1
00:00:00,320 --> 00:00:02,880
Speaker 1: Brought to you by the reinvented two thousand twelve camera.

2
00:00:03,200 --> 00:00:08,920
Speaker 1: It's ready. Are you get in touch with technology with

3
00:00:09,080 --> 00:00:17,799
Speaker 1: text stuff from how stuff works dot com. Hi there, everybody,

4
00:00:17,800 --> 00:00:19,880
Speaker 1: and welcome to tech stuff. My name is Chris Poulette

5
00:00:19,880 --> 00:00:21,680
Speaker 1: and I am the tech editor here at how stuff

6
00:00:21,720 --> 00:00:24,720
Speaker 1: works dot Com. Sitting across from me, as he always does,

7
00:00:25,040 --> 00:00:28,400
Speaker 1: is senior writer Jonathan Strickland. Hey there, guys, we're having

8
00:00:28,440 --> 00:00:34,880
Speaker 1: a bit of an identity crisis today. Who is Yeah, so, um,

9
00:00:34,960 --> 00:00:38,480
Speaker 1: we are going to talk about some some very strange

10
00:00:38,600 --> 00:00:42,360
Speaker 1: combinations of things actually, uh one that you might not

11
00:00:42,440 --> 00:00:47,159
Speaker 1: expect gaming and super computing, super super computing. Yeah. This

12
00:00:47,200 --> 00:00:53,440
Speaker 1: actually comes a courtesy of a little listener text message Liz,

13
00:00:54,280 --> 00:00:58,480
Speaker 1: queue something up, will you? This text message comes from

14
00:00:58,520 --> 00:01:01,320
Speaker 1: Tim who says, what's up with the U. S. Army

15
00:01:01,360 --> 00:01:07,840
Speaker 1: buying p S three's for a parallel supercomputer cloud system? Question? Mark? Well, Tim,

16
00:01:07,880 --> 00:01:09,520
Speaker 1: first of all, it's not the U. S. Army, it's

17
00:01:09,520 --> 00:01:12,119
Speaker 1: the US Air Force. And it wasn't twenty two hundred

18
00:01:12,200 --> 00:01:16,080
Speaker 1: p S three's It was actually twenty five hundred. But uh, yeah,

19
00:01:16,160 --> 00:01:18,760
Speaker 1: we get the point of the question here, which is

20
00:01:19,160 --> 00:01:22,480
Speaker 1: what is up with buying p S three's and how

21
00:01:22,520 --> 00:01:26,160
Speaker 1: does that make a supercomputer? It all boils down to

22
00:01:26,400 --> 00:01:31,600
Speaker 1: the PlayStation three's processor, which is a cell processor. Yes,

23
00:01:31,640 --> 00:01:34,960
Speaker 1: this is a custom chip. UM actually spent quite a

24
00:01:35,040 --> 00:01:37,839
Speaker 1: lot of money in in development and it was under

25
00:01:37,840 --> 00:01:41,720
Speaker 1: development by a joint venture of three companies. In addition

26
00:01:41,760 --> 00:01:45,240
Speaker 1: to Sony, IBM and Toshiba were both involved with this

27
00:01:45,520 --> 00:01:50,400
Speaker 1: with this chips manufacturer. And uh, it's a pretty interesting device.

28
00:01:50,440 --> 00:01:54,040
Speaker 1: It's what gives the p S three it's mph and uh,

29
00:01:54,200 --> 00:01:56,600
Speaker 1: you know, allows it to do what it can do.

30
00:01:56,640 --> 00:02:00,400
Speaker 1: It can crunch lots and lots of numbers, right, and

31
00:02:00,480 --> 00:02:03,480
Speaker 1: so let's uh, let's just do a quick overview of

32
00:02:03,480 --> 00:02:06,160
Speaker 1: what a processor does and then we talk about why

33
00:02:06,240 --> 00:02:09,320
Speaker 1: the the cell processor is slightly different. So, in general,

34
00:02:09,720 --> 00:02:14,000
Speaker 1: a processor's job is to take a uh information, So

35
00:02:14,000 --> 00:02:18,520
Speaker 1: it's taking data and executing any sort of function upon

36
00:02:18,560 --> 00:02:21,320
Speaker 1: that data to get a result. So there are two

37
00:02:21,360 --> 00:02:24,440
Speaker 1: different lines of data going into a processor at any

38
00:02:24,520 --> 00:02:28,040
Speaker 1: one time. There's the source information and then there are

39
00:02:28,200 --> 00:02:32,200
Speaker 1: there the the the functions, the various code that comes

40
00:02:32,200 --> 00:02:35,320
Speaker 1: in and says, all right, add these numbers together, or

41
00:02:35,400 --> 00:02:38,520
Speaker 1: divide this number, or find the uh the largest two

42
00:02:38,639 --> 00:02:42,240
Speaker 1: prime integers of this number. Um. And so that's what

43
00:02:42,280 --> 00:02:46,119
Speaker 1: a processor does, is it It It executes the command

44
00:02:46,360 --> 00:02:49,440
Speaker 1: upon the data and then gives the result. Uh. So

45
00:02:49,520 --> 00:02:52,920
Speaker 1: that's the basic function of a processor. Now the cell

46
00:02:53,000 --> 00:02:56,560
Speaker 1: processor is a little different. It's it's kind of a

47
00:02:56,560 --> 00:03:00,760
Speaker 1: specialized processor and uh it has a cup of different

48
00:03:00,800 --> 00:03:04,080
Speaker 1: elements to it that allow it to behave in a

49
00:03:04,120 --> 00:03:06,919
Speaker 1: slightly different way. Um. In a way, it actually kind

50
00:03:06,919 --> 00:03:11,680
Speaker 1: of acts like multiple computers working together on a single problem.

51
00:03:11,720 --> 00:03:16,200
Speaker 1: It's just it's all been consolidated onto a single processor chip. Now,

52
00:03:16,200 --> 00:03:20,000
Speaker 1: this processor chip has a power processor element or p

53
00:03:20,000 --> 00:03:24,080
Speaker 1: p E. The power processor elements job is kind of

54
00:03:24,120 --> 00:03:27,120
Speaker 1: like a manager at uh in a in an office,

55
00:03:27,200 --> 00:03:29,680
Speaker 1: all right. The manager's job is they take in an

56
00:03:29,720 --> 00:03:32,800
Speaker 1: assignment and then they give that they delegate that assignment

57
00:03:32,840 --> 00:03:37,280
Speaker 1: to an employee complete all right, and then uh that

58
00:03:37,360 --> 00:03:42,520
Speaker 1: employee would be represented by a synergistic processor element or SPE.

59
00:03:43,040 --> 00:03:46,880
Speaker 1: Now the PS three cell processors have a one PPE

60
00:03:47,040 --> 00:03:52,800
Speaker 1: and seven SPEs. So the PPE takes the problem, divides

61
00:03:52,840 --> 00:03:55,960
Speaker 1: it up into individual tasks, and gives each task to

62
00:03:56,080 --> 00:03:59,680
Speaker 1: an SPE. Each SPE is able to work as its

63
00:03:59,720 --> 00:04:04,040
Speaker 1: own processor, with all of them working together in concert.

64
00:04:04,280 --> 00:04:08,120
Speaker 1: You have a very powerful chip because they are all.

65
00:04:08,240 --> 00:04:10,680
Speaker 1: That means all seven of these can work towards solving

66
00:04:10,920 --> 00:04:15,360
Speaker 1: each individual problem, which ultimately will solve a much larger problem.

67
00:04:15,840 --> 00:04:18,320
Speaker 1: Uh in in a fraction of the time of what

68
00:04:18,440 --> 00:04:24,640
Speaker 1: would take a similarly powerful single core processor. Okay, So

69
00:04:25,160 --> 00:04:27,320
Speaker 1: the other way that the analogy I used I use

70
00:04:27,400 --> 00:04:29,680
Speaker 1: this on an episode of Tech Stuff Live not too

71
00:04:29,720 --> 00:04:34,320
Speaker 1: long ago. But imagine that you have two rooms. Inside

72
00:04:34,360 --> 00:04:38,360
Speaker 1: one room, you've got a mathematician genius, someone who can

73
00:04:38,400 --> 00:04:41,599
Speaker 1: take a math problem and solve it. Uh It doesn't

74
00:04:41,600 --> 00:04:44,479
Speaker 1: matter how complex you make it. Eventually, sooner or later,

75
00:04:44,600 --> 00:04:48,400
Speaker 1: this genius will figure it out. In a second room,

76
00:04:48,680 --> 00:04:51,320
Speaker 1: you have a room full of smart people who are

77
00:04:51,320 --> 00:04:53,840
Speaker 1: good at math. But they are not mathematicians. They are

78
00:04:53,880 --> 00:04:56,880
Speaker 1: not geniuses. They're just they're diligent and they work hard

79
00:04:56,920 --> 00:04:58,680
Speaker 1: and they can get the right answer. It just usually

80
00:04:58,720 --> 00:05:01,480
Speaker 1: takes them a little bit longer than it would the

81
00:05:01,640 --> 00:05:04,600
Speaker 1: genius in the other room. Now, let's say you've got

82
00:05:04,600 --> 00:05:07,679
Speaker 1: a math problem. The math problem is a really really

83
00:05:07,760 --> 00:05:10,680
Speaker 1: long problem that can be broken down into several steps,

84
00:05:11,160 --> 00:05:13,440
Speaker 1: And you give the math problem to the genius, and

85
00:05:13,480 --> 00:05:15,200
Speaker 1: you give the math problem to the room full of

86
00:05:15,240 --> 00:05:18,960
Speaker 1: smart people. The smart people can break that problem up

87
00:05:19,000 --> 00:05:22,400
Speaker 1: into lots of smaller problems, and each person is working

88
00:05:22,440 --> 00:05:25,040
Speaker 1: on a section of it, and then ultimately they can

89
00:05:25,040 --> 00:05:28,000
Speaker 1: put all their answers together and solve it faster than

90
00:05:28,040 --> 00:05:33,080
Speaker 1: the genius can. However, if you have a big problem

91
00:05:33,120 --> 00:05:36,520
Speaker 1: that cannot be broken down, the genius is going to

92
00:05:36,560 --> 00:05:39,400
Speaker 1: be able to work through that problem faster because no

93
00:05:39,440 --> 00:05:42,159
Speaker 1: matter how you, how how much the smart people in

94
00:05:42,160 --> 00:05:44,279
Speaker 1: the other room look at it, they can't break that

95
00:05:44,360 --> 00:05:47,440
Speaker 1: problem into smaller pieces to solve each on their own.

96
00:05:47,480 --> 00:05:49,200
Speaker 1: So they just have to work on it and get

97
00:05:49,240 --> 00:05:51,680
Speaker 1: through the problem. That they're not gonna be any faster

98
00:05:52,200 --> 00:05:54,840
Speaker 1: than the other guy in the other room. So that's

99
00:05:54,880 --> 00:05:58,560
Speaker 1: that's kind of a comparison of a cell processor versus

100
00:05:58,680 --> 00:06:04,679
Speaker 1: a very powerful, uh regular microprocessor. Uh. It all depends

101
00:06:04,720 --> 00:06:07,160
Speaker 1: on the kind of information you feed into it. If

102
00:06:07,200 --> 00:06:09,240
Speaker 1: it's the kind of information that can be broken down

103
00:06:09,240 --> 00:06:12,680
Speaker 1: into smaller problems, you're set. You're going to be able

104
00:06:12,680 --> 00:06:16,039
Speaker 1: to solve that in in no time flat, relatively speaking,

105
00:06:16,560 --> 00:06:19,520
Speaker 1: because you can do something that's called parallel processing. All

106
00:06:19,560 --> 00:06:26,680
Speaker 1: the processors are working in parallel. Awesome. Wow, I'm I'm

107
00:06:26,680 --> 00:06:28,560
Speaker 1: already wiped out. We haven't even gotten to the part

108
00:06:28,560 --> 00:06:31,560
Speaker 1: about the US Air Force. Now, the Air Force is

109
00:06:31,560 --> 00:06:34,600
Speaker 1: not actually they're not. It's not the first organization to

110
00:06:34,960 --> 00:06:38,320
Speaker 1: use PS three's as a supercomputer. Yeah. As a matter

111
00:06:38,360 --> 00:06:42,160
Speaker 1: of fact, um Sony was trying to illustrate this as

112
00:06:42,200 --> 00:06:45,560
Speaker 1: far back as two thousand six. Um they have been

113
00:06:46,040 --> 00:06:49,839
Speaker 1: uh promoting the idea that the cell processor is capable

114
00:06:49,880 --> 00:06:54,840
Speaker 1: of handling advanced computing UM basically as an if you will,

115
00:06:54,880 --> 00:06:57,320
Speaker 1: as sort of a marketing effort, although sort of an

116
00:06:57,400 --> 00:07:01,560
Speaker 1: unusual one by you know, I guess most aiming console standards,

117
00:07:02,320 --> 00:07:04,760
Speaker 1: but it's it's basically a marketing effort to show, look,

118
00:07:04,880 --> 00:07:07,719
Speaker 1: this is this processor is so advanced it can handle

119
00:07:08,240 --> 00:07:11,120
Speaker 1: serious computing stuff, not just handle your games. It's not

120
00:07:11,240 --> 00:07:14,360
Speaker 1: just a Blu ray player. It really is a super

121
00:07:14,840 --> 00:07:18,840
Speaker 1: heavy duty monster of a chip. And um, they actually

122
00:07:18,880 --> 00:07:23,600
Speaker 1: have been very uh cooperative with people who are are

123
00:07:23,640 --> 00:07:26,120
Speaker 1: trying to build them into and network them together to

124
00:07:26,200 --> 00:07:31,640
Speaker 1: make a super computer as um as certain uh researcher

125
00:07:31,720 --> 00:07:34,840
Speaker 1: did up in the Northeastern United States. Are you familiar

126
00:07:34,840 --> 00:07:38,640
Speaker 1: with this person? This is the astrophysicist Dr Garov Kana. Yes.

127
00:07:38,720 --> 00:07:43,680
Speaker 1: Dr Kanna, who who networked eight PlayStation three's together originally

128
00:07:43,720 --> 00:07:47,720
Speaker 1: in order to solve problems relating to gravitational fields and

129
00:07:47,720 --> 00:07:49,840
Speaker 1: things of that nature. Yeah, he called it his gravity

130
00:07:49,840 --> 00:07:54,000
Speaker 1: grid according to the research that I used. UM. And basically,

131
00:07:54,120 --> 00:07:56,960
Speaker 1: what he wanted to do was to find out, uh,

132
00:07:57,000 --> 00:08:02,000
Speaker 1: what would happen if a supermassive black ol swallows a star,

133
00:08:02,440 --> 00:08:05,080
Speaker 1: And so he wanted he really needed massive amounts of

134
00:08:05,320 --> 00:08:08,040
Speaker 1: computing ability to do that, and he had been spending

135
00:08:08,040 --> 00:08:10,480
Speaker 1: money that he he was receiving as grants from the

136
00:08:10,560 --> 00:08:15,440
Speaker 1: National Science Foundation UM to distribute his project out to supercomputers,

137
00:08:15,440 --> 00:08:18,080
Speaker 1: but that was costing about five thousand dollars a pop.

138
00:08:18,960 --> 00:08:21,800
Speaker 1: And UM, he realized that he could do this more

139
00:08:21,840 --> 00:08:26,400
Speaker 1: affordably with the processors inside a PlayStation three if he

140
00:08:26,440 --> 00:08:30,680
Speaker 1: networked them together and also installed Lennox. That's an important

141
00:08:30,720 --> 00:08:34,480
Speaker 1: element of this is that they Sony actually allows people

142
00:08:34,600 --> 00:08:38,400
Speaker 1: to install and an open source operating system onto the

143
00:08:38,440 --> 00:08:42,040
Speaker 1: p S three if they want to. It's it's actually allowed. Um,

144
00:08:42,080 --> 00:08:45,960
Speaker 1: there's no barrier in place, which I gotta say is

145
00:08:46,080 --> 00:08:48,920
Speaker 1: really weird for Sony. I mean, Sony is not known

146
00:08:48,960 --> 00:08:53,240
Speaker 1: for being the most open platform kind of company, but um, yeah,

147
00:08:53,320 --> 00:08:58,880
Speaker 1: by installing lenox and creating a cluster of these PS three's.

148
00:08:59,040 --> 00:09:02,800
Speaker 1: You have this very power full processing machine. What that

149
00:09:02,800 --> 00:09:06,760
Speaker 1: that is? Uh, that's one illustration of how behind this

150
00:09:06,800 --> 00:09:10,599
Speaker 1: project Sony is. But I haven't a more even uh

151
00:09:10,960 --> 00:09:15,839
Speaker 1: vivid thing because see Dr Kano figured it would take

152
00:09:15,840 --> 00:09:19,160
Speaker 1: about to buy the PlayStation three and this was you know,

153
00:09:19,160 --> 00:09:21,080
Speaker 1: a couple of years ago. They were still more they

154
00:09:21,120 --> 00:09:25,480
Speaker 1: were more expensive than since then. Um, but he was

155
00:09:25,559 --> 00:09:27,599
Speaker 1: able to talk Sony. He explained to Sony what he

156
00:09:27,640 --> 00:09:30,160
Speaker 1: wanted to do, and Sony Sony donated the eight machines

157
00:09:30,200 --> 00:09:33,120
Speaker 1: that he needed to create his gravity grid. So they

158
00:09:33,120 --> 00:09:38,080
Speaker 1: were absolutely saying yeah and are absolutely so, um, you

159
00:09:38,080 --> 00:09:40,160
Speaker 1: know they were. They were totally behind it, and it

160
00:09:40,360 --> 00:09:43,280
Speaker 1: made some noise at the time, which was good because

161
00:09:43,960 --> 00:09:46,400
Speaker 1: at the time Sony was having some difficulties selling them

162
00:09:46,400 --> 00:09:50,680
Speaker 1: because they were very expensive. And so when you know,

163
00:09:50,800 --> 00:09:54,400
Speaker 1: I was talking earlier about the the one PPE handing

164
00:09:54,400 --> 00:09:57,880
Speaker 1: out the the the individual jobs to the sp s.

165
00:09:58,440 --> 00:10:00,240
Speaker 1: The same thing can be said. If you Clu suster

166
00:10:00,400 --> 00:10:03,640
Speaker 1: these these devices together, you just you become you create

167
00:10:03,679 --> 00:10:06,080
Speaker 1: an even larger group. So it's like a group of

168
00:10:06,120 --> 00:10:10,280
Speaker 1: managers and a group of employees and so uh, you

169
00:10:10,640 --> 00:10:15,080
Speaker 1: effectively increase the processing ability of this these machines, you know,

170
00:10:15,280 --> 00:10:20,120
Speaker 1: by however many you managed to network together, and you

171
00:10:20,120 --> 00:10:22,600
Speaker 1: can network lots of these together. You don't. It doesn't

172
00:10:22,600 --> 00:10:24,960
Speaker 1: have to just be eight. That was the size of

173
00:10:24,960 --> 00:10:28,240
Speaker 1: the cluster that um Dr Khanna decided to to go with.

174
00:10:28,760 --> 00:10:32,160
Speaker 1: But uh, and it admit the processing needs that you

175
00:10:32,200 --> 00:10:35,080
Speaker 1: had at the time. The US Air Force is doing

176
00:10:35,120 --> 00:10:39,000
Speaker 1: a similar project. You know, we were talking abouts threes.

177
00:10:39,040 --> 00:10:42,840
Speaker 1: It's actually closer to undred. They already had over three

178
00:10:42,880 --> 00:10:48,800
Speaker 1: hundred uh already set aside for supercomputing. But it's the

179
00:10:48,840 --> 00:10:52,280
Speaker 1: same sort of concept is linking these together. You install

180
00:10:52,360 --> 00:10:56,360
Speaker 1: Linux on them, you create clusters, and using these clusters

181
00:10:56,360 --> 00:11:01,960
Speaker 1: you can solve really complex problems and and and parallel problems. Again,

182
00:11:02,160 --> 00:11:04,959
Speaker 1: once again we have to differentiate. Um has to be

183
00:11:05,040 --> 00:11:08,640
Speaker 1: a complex parallel problem for these clusters to actually be effective.

184
00:11:09,160 --> 00:11:12,400
Speaker 1: This is very similar in a way to what quantum

185
00:11:12,400 --> 00:11:16,280
Speaker 1: computing promises us. You may remember we talked about quantum

186
00:11:16,280 --> 00:11:19,000
Speaker 1: computing in a previous podcast. Actually we may have talked

187
00:11:19,040 --> 00:11:23,079
Speaker 1: about a couple of times. But quantum computers use cubits

188
00:11:23,280 --> 00:11:26,160
Speaker 1: rather than just regular bits, and a cubit can be

189
00:11:26,240 --> 00:11:29,079
Speaker 1: a zero or a one, or theoretically anything in between

190
00:11:29,520 --> 00:11:31,800
Speaker 1: at the same time, so you are able to do

191
00:11:31,920 --> 00:11:35,800
Speaker 1: multiple You're able to execute multiple processes all at once

192
00:11:35,920 --> 00:11:40,640
Speaker 1: because these cubits are fulfilling all possibilities at the same time.

193
00:11:41,040 --> 00:11:44,520
Speaker 1: The difference between quantum computers and networking a bunch of

194
00:11:44,520 --> 00:11:47,240
Speaker 1: p S three's together is the results from a quantum

195
00:11:47,280 --> 00:11:51,600
Speaker 1: computer are going to be generated in probabilities, so you

196
00:11:51,679 --> 00:11:56,400
Speaker 1: get you'll get maybe a range of potential answers with

197
00:11:56,480 --> 00:12:00,200
Speaker 1: a percentage of how likely each one is correct, so

198
00:12:00,280 --> 00:12:03,960
Speaker 1: you'll never be certain that the answer that has the

199
00:12:04,000 --> 00:12:06,720
Speaker 1: highest percentage, you know, the highest probability of being correct,

200
00:12:06,880 --> 00:12:09,480
Speaker 1: is in fact the really correct one, which is kind

201
00:12:09,480 --> 00:12:15,959
Speaker 1: of weird, but that's quantum for you. Yeah, probably, yeah. Um.

202
00:12:16,000 --> 00:12:18,319
Speaker 1: You know it's funny too because, just as an aside,

203
00:12:18,480 --> 00:12:22,640
Speaker 1: every time somebody says cubits, I think of the the

204
00:12:22,760 --> 00:12:27,079
Speaker 1: length measurements, not which is cub i T you're thinking

205
00:12:27,080 --> 00:12:30,120
Speaker 1: about building an arc, Well, yeah, everybody thinks of the arc,

206
00:12:30,360 --> 00:12:32,760
Speaker 1: probably because of the Bill Cosby bit. What the hell

207
00:12:32,800 --> 00:12:37,440
Speaker 1: is a cubit? Yeah, but now these are q bits, Yeah,

208
00:12:37,760 --> 00:12:40,920
Speaker 1: which is an entirely different. Animals say a distant cousin

209
00:12:40,960 --> 00:12:47,360
Speaker 1: to q Bert. Right. UM. That amazes me. That's why

210
00:12:47,360 --> 00:12:51,160
Speaker 1: I bring up Cubert as often as possible. The Air

211
00:12:51,200 --> 00:12:53,079
Speaker 1: Force is actually going to be using this cluster to

212
00:12:53,679 --> 00:12:58,960
Speaker 1: test a way of processing multiple radar images. Basically, they

213
00:12:59,000 --> 00:13:01,920
Speaker 1: want to build them into fire resolution versions of those

214
00:13:01,920 --> 00:13:07,960
Speaker 1: images UM, and other projects including high definition video processing. UM.

215
00:13:08,040 --> 00:13:11,920
Speaker 1: You know, pretty heavy stuff. UM. But you know it

216
00:13:11,960 --> 00:13:15,199
Speaker 1: requires that much processing power and it actually apparently is

217
00:13:15,240 --> 00:13:17,960
Speaker 1: going to be handled at the Air Force Reach Research

218
00:13:18,080 --> 00:13:23,199
Speaker 1: Laboratories Information Center in Rome, New York. So. UM, it's

219
00:13:23,200 --> 00:13:26,520
Speaker 1: a very very specialized unit. It sounds like from the

220
00:13:26,520 --> 00:13:31,520
Speaker 1: from the descriptions I've read, UM. But what's really interesting

221
00:13:31,679 --> 00:13:34,080
Speaker 1: is you can do this yourself if you are. I

222
00:13:34,160 --> 00:13:35,800
Speaker 1: was wondering if we were going to go into the

223
00:13:35,840 --> 00:13:38,360
Speaker 1: how too, because this is this gets kind of complicated. No,

224
00:13:38,440 --> 00:13:40,120
Speaker 1: I wasn't going to go in depth into the how to,

225
00:13:40,240 --> 00:13:42,640
Speaker 1: but I was going to point out that uh Dr

226
00:13:42,720 --> 00:13:48,760
Speaker 1: Khanna and another UM, the U Mass Dartmouth Principal investigator H.

227
00:13:49,000 --> 00:13:53,079
Speaker 1: Chris Poulin, have a guide available to It's free, it's

228
00:13:53,120 --> 00:13:55,120
Speaker 1: open source. All you need to do is go to

229
00:13:55,200 --> 00:13:59,360
Speaker 1: PS three cluster dot org and you can find out

230
00:13:59,360 --> 00:14:02,760
Speaker 1: how to build your on PlayStation three uh cluster computer,

231
00:14:03,200 --> 00:14:05,480
Speaker 1: and they even tell you how you can still play

232
00:14:05,520 --> 00:14:07,800
Speaker 1: games on it, although it does require a special boot

233
00:14:08,520 --> 00:14:11,839
Speaker 1: system in order to boot back into the regular proprietary

234
00:14:11,960 --> 00:14:14,800
Speaker 1: PS three operating system, which is very very cool. Yeah,

235
00:14:15,000 --> 00:14:18,120
Speaker 1: so in general, I can run through like the high

236
00:14:18,240 --> 00:14:20,760
Speaker 1: level steps because we can't really get into the details.

237
00:14:20,880 --> 00:14:23,400
Speaker 1: It's really really complicated. Well, yeah, I'm not sure that

238
00:14:23,440 --> 00:14:26,600
Speaker 1: anybody is going to take the podcast and yeah, but

239
00:14:26,760 --> 00:14:28,560
Speaker 1: just just so you kind of get an idea of

240
00:14:28,560 --> 00:14:31,360
Speaker 1: what goes into this is that you you have to

241
00:14:31,400 --> 00:14:35,320
Speaker 1: be able to uh get a particular they they recommend

242
00:14:35,320 --> 00:14:39,440
Speaker 1: a specific Linux distribution that you then uh you uh

243
00:14:39,600 --> 00:14:42,720
Speaker 1: image onto a DVD and you use that as a

244
00:14:42,920 --> 00:14:45,920
Speaker 1: part of what you're going to do for the the

245
00:14:46,000 --> 00:14:49,000
Speaker 1: operating system for the Sony PS three. You also have

246
00:14:49,080 --> 00:14:51,720
Speaker 1: to have a USB memory stick to get a special

247
00:14:51,760 --> 00:14:55,440
Speaker 1: boot command on that um uh you put the image

248
00:14:55,440 --> 00:14:58,200
Speaker 1: of that on there. They actually host that file on

249
00:14:57,920 --> 00:15:03,400
Speaker 1: the site itself, and you using these together, you can

250
00:15:03,440 --> 00:15:08,280
Speaker 1: go into a Sony PlayStation three menu and there's actually

251
00:15:08,320 --> 00:15:12,480
Speaker 1: a choice to um switch the operating system to other

252
00:15:13,160 --> 00:15:15,440
Speaker 1: So you I mean, it's built in you can change

253
00:15:15,440 --> 00:15:17,400
Speaker 1: it that way. I mean that shows you that Sony

254
00:15:17,480 --> 00:15:21,320
Speaker 1: was thinking ahead and was being very accommodating to this

255
00:15:21,400 --> 00:15:25,040
Speaker 1: kind of hacking. Um, and we're using hacking in the uh,

256
00:15:25,680 --> 00:15:28,840
Speaker 1: not not in the malicious sense, but in the hey,

257
00:15:28,960 --> 00:15:32,240
Speaker 1: let's make this thing that does this particular task do

258
00:15:32,400 --> 00:15:35,480
Speaker 1: something totally unrelated, which is really cool. Let's take it

259
00:15:35,520 --> 00:15:37,880
Speaker 1: apart and see what else we can make it. So

260
00:15:38,040 --> 00:15:41,920
Speaker 1: you has they have steps for installing Linux onto the machine,

261
00:15:41,920 --> 00:15:44,800
Speaker 1: which is actually it's a multi step process, um and

262
00:15:44,800 --> 00:15:47,480
Speaker 1: it it might be a little intimidating, but if you

263
00:15:47,520 --> 00:15:51,320
Speaker 1: actually follow the instructions that are on the website, they're

264
00:15:51,440 --> 00:15:55,560
Speaker 1: very comprehensive and they you know you don't. It's one

265
00:15:55,600 --> 00:15:57,840
Speaker 1: of those things where it's better if you do know everything,

266
00:15:58,160 --> 00:16:01,280
Speaker 1: like the reasons why you are are going through these steps,

267
00:16:01,720 --> 00:16:04,320
Speaker 1: but it's not necessary because they tell you what to

268
00:16:04,360 --> 00:16:07,040
Speaker 1: do anyway, so you could just blindly follow the steps

269
00:16:07,040 --> 00:16:08,960
Speaker 1: and as long as you do that, you should be

270
00:16:09,000 --> 00:16:12,520
Speaker 1: all right. And uh. After you've installed Lenox, it's time

271
00:16:12,560 --> 00:16:17,080
Speaker 1: to create something called the message passing interface, which is

272
00:16:17,520 --> 00:16:21,560
Speaker 1: that's the way that it it processes information and networks information.

273
00:16:23,000 --> 00:16:26,200
Speaker 1: And you also have to get the uh software Developer Kit,

274
00:16:26,360 --> 00:16:29,320
Speaker 1: so that you can actually make it do whatever it

275
00:16:29,400 --> 00:16:31,480
Speaker 1: is that you plan on having it do, whether that's

276
00:16:31,520 --> 00:16:36,800
Speaker 1: search for intelligent life or plot thermonuclear war. I mean,

277
00:16:36,840 --> 00:16:40,520
Speaker 1: it's really up to you basically at that point, it's

278
00:16:40,560 --> 00:16:42,560
Speaker 1: just a tool. It's not that you know, it's not

279
00:16:42,600 --> 00:16:44,880
Speaker 1: gonna run your games better, let me put it that way.

280
00:16:44,920 --> 00:16:48,520
Speaker 1: You're not gonna suddenly see and improve performance from your

281
00:16:48,560 --> 00:16:50,400
Speaker 1: p S three as far as gaming is concerned. But

282
00:16:50,440 --> 00:16:52,800
Speaker 1: you can use it as a very powerful computer. And

283
00:16:52,840 --> 00:16:56,440
Speaker 1: if you have a particular project that requires parallel processing

284
00:16:56,440 --> 00:16:59,280
Speaker 1: and that's a lot of ps. Yeah, that thing we've

285
00:16:59,320 --> 00:17:02,120
Speaker 1: got a pop filed are here. Um. But if you

286
00:17:02,560 --> 00:17:05,280
Speaker 1: if you happen to have that that need, this could

287
00:17:05,520 --> 00:17:09,760
Speaker 1: be a resource for you, especially now that the prices dropped. Um.

288
00:17:09,800 --> 00:17:12,840
Speaker 1: And it's that's it's kind of interesting too that it's

289
00:17:12,880 --> 00:17:15,040
Speaker 1: gained so much traction as an idea to use the

290
00:17:15,280 --> 00:17:18,720
Speaker 1: PlayStation three as a model for the or and I

291
00:17:18,760 --> 00:17:21,920
Speaker 1: guess a unit to be built into a cluster UM.

292
00:17:22,280 --> 00:17:25,080
Speaker 1: The Cluster Workshop project, the one that we're just talking about,

293
00:17:25,160 --> 00:17:29,639
Speaker 1: is as now partially funded by the National Science Foundation. UM.

294
00:17:29,720 --> 00:17:32,760
Speaker 1: So it's you know, it's received that level of support

295
00:17:32,960 --> 00:17:36,359
Speaker 1: in the scientific community. So um, and from what I

296
00:17:36,359 --> 00:17:37,879
Speaker 1: could tell, it looks like a lot of people have

297
00:17:37,920 --> 00:17:40,800
Speaker 1: adopted the idea. Um, it's just one of those things

298
00:17:40,880 --> 00:17:44,120
Speaker 1: where when the military goes out and buys a whole

299
00:17:44,119 --> 00:17:47,920
Speaker 1: lot of game consoles and networked into a cluster computer.

300
00:17:48,040 --> 00:17:50,800
Speaker 1: It's it's it makes headlines because it's an interesting story.

301
00:17:50,840 --> 00:17:53,520
Speaker 1: Well yeah, and when you think about it, having to

302
00:17:53,600 --> 00:17:56,919
Speaker 1: submit that budget request up the chain of command to

303
00:17:56,960 --> 00:17:59,439
Speaker 1: the point where you know ultimately it's going to go

304
00:17:59,480 --> 00:18:03,400
Speaker 1: to some sort of politicians, it can be a challenge

305
00:18:03,480 --> 00:18:07,840
Speaker 1: to explain to someone not it's not not I don't

306
00:18:07,840 --> 00:18:11,719
Speaker 1: want to use the word savvy, but not particularly informed

307
00:18:12,200 --> 00:18:16,000
Speaker 1: about the possibility of of networking these devices together in

308
00:18:16,080 --> 00:18:18,719
Speaker 1: order to make an actual computer. Uh. You know, on

309
00:18:18,760 --> 00:18:20,359
Speaker 1: the surface, you just say, why the heck do you

310
00:18:20,400 --> 00:18:25,120
Speaker 1: want video game consoles? Uh? So, I mean that would

311
00:18:25,160 --> 00:18:26,800
Speaker 1: definitely be a challenge. I would not want to be

312
00:18:26,840 --> 00:18:29,840
Speaker 1: the guy who has to explain, especially if if I

313
00:18:29,880 --> 00:18:32,439
Speaker 1: were in a chain of command situation, I might not

314
00:18:32,520 --> 00:18:35,120
Speaker 1: even understand it. You know, it might be like three

315
00:18:35,160 --> 00:18:39,800
Speaker 1: people below me have have that information and then I'm

316
00:18:40,160 --> 00:18:43,720
Speaker 1: kind of stuck. But that is It is a nice

317
00:18:44,240 --> 00:18:47,520
Speaker 1: interesting way to solve the problem because supercomputers are not

318
00:18:47,600 --> 00:18:52,560
Speaker 1: just expensive. I mean, not all research facilities purchase supercomputers.

319
00:18:52,560 --> 00:18:57,160
Speaker 1: In fact, most don't. Most end up essentially, um not

320
00:18:57,160 --> 00:19:00,720
Speaker 1: not renting time, but getting having to I'm granted to

321
00:19:00,920 --> 00:19:04,840
Speaker 1: them to use a supercomputer, either through grant money or

322
00:19:04,960 --> 00:19:10,680
Speaker 1: sometimes the supercomputer, uh resource has a quota to meet

323
00:19:10,760 --> 00:19:13,320
Speaker 1: of a certain number of hours of processing time they

324
00:19:13,359 --> 00:19:18,320
Speaker 1: have to dedicate to research projects. But you've got you've

325
00:19:18,359 --> 00:19:21,960
Speaker 1: got more projects, and you have supercomputers, so you know

326
00:19:22,040 --> 00:19:24,720
Speaker 1: that time becomes a precious commodity. And if you if

327
00:19:24,760 --> 00:19:26,880
Speaker 1: you aren't able to book that time when your research

328
00:19:26,960 --> 00:19:29,879
Speaker 1: project is active, then it could be really hard to

329
00:19:29,920 --> 00:19:32,439
Speaker 1: keep that project going long enough for you to be

330
00:19:32,480 --> 00:19:34,399
Speaker 1: able to get the access. So this is a really

331
00:19:34,480 --> 00:19:39,240
Speaker 1: it's a viable alternative. Yeah. Yeah, And another nice thing

332
00:19:39,280 --> 00:19:42,200
Speaker 1: too is that, you know, it's one of the things

333
00:19:42,240 --> 00:19:45,320
Speaker 1: that keeps Google's data centers running. Mean, they use thousands

334
00:19:45,320 --> 00:19:48,040
Speaker 1: and thousands of computers as servers. You know, if one

335
00:19:48,080 --> 00:19:50,640
Speaker 1: of them goes down, there's some redundancy. So I would

336
00:19:50,680 --> 00:19:52,800
Speaker 1: assume although I you know, it's just one of those

337
00:19:52,800 --> 00:19:54,240
Speaker 1: things that just came to me I haven't researched it,

338
00:19:54,280 --> 00:19:57,240
Speaker 1: but I would assume if one of the PlayStations, um,

339
00:19:57,280 --> 00:19:59,840
Speaker 1: you know, gives up the ghost, then you could pull

340
00:19:59,840 --> 00:20:02,320
Speaker 1: it out and you know, put another machine in there

341
00:20:02,359 --> 00:20:05,520
Speaker 1: far more affordably than you could if you're a supercomputer,

342
00:20:06,000 --> 00:20:09,720
Speaker 1: you know, your crazy definitely, definitely, because you're talking about

343
00:20:09,800 --> 00:20:14,800
Speaker 1: you know, off the shelf components versus sometimes uh very

344
00:20:14,840 --> 00:20:19,720
Speaker 1: specific um unique parts. I mean, some of these supercomputers

345
00:20:19,720 --> 00:20:22,440
Speaker 1: are running on chips that you know, they essentially come

346
00:20:22,480 --> 00:20:24,720
Speaker 1: out of the lab and go into a supercomputer, so

347
00:20:24,760 --> 00:20:27,439
Speaker 1: they're not even at the mass production phase necessarily for

348
00:20:27,480 --> 00:20:29,840
Speaker 1: some of these chips, whereas with the p S three

349
00:20:29,920 --> 00:20:32,480
Speaker 1: you just go to best Buy or you know something

350
00:20:32,520 --> 00:20:34,520
Speaker 1: and just pick one off the shelf and go and

351
00:20:34,560 --> 00:20:36,119
Speaker 1: you buy it and you're done. You have to go

352
00:20:36,160 --> 00:20:37,800
Speaker 1: back and you you know, you have to of course

353
00:20:37,880 --> 00:20:41,000
Speaker 1: install the operating system and and and attach it to

354
00:20:41,040 --> 00:20:43,680
Speaker 1: your cluster part of the cluster. It's not like it's

355
00:20:43,680 --> 00:20:46,120
Speaker 1: just plug in play. But it's a lot easier than

356
00:20:46,280 --> 00:20:49,639
Speaker 1: if you suddenly had your supercomputer go oopsie on you. Yeah,

357
00:20:49,840 --> 00:20:52,320
Speaker 1: not that I've ever heard of a PlayStation three having

358
00:20:52,440 --> 00:20:54,080
Speaker 1: that kind of a problem. Now that's more of an

359
00:20:54,280 --> 00:20:57,679
Speaker 1: Xbox three sixty kind of thing. See, now people are

360
00:20:57,720 --> 00:20:59,119
Speaker 1: gonna write to me and say that I hate the

361
00:20:59,200 --> 00:21:01,640
Speaker 1: Xbox three six, despite the fact that I am an

362
00:21:01,680 --> 00:21:05,280
Speaker 1: avid Xbox fan. But come on, guys, that red ring

363
00:21:05,320 --> 00:21:09,800
Speaker 1: of death. Seriously. Oh you know, I I wouldn't mind having.

364
00:21:09,840 --> 00:21:11,720
Speaker 1: I don't own either one, and I would like at

365
00:21:11,760 --> 00:21:15,520
Speaker 1: PlayStation three and an Xbox reason so would I put

366
00:21:15,560 --> 00:21:18,159
Speaker 1: it on my Amazon wish list, And no one has

367
00:21:18,200 --> 00:21:20,239
Speaker 1: picked it up yet. I keep checking. Well, you know,

368
00:21:20,400 --> 00:21:22,080
Speaker 1: look in your stocking. We'll see what we can do.

369
00:21:23,080 --> 00:21:25,080
Speaker 1: And I have been gaining a lot of weight. My

370
00:21:25,160 --> 00:21:30,000
Speaker 1: stocking has been getting larger, Leon is getting larger. Well,

371
00:21:30,040 --> 00:21:32,000
Speaker 1: I guess, uh, that's a good discussion about the p

372
00:21:32,160 --> 00:21:35,880
Speaker 1: S three supercomputer connection. Um, that was a great question, Tim.

373
00:21:36,640 --> 00:21:38,399
Speaker 1: He actually he actually, Although I said it was a

374
00:21:38,440 --> 00:21:41,480
Speaker 1: listener text message, he sent his request via phone, which

375
00:21:41,520 --> 00:21:43,600
Speaker 1: a lot of you have been doing. And it just

376
00:21:43,640 --> 00:21:47,840
Speaker 1: shows that our listeners are super cool smartphone users, which

377
00:21:47,880 --> 00:21:52,240
Speaker 1: is awesome. Um. Well, I guess that wraps up that discussion,

378
00:21:52,280 --> 00:21:55,080
Speaker 1: but it does lead us to a little listener mail.

379
00:21:58,720 --> 00:22:01,960
Speaker 1: This listener mail comes from Immanual Emmanuel says, Hey, they're

380
00:22:02,560 --> 00:22:06,919
Speaker 1: imitating Jonathan because you guys are awesome. Thank you. You

381
00:22:06,960 --> 00:22:09,879
Speaker 1: guys mentioned that NATAL uses only gestures and it is

382
00:22:09,960 --> 00:22:12,200
Speaker 1: unsure of whether it will be too much of a

383
00:22:12,280 --> 00:22:15,040
Speaker 1: leap in technology. However, one thing you forgot to mention

384
00:22:15,080 --> 00:22:18,880
Speaker 1: is that NATAL can actually scan physical objects, which then

385
00:22:18,920 --> 00:22:21,439
Speaker 1: allows you to use real world objects in your games.

386
00:22:21,760 --> 00:22:24,560
Speaker 1: This can range from skateboards, tennis rackets, to an air

387
00:22:24,600 --> 00:22:28,080
Speaker 1: soft gun for first person shoots, etcetera. In my humble opinion,

388
00:22:28,160 --> 00:22:30,480
Speaker 1: the developers are really in control of whether or not

389
00:22:30,520 --> 00:22:33,199
Speaker 1: their game will implement some sort of physical object. The

390
00:22:33,240 --> 00:22:34,960
Speaker 1: only way to find out is to wait until the

391
00:22:35,000 --> 00:22:38,800
Speaker 1: release date. We shall see. I'm Emmanuel from Bermuda and

392
00:22:38,840 --> 00:22:43,560
Speaker 1: this is my okay. Now, this was a request from Immanual.

393
00:22:43,720 --> 00:22:46,879
Speaker 1: So anyone who gets annoyed by me yelling into the microphone,

394
00:22:46,880 --> 00:22:50,320
Speaker 1: this is your opportunity. Turned down the volume now, okay,

395
00:22:50,320 --> 00:22:52,480
Speaker 1: I'm going to count to three, and after that I'm

396
00:22:52,480 --> 00:22:56,000
Speaker 1: gonna fulfill Emmanual's requests. Remember this is a manual asking

397
00:22:56,040 --> 00:23:01,960
Speaker 1: for this, so please turn down your volume one, two, three.

398
00:23:02,960 --> 00:23:12,080
Speaker 1: Listener Maid in the tone that you used to use

399
00:23:12,119 --> 00:23:15,320
Speaker 1: because it always managed to make me laugh. Thanks Emmanuel,

400
00:23:15,480 --> 00:23:17,120
Speaker 1: I hope you guys were able to turn your volume

401
00:23:17,160 --> 00:23:20,760
Speaker 1: down in time. Don't send me iTunes reviews and say

402
00:23:20,760 --> 00:23:23,080
Speaker 1: that you hated it. I gave you plenty of opportunity

403
00:23:23,119 --> 00:23:25,520
Speaker 1: that time, guys. You know, I have to say it's

404
00:23:25,520 --> 00:23:29,680
Speaker 1: it's funny because, um, both sides are pretty vocal about

405
00:23:29,680 --> 00:23:32,399
Speaker 1: whether they liked or hated that. Yeah, we get email

406
00:23:32,440 --> 00:23:35,159
Speaker 1: all the time about people who say, thank you for

407
00:23:35,280 --> 00:23:38,439
Speaker 1: changing that, that was irritating, or I really miss the

408
00:23:38,480 --> 00:23:40,439
Speaker 1: way you used to say listener mail. And then we

409
00:23:40,480 --> 00:23:43,760
Speaker 1: still get reviews on iTunes that say I hate how

410
00:23:43,760 --> 00:23:46,400
Speaker 1: he yells into the microphone, Like I haven't done that

411
00:23:46,480 --> 00:23:50,080
Speaker 1: for like three months anyway, So I guess they're just

412
00:23:50,119 --> 00:23:52,920
Speaker 1: listening to all the old episodes. Thanks a lot, Emanuel.

413
00:23:52,960 --> 00:23:55,880
Speaker 1: That is a good point. Yes, Natal can scan objects

414
00:23:55,920 --> 00:23:58,159
Speaker 1: like like a skateboard. So it's kind of similar to

415
00:23:58,200 --> 00:24:00,560
Speaker 1: the Tony Hawk Ride game that came mount where you

416
00:24:00,560 --> 00:24:02,720
Speaker 1: have a very special controller that's in the shape of

417
00:24:02,720 --> 00:24:06,720
Speaker 1: a skateboard. Use that as a controller. Why are you

418
00:24:06,800 --> 00:24:09,840
Speaker 1: looking at me like that? Nothing? So anyway, you can

419
00:24:09,920 --> 00:24:12,800
Speaker 1: scan like a board and use it as if it

420
00:24:12,840 --> 00:24:14,879
Speaker 1: were one of those controllers. Now, granted the board you

421
00:24:14,920 --> 00:24:17,800
Speaker 1: scan would just be a normal board. It wouldn't even

422
00:24:17,920 --> 00:24:20,800
Speaker 1: have any controls in it or anything. It's an't all

423
00:24:20,840 --> 00:24:23,960
Speaker 1: scanning it and then interpreting that. So when I said

424
00:24:24,000 --> 00:24:26,919
Speaker 1: that you would have to use like just your empty

425
00:24:26,960 --> 00:24:29,960
Speaker 1: hand for like a lightsaber game, I totally forgot. No,

426
00:24:30,080 --> 00:24:32,320
Speaker 1: you don't necessarily have to do that. It may allow

427
00:24:32,359 --> 00:24:34,639
Speaker 1: you to scan an object and use that as the

428
00:24:34,720 --> 00:24:37,520
Speaker 1: hilt of a lightsaber, which could then help you defeat

429
00:24:37,560 --> 00:24:40,280
Speaker 1: the forces of evil, or if you are like me,

430
00:24:40,760 --> 00:24:45,080
Speaker 1: make everyone join the empire. I have a plus three

431
00:24:45,119 --> 00:24:49,119
Speaker 1: slice of pizza. Fear me. Okay, now you're mixing genres

432
00:24:49,320 --> 00:24:52,119
Speaker 1: and and different kinds of games. I'm gonna have to

433
00:24:52,160 --> 00:24:54,919
Speaker 1: close this out. If any of you would like to

434
00:24:54,920 --> 00:24:58,040
Speaker 1: send us mail, our address is tex Stuff at how

435
00:24:58,119 --> 00:25:00,879
Speaker 1: stuff works dot com. Remember we got a live show,

436
00:25:00,920 --> 00:25:03,680
Speaker 1: tech Stuff Live every Tuesday one pm Eastern. You can

437
00:25:03,720 --> 00:25:06,080
Speaker 1: find a link to that through the blogs and the

438
00:25:06,080 --> 00:25:08,439
Speaker 1: blog links are on how stuff works dot com on

439
00:25:08,480 --> 00:25:11,520
Speaker 1: the right hand side and Crispy and I will talk

440
00:25:11,560 --> 00:25:16,919
Speaker 1: to you again really soon. For more on this and

441
00:25:16,960 --> 00:25:20,200
Speaker 1: thousands of other topics, visit how stuff works dot com

442
00:25:20,240 --> 00:25:21,960
Speaker 1: and be sure to check out the new tech stuff

443
00:25:21,960 --> 00:25:29,320
Speaker 1: blog now on the house, Stuff Works homepage, brought to

444
00:25:29,359 --> 00:25:32,439
Speaker 1: you by the reinvented two thousand twelve camera. It's ready,

445
00:25:32,640 --> 00:25:33,120
Speaker 1: are you