,

370
00:21:01,080 --> 00:21:03,359
Speaker 1: which as at this point was snow Leppard and uses

371
00:21:03,800 --> 00:21:06,960
Speaker 1: the Grand Central technology, which sort of coopts your graphics

372
00:21:07,000 --> 00:21:10,320
Speaker 1: processor if it if it isn't busy with something. Uh.

373
00:21:10,520 --> 00:21:16,440
Speaker 1: So it seems like the uh operating system manufacturers and

374
00:21:16,600 --> 00:21:19,240
Speaker 1: the chip manufacturers are both sort of aware, you know what,

375
00:21:19,280 --> 00:21:22,639
Speaker 1: we could probably be using you know, one chip to

376
00:21:22,720 --> 00:21:25,480
Speaker 1: do multiple things, and if you have multiple chips, you

377
00:21:25,480 --> 00:21:28,360
Speaker 1: can have them sort of you know, pinch hit when

378
00:21:28,400 --> 00:21:31,680
Speaker 1: required in other areas. So it seems to make the

379
00:21:32,200 --> 00:21:36,919
Speaker 1: architecture more computer itself architecture, not the the chip architecture

380
00:21:37,000 --> 00:21:40,679
Speaker 1: more flexible because that way, say you have a Sandy

381
00:21:40,720 --> 00:21:44,960
Speaker 1: Bridge chip which has the onboard ability to graphics process

382
00:21:45,119 --> 00:21:47,760
Speaker 1: or process graphics. Sorry. Uh, and you have a GPU

383
00:21:47,880 --> 00:21:50,439
Speaker 1: as well, it seems that that you would have a

384
00:21:50,480 --> 00:21:54,879
Speaker 1: lot of ability for your computer to use those computing cycles,

385
00:21:55,359 --> 00:21:57,480
Speaker 1: you know, in both if it's if the operating system

386
00:21:57,520 --> 00:22:00,520
Speaker 1: is capable of handling or you know, routing those instructions

387
00:22:00,520 --> 00:22:03,080
Speaker 1: to different places like that. Yeah, and you've got some

388
00:22:03,320 --> 00:22:07,560
Speaker 1: GPU manufacturers that are looking into doing CPUs now. So

389
00:22:08,040 --> 00:22:11,320
Speaker 1: you know, while we're seeing Intel kind of push its

390
00:22:11,320 --> 00:22:14,960
Speaker 1: way into the graphics processing unit world just by incorporating

391
00:22:14,960 --> 00:22:17,520
Speaker 1: it into the chips, we're seeing the opposite from the

392
00:22:17,560 --> 00:22:21,000
Speaker 1: graphics processing world as well. So uh yeah, there's a

393
00:22:21,000 --> 00:22:23,480
Speaker 1: lot of competition in this space, and that's one of

394
00:22:23,480 --> 00:22:25,959
Speaker 1: the things. As you make these elements smaller and smaller,

395
00:22:26,000 --> 00:22:28,520
Speaker 1: you can you can really diversify what they can do.

396
00:22:29,000 --> 00:22:30,719
Speaker 1: So let's let's look a little bit into the future,

397
00:22:31,160 --> 00:22:33,160
Speaker 1: all right and look at some of the chips will

398
00:22:33,200 --> 00:22:36,400
Speaker 1: be coming out over the next few years. So following

399
00:22:36,480 --> 00:22:40,200
Speaker 1: sandy Bridge will be ivy Bridge, which of course will

400
00:22:40,200 --> 00:22:43,160
Speaker 1: be another tick. So we're talking about a reduction in size.

401
00:22:43,720 --> 00:22:46,440
Speaker 1: This chip is going to have elements on the twenty

402
00:22:46,480 --> 00:22:51,000
Speaker 1: two nanometer scale. That's tiny. This is we're really getting

403
00:22:51,000 --> 00:22:52,960
Speaker 1: the point where my mind's being blown. Keep in mind

404
00:22:53,000 --> 00:22:55,679
Speaker 1: that the nanometer scale is approximate. It's about, you know,

405
00:22:55,720 --> 00:22:59,280
Speaker 1: ten times the size of the atomic scale. So when

406
00:22:59,280 --> 00:23:02,000
Speaker 1: you get to one nanometer, that's about the size of

407
00:23:02,840 --> 00:23:06,800
Speaker 1: this is. This is oversimplifying, but ten atoms next to

408
00:23:06,800 --> 00:23:11,000
Speaker 1: each other. So yeah, so this is a twenty two

409
00:23:11,080 --> 00:23:13,600
Speaker 1: nanometer scale chip. Uh, and it's built. It will be

410
00:23:13,600 --> 00:23:16,119
Speaker 1: built on the sandy Bridge micro architecture, so it follows

411
00:23:16,160 --> 00:23:19,680
Speaker 1: the same plan. But after Ivy Bridge, what what happens, Well,

412
00:23:20,080 --> 00:23:23,200
Speaker 1: then we're going to get has Well should be the talk.

413
00:23:23,520 --> 00:23:26,000
Speaker 1: So that's gonna be new micro architecture based on this

414
00:23:26,040 --> 00:23:30,840
Speaker 1: twenty two nanometer scale. And then following has Well, we

415
00:23:30,960 --> 00:23:34,280
Speaker 1: get rockwell, which is the next tick, and that's going

416
00:23:34,320 --> 00:23:39,040
Speaker 1: to be at an insanely small sixteen nanometer scale. And

417
00:23:39,160 --> 00:23:42,439
Speaker 1: I remember thinking that I couldn't imagine them breaking the

418
00:23:42,440 --> 00:23:45,359
Speaker 1: forty five nanometer barrier. I didn't think that they were

419
00:23:45,359 --> 00:23:47,000
Speaker 1: going to get down to thirty two. I just didn't

420
00:23:47,000 --> 00:23:50,280
Speaker 1: think it was gonna be physically possible, knowing what I knew,

421
00:23:50,680 --> 00:23:56,360
Speaker 1: which granted, was very little about the physical limitations of

422
00:23:56,359 --> 00:23:59,320
Speaker 1: of the materials they were using. And then you know,

423
00:23:59,520 --> 00:24:03,240
Speaker 1: it's not just that you know the gates are have

424
00:24:03,400 --> 00:24:05,159
Speaker 1: to be thick enough or made out of the right

425
00:24:05,200 --> 00:24:08,560
Speaker 1: material to prevent electron tunneling. It's how do you actually

426
00:24:08,640 --> 00:24:15,000
Speaker 1: design technology that can create things that's small and make

427
00:24:15,040 --> 00:24:17,720
Speaker 1: it efficient enough so you can mass produce it, right,

428
00:24:17,760 --> 00:24:20,159
Speaker 1: I mean, it's not just that it's one thing to

429
00:24:20,200 --> 00:24:22,520
Speaker 1: figure out a way of making an element so small

430
00:24:22,560 --> 00:24:28,400
Speaker 1: that it's that tiny. We've seen companies including IBM, use

431
00:24:28,480 --> 00:24:34,800
Speaker 1: technology to manipulate single atoms and and create pictures with them,

432
00:24:35,400 --> 00:24:39,480
Speaker 1: including the IBM logo. Right, that's a great one, right

433
00:24:39,520 --> 00:24:43,879
Speaker 1: where they've used an electron microscope and they use the

434
00:24:44,000 --> 00:24:47,840
Speaker 1: very tip of it to pull individual atoms and spell

435
00:24:47,880 --> 00:24:50,800
Speaker 1: out IBM Um, there's a great picture of it online.

436
00:24:50,840 --> 00:24:52,760
Speaker 1: Just do a Google search and you'll find it. But

437
00:24:52,880 --> 00:24:55,680
Speaker 1: the you know, the fact that you can do this,

438
00:24:56,080 --> 00:24:58,280
Speaker 1: that's not me. That doesn't mean that it's efficient or

439
00:24:58,320 --> 00:25:00,520
Speaker 1: that you could do it on a mask. Lees So

440
00:25:00,920 --> 00:25:02,960
Speaker 1: it's phenomenal to me. The note did they find the

441
00:25:02,960 --> 00:25:06,040
Speaker 1: technology to to build things at the scale, but do

442
00:25:06,119 --> 00:25:08,159
Speaker 1: it in an efficient way where you can actually make chips.

443
00:25:09,600 --> 00:25:14,200
Speaker 1: I'm just you know, it's it's it's very impressive, and

444
00:25:14,200 --> 00:25:18,880
Speaker 1: it's I'm just wondering how long Mr Moore's Law will

445
00:25:18,880 --> 00:25:21,119
Speaker 1: continue to be a law. I mean, they're they're trying

446
00:25:21,119 --> 00:25:25,480
Speaker 1: their hardness. Yeah, And it's it's funny because you said

447
00:25:25,480 --> 00:25:27,560
Speaker 1: that people have been predicting the end of Moore's law.

448
00:25:27,600 --> 00:25:29,480
Speaker 1: People have been predicting the end of Moore's Law since

449
00:25:29,520 --> 00:25:31,960
Speaker 1: like the eighties. Well, and as you point out in

450
00:25:31,960 --> 00:25:35,080
Speaker 1: your article, it would have gone by the wayside if

451
00:25:35,119 --> 00:25:39,159
Speaker 1: they weren't actively trying to make it happen. Now that

452
00:25:39,240 --> 00:25:44,600
Speaker 1: the companies it's become, it's almost like a self fulfilling prophecy, right, Uh,

453
00:25:44,720 --> 00:25:48,720
Speaker 1: No one wants to admit that Moore's Law is has

454
00:25:48,760 --> 00:25:51,560
Speaker 1: reached its end. Everyone wants to be able to keep

455
00:25:51,600 --> 00:25:55,280
Speaker 1: pushing that innovation. For one thing, it is a motivator

456
00:25:55,600 --> 00:25:59,159
Speaker 1: to innovate, and we want innovation. If we are we

457
00:25:59,240 --> 00:26:02,200
Speaker 1: become unmodi vata, demotivated whatever you want to call it,

458
00:26:02,560 --> 00:26:05,760
Speaker 1: to innovate, then we just stagnate. You know, we're gonna

459
00:26:05,800 --> 00:26:07,919
Speaker 1: be like, well, we've gotten as far as we can

460
00:26:07,960 --> 00:26:10,520
Speaker 1: go and that's it. No, No, it's better to sit

461
00:26:10,520 --> 00:26:12,399
Speaker 1: there and say no, no, there's got to be a

462
00:26:12,400 --> 00:26:15,000
Speaker 1: better way to make this even faster. Uh. And that

463
00:26:15,040 --> 00:26:19,000
Speaker 1: way we keep moving forward, we progress, we advance, and um,

464
00:26:19,800 --> 00:26:21,840
Speaker 1: that's kind of what Moore's laws helped us do. It's

465
00:26:21,880 --> 00:26:25,560
Speaker 1: really pushed us to innovate so that we can keep

466
00:26:25,680 --> 00:26:29,280
Speaker 1: up with this observation. And uh yeah, we'll probably reach

467
00:26:29,320 --> 00:26:32,280
Speaker 1: a day at some point where at least the approach

468
00:26:32,280 --> 00:26:36,200
Speaker 1: we're using now will no longer be feasible. In order

469
00:26:36,240 --> 00:26:39,679
Speaker 1: to maintain More's law, we may have to have a

470
00:26:39,760 --> 00:26:44,960
Speaker 1: complete shift in what it means to to build a computer, right,

471
00:26:45,400 --> 00:26:48,359
Speaker 1: and may mean that we go through quantum computers which

472
00:26:48,520 --> 00:26:52,240
Speaker 1: are still unproven, or we may have biological computers that

473
00:26:52,840 --> 00:26:58,199
Speaker 1: uh that that utilized DNA as a computing technology. You know,

474
00:26:58,240 --> 00:27:05,000
Speaker 1: once to get a taste at DNA. Yeah, but I

475
00:27:05,280 --> 00:27:08,680
Speaker 1: do think that until really got onto something when they

476
00:27:08,760 --> 00:27:12,240
Speaker 1: when they developed the strategy though, because it seems like

477
00:27:12,520 --> 00:27:18,120
Speaker 1: by concentrating on either the size or the architecture and

478
00:27:18,359 --> 00:27:20,960
Speaker 1: building around that, it gives them something to a point

479
00:27:20,960 --> 00:27:23,719
Speaker 1: from which they can start and they can build a

480
00:27:23,720 --> 00:27:29,160
Speaker 1: new chip without having to necessarily building everything from scratch.

481
00:27:29,640 --> 00:27:32,199
Speaker 1: I think it probably cuts down I'm saying probably. I

482
00:27:32,200 --> 00:27:36,360
Speaker 1: don't know. I'm an outsider here, you know, but uh,

483
00:27:36,560 --> 00:27:38,399
Speaker 1: I would imagine it. It cuts down on the amount

484
00:27:38,440 --> 00:27:41,639
Speaker 1: of time they have to spend preparing, uh, you know,

485
00:27:41,720 --> 00:27:43,600
Speaker 1: a design for the new chip, because they have an

486
00:27:43,640 --> 00:27:46,639
Speaker 1: idea of where they want to start. Um, it cuts

487
00:27:46,680 --> 00:27:49,800
Speaker 1: down on the possibility of mistakes, which at this point,

488
00:27:49,880 --> 00:27:54,480
Speaker 1: at this scale, a mistake would be a huge hit

489
00:27:54,520 --> 00:27:59,360
Speaker 1: to reputation. I mean, yes, Intel is the giant chip manufacturer.

490
00:27:59,400 --> 00:28:02,760
Speaker 1: They're They're storied in their past. Um, they have been

491
00:28:03,560 --> 00:28:07,240
Speaker 1: sued for for monopoly. Yeah, they are definitely the dominant

492
00:28:07,240 --> 00:28:09,840
Speaker 1: player in the market. And we will, I'm sure get

493
00:28:09,920 --> 00:28:11,560
Speaker 1: someone who wants us to do the A M. D

494
00:28:11,720 --> 00:28:14,159
Speaker 1: story or we have had people asked us to do

495
00:28:14,240 --> 00:28:16,960
Speaker 1: A and D. And we will. We just we figured

496
00:28:17,000 --> 00:28:21,240
Speaker 1: this would you know, when you're talking about micro processors,

497
00:28:21,280 --> 00:28:25,800
Speaker 1: to not start with until seems ridiculous, just because it

498
00:28:25,920 --> 00:28:29,560
Speaker 1: is such a huge player in that market. Yes it is.

499
00:28:29,960 --> 00:28:33,320
Speaker 1: And that's the thing, even at its size. Uh. The

500
00:28:33,320 --> 00:28:39,160
Speaker 1: the the semi scandal was sandy Bridge um was already

501
00:28:39,360 --> 00:28:41,760
Speaker 1: starting to have an effect on intelent. You could see

502
00:28:41,760 --> 00:28:44,680
Speaker 1: in the way that that the company reacted to the

503
00:28:44,720 --> 00:28:48,680
Speaker 1: problem that it discovered that probably you know, they pulled

504
00:28:48,680 --> 00:28:51,280
Speaker 1: those chips and stopped making the ones that they were

505
00:28:51,280 --> 00:28:53,280
Speaker 1: making with that architecture. Yeah, do you want to just

506
00:28:53,360 --> 00:28:56,400
Speaker 1: really quickly, Yeah, that problem had to do with other

507
00:28:56,800 --> 00:28:59,480
Speaker 1: processor chips, not the not the actual sandy Bridge chip,

508
00:28:59,600 --> 00:29:02,640
Speaker 1: but other hips on the motherboard that Intel was shipping

509
00:29:02,640 --> 00:29:06,040
Speaker 1: out that would support the sandy Bridge chip. And uh,

510
00:29:06,320 --> 00:29:09,600
Speaker 1: the problem was discovered that over time, uh, and we're

511
00:29:09,600 --> 00:29:13,520
Speaker 1: talking about a fairly short time frame, the the performance

512
00:29:13,760 --> 00:29:18,640
Speaker 1: of those chips would degrade fairly rapidly. And that you know,

513
00:29:18,680 --> 00:29:21,200
Speaker 1: for most people it wouldn't really be a huge deal

514
00:29:21,240 --> 00:29:24,240
Speaker 1: because most people are not really pushing their machine to

515
00:29:24,480 --> 00:29:26,800
Speaker 1: its limits. But for the people who were pushing their

516
00:29:26,840 --> 00:29:28,800
Speaker 1: machines too, as far as they're gonna go, Like the

517
00:29:28,920 --> 00:29:32,840
Speaker 1: video gamers and the media editors out there, they would

518
00:29:33,360 --> 00:29:38,040
Speaker 1: potentially notice a decrease in performance much more quickly than

519
00:29:38,080 --> 00:29:41,800
Speaker 1: you would have anticipate for a typical computer. And so

520
00:29:42,040 --> 00:29:45,080
Speaker 1: that meant that a lot of manufacturers and a lot

521
00:29:45,120 --> 00:29:47,960
Speaker 1: of you know, a lot of computer retailers began to

522
00:29:48,040 --> 00:29:52,600
Speaker 1: hold off on incorporating sandy Bridge motherboards into their systems

523
00:29:52,960 --> 00:29:56,800
Speaker 1: until this was addressed, until these chips had been fixed

524
00:29:56,840 --> 00:30:00,200
Speaker 1: and new motherboards were being shipped out. So it was

525
00:30:00,280 --> 00:30:03,000
Speaker 1: kind of a black eye for Intel, But ultimately it

526
00:30:03,080 --> 00:30:07,400
Speaker 1: was a problem not with the actual sandy Bridge microprocessor.

527
00:30:08,040 --> 00:30:10,960
Speaker 1: UM but I'm sorry, go ahead, I'm sorry. I was

528
00:30:10,960 --> 00:30:13,400
Speaker 1: just gonna say though that that that sort of illustrates

529
00:30:13,600 --> 00:30:17,880
Speaker 1: my point though that um, this will help I think

530
00:30:17,920 --> 00:30:20,360
Speaker 1: that this will help Intel cut down on the possibility

531
00:30:20,400 --> 00:30:24,480
Speaker 1: that there will be um architecture problems with these chips.

532
00:30:24,520 --> 00:30:26,320
Speaker 1: Because they have a place from which to start, they

533
00:30:26,320 --> 00:30:30,000
Speaker 1: can go ahead and get moving on it. Obviously, with

534
00:30:30,080 --> 00:30:33,920
Speaker 1: the roadmap set out years in advance, the company already

535
00:30:33,960 --> 00:30:36,280
Speaker 1: has an idea of where it's going, so it can

536
00:30:36,320 --> 00:30:38,880
Speaker 1: be it can be working on the next chip even

537
00:30:38,920 --> 00:30:42,000
Speaker 1: before this chip is released. Actually, yeah, they're working on

538
00:30:42,040 --> 00:30:44,000
Speaker 1: like the next three chips. By the time a chip

539
00:30:44,040 --> 00:30:46,320
Speaker 1: has come out, you can get your it's it's a

540
00:30:46,320 --> 00:30:49,240
Speaker 1: guarantee that they're working on at least the next two,

541
00:30:49,280 --> 00:30:52,600
Speaker 1: if not three generations. And that's just impressive that they

542
00:30:52,800 --> 00:30:56,360
Speaker 1: that the company is is that efficient and programmed out

543
00:30:56,360 --> 00:30:58,280
Speaker 1: that it knows what it's doing and it can move

544
00:30:58,320 --> 00:31:01,080
Speaker 1: ahead with confidence. And in case you're curious as to

545
00:31:01,120 --> 00:31:04,960
Speaker 1: what strategy they used before TikTok, uh, it was that

546
00:31:05,040 --> 00:31:08,840
Speaker 1: they were concentrating on reducing the size of their transistors

547
00:31:08,880 --> 00:31:10,960
Speaker 1: every year. So each year they were trying to to

548
00:31:11,080 --> 00:31:13,520
Speaker 1: double the number of transistors on a chip, but they

549
00:31:13,520 --> 00:31:16,360
Speaker 1: were only looking at the micro architecture every two to

550
00:31:16,560 --> 00:31:20,840
Speaker 1: four to sometimes six years, so they were only adjusting

551
00:31:20,880 --> 00:31:24,400
Speaker 1: the efficiency of the chips, uh, you know, sporadically, while

552
00:31:24,560 --> 00:31:27,640
Speaker 1: whereas they were reducing the size year over year over year.

553
00:31:27,640 --> 00:31:29,840
Speaker 1: And that's when they realized that, well, this is not

554
00:31:29,880 --> 00:31:33,960
Speaker 1: really sustainable if we want to truly stay twice as power,

555
00:31:34,320 --> 00:31:37,080
Speaker 1: keep a chip twice as powerful two years out. Uh

556
00:31:37,200 --> 00:31:41,480
Speaker 1: So they readjusted their strategy and that's when they adopted TikTok.

557
00:31:42,040 --> 00:31:44,320
Speaker 1: So it seems to be working for them. So good

558
00:31:44,400 --> 00:31:49,120
Speaker 1: job Intel. I'm sure other companies use similar strategies. Uh,

559
00:31:49,160 --> 00:31:52,320
Speaker 1: this was just one that has actually become fairly famous,

560
00:31:52,320 --> 00:31:54,440
Speaker 1: at least in the technology world, so we wanted to

561
00:31:54,440 --> 00:31:57,160
Speaker 1: to tackle it. Oh, that kind of wraps up this

562
00:31:57,200 --> 00:32:01,320
Speaker 1: discussion about the TikTok strategy. So if you have any questions,

563
00:32:01,440 --> 00:32:05,480
Speaker 1: or you want to suggest your own topic of favorite topic,

564
00:32:05,520 --> 00:32:07,680
Speaker 1: if it's A and D, or perhaps it has nothing

565
00:32:07,720 --> 00:32:09,160
Speaker 1: to do with chips at all. You just want to

566
00:32:09,160 --> 00:32:11,719
Speaker 1: hear us talk about video game music more, let us

567
00:32:11,760 --> 00:32:16,719
Speaker 1: know tunes, Yes, you can let us know on Twitter

568
00:32:16,840 --> 00:32:20,200
Speaker 1: or Facebook are handled. There is text stuff HSW or

569
00:32:20,240 --> 00:32:23,040
Speaker 1: you can send us an email. That address is tech

570
00:32:23,160 --> 00:32:26,000
Speaker 1: stuff at how stuff works dot com and Chris and

571
00:32:26,040 --> 00:32:30,520
Speaker 1: I will talk you again really soon for more on

572
00:32:30,560 --> 00:32:32,880
Speaker 1: this and thousands of other topics. Is it how stuff

573
00:32:32,880 --> 00:32:35,720
Speaker 1: works dot com. To learn more about the podcast, click

574
00:32:35,760 --> 00:32:38,080
Speaker 1: on the podcast icon in the upper right corner of

575
00:32:38,120 --> 00:32:42,160
Speaker 1: our homepage. The How Stuff Works iPhone app has arrived.

576
00:32:42,280 --> 00:32:49,560
Speaker 1: Download it today on iTunes, brought to you by the

577
00:32:49,560 --> 00:32:52,920
Speaker 1: reinvented two thousand twelve camera. It's ready, are you