1
00:00:00,570 --> 00:00:03,610
>>Dr. McGovern: I’m Dr. McGovern, and I’m
here today with Dr. Henry Neeman who is the

2
00:00:03,610 --> 00:00:08,510
director of the OU Supercomputing Center for
Education and Research. So Henry, can you

3
00:00:08,510 --> 00:00:09,950
tell us more about what OSCER does?

4
00:00:09,950 --> 00:00:16,950
>>Dr. Neeman: Sure! So we provide supercomputing
education, expertise and resources not only

5
00:00:16,980 --> 00:00:22,789
to the OU community, but to researchers across
the state and to the collaborators all around the

6
00:00:22,789 --> 00:00:28,259
world. So on the education side, we teach
people both about supercomputing and how to

7
00:00:28,259 --> 00:00:33,860
use supercomputing. In terms of expertise,
we work with researchers to get them productive

8
00:00:33,860 --> 00:00:38,980
quickly doing research on the supercomputer.
And then of course resources, that’s the

9
00:00:38,980 --> 00:00:40,400
machines themselves.

10
00:00:40,400 --> 00:00:43,750
>>Dr. McGovern: Okay, so can you tell our
audience what a supercomputer is? What makes

11
00:00:43,750 --> 00:00:45,200
a supercomputer “super”?

12
00:00:45,200 --> 00:00:51,090
>>Dr. Neeman: Sure. In order to be a supercomputer,
all you have to be is really, really big and

13
00:00:51,090 --> 00:00:55,640
really, really fast. In fact, you have to
be one of the biggest, fastest computers in

14
00:00:55,640 --> 00:00:59,870
the world right this minute. And the reason
I say right this minute is because it’s

15
00:00:59,870 --> 00:01:04,839
always changing. Computers are always getting
bigger and faster. So if something is a supercomputer

16
00:01:04,839 --> 00:01:09,150
today, it certainly won’t be a supercomputer
10 years from now. In fact, then it’ll be

17
00:01:09,150 --> 00:01:15,000
a laptop. Five years from now, maybe three
years from now – yeah, it’s probably still

18
00:01:15,000 --> 00:01:15,900
a supercomputer.

19
00:01:15,900 --> 00:01:18,470
>>Dr. McGovern: So how big is your supercomputer?
Can you tell us about it?

20
00:01:18,470 --> 00:01:21,810
>>Dr. Neeman: The one we currently have, which
is actually getting towards the end of its

21
00:01:21,810 --> 00:01:29,400
life, can do a little over 100 trillion calculations
per second. By comparison, the one that we had

22
00:01:29,479 --> 00:01:35,210
10 years ago could do one trillion calculations
per second. So we’ve increased by a factor

23
00:01:35,210 --> 00:01:37,530
of 100 over the course of 10 years.

24
00:01:37,530 --> 00:01:41,000
>>Dr. McGovern: And how many cores? They’ve
learned about what a core and a process—a CPU.

25
00:01:41,000 --> 00:01:41,500
>>Dr. Neeman: Sure.

26
00:01:41,500 --> 00:01:42,400
>>Dr. McGovern: How many do you have?

27
00:01:42,400 --> 00:01:49,300
>>Dr. Neeman: So we’ve got just under 7,000
CPU cores. And these are standard Intel CPUs,

28
00:01:49,360 --> 00:01:53,600
just like the kind of CPUs that you have in
a laptop or desktop PC.

29
00:01:53,600 --> 00:01:56,500
>>Dr. McGovern: Okay 7,000. And can they use
all 7,000 at once?

30
00:01:56,500 --> 00:02:00,820
>>Dr. Neeman: I don’t think we usually see
that. We’ve seen people use the majority

31
00:02:00,820 --> 00:02:04,990
of them. It gets a little bit tricky because
there’s lots of other people wanting to

32
00:02:04,990 --> 00:02:09,210
use them at the same time, and so if you get
into line and you’re waiting your turn,

33
00:02:09,210 --> 00:02:13,770
you might end up waiting a long time. Usually
when someone wants to run a job that big we

34
00:02:13,770 --> 00:02:17,640
recommend that they work with us, so that
for example we can get their job to start

35
00:02:17,640 --> 00:02:21,060
right after the end of a maintenance period when nobody
else is on the machine.

36
00:02:21,060 --> 00:02:25,280
>>Dr. McGovern: Okay. So for another aspect
of a supercomputer – you said it has 7,000

37
00:02:25,280 --> 00:02:26,500
cores. How much memory does it have?

38
00:02:26,500 --> 00:02:31,820
>>Dr. Neeman: It's about 15 and a half terabytes
of RAM, and then by comparison about 250 terabytes

39
00:02:31,820 --> 00:02:32,670
of disc.

40
00:02:32,670 --> 00:02:35,500
>>Dr. McGovern: Okay. And one other thing
they learned about was the GPUs.

41
00:02:35,500 --> 00:02:36,000
>>Dr. Neeman: Mhmmm.

42
00:02:36,000 --> 00:02:37,700
>>Dr. McGovern: Do you have a GPU-based supercomputer?

43
00:02:37,700 --> 00:02:45,400
>>Dr. Neeman: We have both GPUs and non-GPU
accelerators in the supercomputer. We have

44
00:02:45,470 --> 00:02:55,400
about 25 GPU cards in the supercomputer, and
then we have a like number of Intel Xeon Phi

45
00:02:55,460 --> 00:03:00,700
accelerator cards, which are essentially made
up of souped-up Intel Atom cores.

46
00:03:00,700 --> 00:03:03,100
>>Dr. McGovern: Okay. They’re not GPUs?

47
00:03:03,100 --> 00:03:07,160
>>Dr. Neeman: They’re not technically GPUs;
you can’t use them to do rendering, but

48
00:03:07,160 --> 00:03:11,900
they do number crunching in a way that’s
somewhat similar to how GPUs do number crunching.

49
00:03:11,900 --> 00:03:14,700
>>Dr. McGovern: Okay, and they can be used
to build their own special segment of the

50
00:03:14,700 --> 00:03:16,930
program? So they’re really fast – you
can run on just those?

51
00:03:16,930 --> 00:03:21,150
>>Dr. Neeman: You can. And in fact, 
with the Intel accelerators you can actually

52
00:03:21,150 --> 00:03:26,590
treat the Intel accelerators as if they’re
servers. With the GPUs you can’t really

53
00:03:26,590 --> 00:03:31,100
do that. You have a server that they’re
contained in – that’s true with the Intel

54
00:03:31,100 --> 00:03:35,910
accelerators as well. But with the GPUs you
have to use them as what’s sometimes referred

55
00:03:35,910 --> 00:03:40,350
to as offload engines, where you take part
of your computing and you stick it in the

56
00:03:40,350 --> 00:03:43,300
card, and then the card does that computing
really fast.

57
00:03:43,300 --> 00:03:46,860
>>Dr. McGovern: So they’ve learned how to
program in a language called Snap!, which

58
00:03:46,860 --> 00:03:48,070
is a graphical language. And I know –

59
00:03:48,070 --> 00:03:48,250
>>Dr. Neeman: Okay.

60
00:03:48,250 --> 00:03:51,060
>>Dr. McGovern: – you teach the students,
and they could be interested in your course

61
00:03:51,060 --> 00:03:52,670
next. You teach the students how to program
in C.

62
00:03:52,670 --> 00:03:53,240
>>Dr. Neeman: That’s correct.

63
00:03:53,240 --> 00:03:55,430
>>Dr. McGovern: What would they use if they
were going to use the supercomputer?

64
00:03:55,430 --> 00:04:01,880
>>Dr. Neeman: So the most popular languages
in supercomputing today are C, C++, and Fortran,

65
00:04:01,880 --> 00:04:06,530
and sort of not in that order. Fortran is
probably still the most popular. Fortran is

66
00:04:06,530 --> 00:04:11,600
actually – if not the oldest then one of
the oldest higher-level languages in history;

67
00:04:11,600 --> 00:04:18,800
it was developed in the late 1950s. But it’s
still very popular, largely because A) there’s

68
00:04:18,850 --> 00:04:24,450
a huge base of what we call legacy code out
there, which is old Fortran programs that

69
00:04:24,450 --> 00:04:30,600
are still useful and still being used in production
today, and partly because there are old Fortran

70
00:04:30,600 --> 00:04:32,530
programmers who are still useful –

71
00:04:32,530 --> 00:04:33,230
>>Dr. McGovern: [laughs]

72
00:04:33,230 --> 00:04:37,430
>>Dr. Neeman: – and they know Fortran very
well. And so they’ve continued to maintain

73
00:04:37,430 --> 00:04:43,240
the Fortran code. And one other reason: if
you’ve got a code that's been around for a long time, you’ve

74
00:04:43,240 --> 00:04:48,180
done lots of good debugging on it. If you
then translate it to a new language, you’re

75
00:04:48,180 --> 00:04:53,260
introducing exciting new bugs into it. That’s
usually not something they can afford to do.

76
00:04:53,260 --> 00:04:56,090
>>Dr. McGovern: And they’ve learned about
debugging this semester. So they’ve learned

77
00:04:56,090 --> 00:04:58,880
about writing code and commenting it, and
trying to find the right code and make sure

78
00:04:58,880 --> 00:04:59,570
it works right.

79
00:04:59,570 --> 00:05:04,560
>>Dr. Neeman: Yes, and remember, why waste
hours on design when you could spend weeks

80
00:05:04,560 --> 00:05:05,160
debugging?

81
00:05:05,160 --> 00:05:10,320
>>Dr. McGovern: Excellent advice. [laughs]
So you said that you used this for teaching

82
00:05:10,320 --> 00:05:14,120
people how to use the supercomputers. What
researchers across campus, what disciplines

83
00:05:14,120 --> 00:05:16,300
do they come from? Who uses OSCER? Can you
tell us?

84
00:05:16,300 --> 00:05:21,300
>>Dr. Neeman: It’s literally everything
from aerospace to zoology. The most popular,

85
00:05:21,300 --> 00:05:26,430
number one – you know, OU is a big meteorology
school. Number one is weather forecasting.

86
00:05:26,430 --> 00:05:32,360
We do an enormous amount of both real-time
weather forecasts and research runs for publications,

87
00:05:32,360 --> 00:05:39,500
and that adds up to close to half the total
usage of the supercomputer. Number two is

88
00:05:39,560 --> 00:05:45,200
molecular dynamics. So if you want to understand
the shapes that molecules make, you’ve got

89
00:05:45,200 --> 00:05:50,210
to calculate the forces among all of those
molecules in order to see how they’ll bend.

90
00:05:50,210 --> 00:05:55,520
Because the shape of a molecule then affects
the structures it can create, and that effects

91
00:05:55,520 --> 00:06:01,110
what you do with it. So if we’ve got DNA
that says build this particular thing, you

92
00:06:01,110 --> 00:06:04,550
could end up with a really smart brain or
you could end up with Alzheimer’s depending

93
00:06:04,550 --> 00:06:08,900
on whether that’s right or wrong, so to
speak. So molecular dynamics is number two;

94
00:06:08,900 --> 00:06:13,240
that’s probably between a quarter and a
third of our usage. Number three is high-energy

95
00:06:13,240 --> 00:06:17,270
physics. So this is banging tiny particles
together at unbelievably high speed. These

96
00:06:17,270 --> 00:06:23,490
folks are producing petabytes of data every
year that they – in order to chew through

97
00:06:23,490 --> 00:06:28,780
that data, they’ve got to farm it out to
zillions of universities not just in the U.S.,

98
00:06:28,780 --> 00:06:32,620
all over the world. I think there’s
something like over 100 countries involved

99
00:06:32,620 --> 00:06:34,090
in these kinds of projects.

100
00:06:34,090 --> 00:06:35,080
>>Dr. McGovern: Wow.

101
00:06:35,080 --> 00:06:42,100
>>Dr. Neeman: And we are part of what’s
called a tier two center. So tier one is one

102
00:06:42,110 --> 00:06:46,010
of the big national labs. We’re a tier two
center. We’re drawing in a lot of this data

103
00:06:46,010 --> 00:06:51,020
from the big accelerators. We’re chewing
through it looking for things like the Higgs

104
00:06:51,020 --> 00:06:53,290
boson, which was discovered about a year ago
–

105
00:06:53,290 --> 00:06:53,540
>>Dr. McGovern: Right.

106
00:06:53,540 --> 00:06:58,290
>>Dr. Neeman: – through that kind of experiment.
We’re also doing simulations to show whether

107
00:06:58,290 --> 00:07:03,210
what we think are the laws of physics that
govern banging tiny particles together actually

108
00:07:03,210 --> 00:07:06,310
are the laws of physics that govern banging
tiny particles together.

109
00:07:06,310 --> 00:07:10,260
>>Dr. McGovern: So that molecular stuff that
you were talking about, is that related to

110
00:07:10,260 --> 00:07:13,389
the protein folding? I know there’s like a program
you can download called “Fold at home”?

111
00:07:13,389 --> 00:07:18,010
>>Dr. Neeman: Right. So molecular dynamics
is what underlies protein folding; that’s

112
00:07:18,010 --> 00:07:22,800
how you do the protein folding. So if I’m
an atom and you’re an atom, there’s a

113
00:07:22,800 --> 00:07:27,720
force between us, and you can trivially calculate
that force. Now if there’s another atom

114
00:07:27,720 --> 00:07:32,050
sitting over here, we can calculate the force
from me to that atom from me to you. If there’s

115
00:07:32,050 --> 00:07:35,950
another one over there, from me to there,
me to there, me to there, and so on. So the

116
00:07:35,950 --> 00:07:40,250
number of forces we have to calculate is proportional
to the square of the number of atoms. That

117
00:07:40,250 --> 00:07:46,320
creates an enormous amount of calculation.
Folding@home is an example of where they discovered

118
00:07:46,320 --> 00:07:52,180
there was an enormous amount of idle computers around the world just sitting around in screensaver mode. So

119
00:07:52,180 --> 00:07:56,630
why not take advantage of that? When they’re
in screensaver mode, have them do number crunching

120
00:07:56,630 --> 00:08:01,240
for folding proteins. And they’re not literally,
like, folding the proteins; they’re calculating

121
00:08:01,240 --> 00:08:06,850
the shapes that would naturally arise based
on those forces. And I mentioned about Alzheimer’s

122
00:08:06,850 --> 00:08:13,650
– what, what gives rise to Alzheimer’s
is the proteins in your brain – they’re

123
00:08:13,650 --> 00:08:19,810
slightly incorrect, but they can fold very
wrong as a result of being slightly incorrect.

124
00:08:19,810 --> 00:08:23,740
They're the wrong shapes, so they don’t
turn into the correct structures, and that

125
00:08:23,740 --> 00:08:29,050
can make big problems for you. This is actually
a fairly acute issue for me, because my grandmother

126
00:08:29,050 --> 00:08:35,050
and my aunt both died of Alzheimer’s when
they were about 40 years older than I am now.

127
00:08:35,050 --> 00:08:37,520
So you’ve got 40 years to solve this problem,
folks.

128
00:08:37,520 --> 00:08:37,820
>>Dr. McGovern: [laughs]

129
00:08:37,820 --> 00:08:38,849
>>Dr. Neeman: Get on it, please.

130
00:08:38,849 --> 00:08:42,240
>>Dr. McGovern: I’ve seen some interesting
things in the news lately about Alzheimer’s

131
00:08:42,240 --> 00:08:42,419
research.

132
00:08:42,419 --> 00:08:42,940
>>Dr. Neeman: Mhmm.

133
00:08:42,940 --> 00:08:44,310
>>Dr. McGovern: So they’re making progress.

134
00:08:44,310 --> 00:08:49,500
>>Dr. Neeman: They are, and it’s partially
thanks to projects like Folding@home and other

135
00:08:49,500 --> 00:08:51,300
molecular dynamics projects.

136
00:08:51,300 --> 00:08:53,720
>>Dr. McGovern: So if they were interested
in the supercomputers – we’ll talk about

137
00:08:53,720 --> 00:08:57,520
OSCER’s resources – but just on the Folding@home
note, is that sort of making a virtual supercomputer

138
00:08:57,520 --> 00:08:58,630
for the students that they can join?

139
00:08:58,630 --> 00:09:02,670
>>Dr. Neeman: Absolutely. And in fact, projects
like Folding@home constitute arguably the

140
00:09:02,670 --> 00:09:04,900
biggest virtual supercomputer in the world.

141
00:09:04,900 --> 00:09:05,500
>>Dr. McGovern: That’s pretty cool.

142
00:09:05,500 --> 00:09:09,930
>>Dr. Neeman: Far bigger than any actual supercomputer has been able to build to, largely because

143
00:09:09,930 --> 00:09:13,000
building a huge supercomputer is incredibly
expensive.

144
00:09:13,000 --> 00:09:15,279
>>Dr. McGovern: I was involved in SETI at
home for a long time.

145
00:09:15,279 --> 00:09:16,520
>>Dr. Neeman: Mhmm, same kind of idea.

146
00:09:16,520 --> 00:09:18,150
>>Dr. McGovern: Yeah, looking for extraterrestrial
signals.

147
00:09:18,150 --> 00:09:21,700
>>Dr. Neeman: And it’s a brilliant solution
to a very tough problem, because they’re

148
00:09:21,700 --> 00:09:29,200
looking for radio signals from bug-eyed monsters
out in space, and leaving aside the fact that

149
00:09:29,250 --> 00:09:32,540
all of the move I’ve seen – when the bug-eyed
monsters find us then they come to eat us,

150
00:09:32,540 --> 00:09:33,600
but leaving that aside –

151
00:09:33,600 --> 00:09:35,120
>>Dr. McGovern: [laughs] We’re looking for
the friendly ones.

152
00:09:35,120 --> 00:09:38,320
>>Dr. Neeman: Yeah, so hopefully these are
the nice, friendly bug-eyed monsters out in

153
00:09:38,320 --> 00:09:44,360
space. But it’s really hard to convince
a funding agency or Congress to put a lot

154
00:09:44,360 --> 00:09:49,970
of money into supercomputers to look for bug-eyed
monsters out in space. And so SETI@home was

155
00:09:49,970 --> 00:09:54,710
this brilliant idea that gave rise to other
projects like Folding@home. Well, all these

156
00:09:54,710 --> 00:09:57,900
idle desktops – let’s see if we can get
people to volunteer –

157
00:09:57,900 --> 00:09:58,400
>>Dr. McGovern: Right.

158
00:09:58,400 --> 00:10:02,750
>>Dr. Neeman: – some of their idle machines
to help us look for the green, bug-eyed monsters

159
00:10:02,750 --> 00:10:03,310
out in space.

160
00:10:03,310 --> 00:10:05,300
>>Dr. McGovern: I hope that we find the friendly
aliens when we do find them.

161
00:10:05,300 --> 00:10:07,100
>>Dr. Neeman: I certainly do, too. That would
be way better.

162
00:10:07,100 --> 00:10:09,310
>>Dr. McGovern: Yeah. [laughs] They don’t
have to be bug-eyed.

163
00:10:09,310 --> 00:10:12,420
>>Dr. Neeman: No. Well you know, they seem to
just turn out that way in the movies.

164
00:10:12,420 --> 00:10:15,790
>>Dr. McGovern: So you mentioned that your
current supercomputer is getting to the end

165
00:10:15,790 --> 00:10:18,310
of its lifetime. What does that mean? What
are your plans for the next one?

166
00:10:18,310 --> 00:10:23,279
>>Dr. Neeman: So we’ve already done a request
for proposals from a variety of vendors to

167
00:10:23,279 --> 00:10:29,440
buy the next supercomputer. And the bids have
come in, we’ve done an evaluation on them,

168
00:10:29,440 --> 00:10:33,790
and we’ve submitted an agenda item to the
OU Board of Regents. So we’re waiting for

169
00:10:33,790 --> 00:10:37,430
the OU Board of Regents to have their next
meeting, which will be in mid March. At that

170
00:10:37,430 --> 00:10:43,010
meeting they’ll announce their decision.
Assuming all of that goes well, then we’ll

171
00:10:43,010 --> 00:10:47,510
move into a final design phase where we’re
working with the chosen vendor to work out

172
00:10:47,510 --> 00:10:53,050
the particulars. And this year for the first
time, we’ll be doing that in collaboration

173
00:10:53,050 --> 00:10:58,060
with faculty, staff, and students across
the University. We’ll be asking all of you

174
00:10:58,060 --> 00:11:03,520
to participate in that process to help us
make better decisions in terms of which kinds

175
00:11:03,520 --> 00:11:08,000
of components, what quantities of components
(within the budget that we’ve got available).

176
00:11:08,000 --> 00:11:11,500
>>Dr. McGovern: So the last time you went
up by 100 fold?

177
00:11:11,500 --> 00:11:14,700
>>Dr. Neeman: Well over 10 years we did; that
was over four different systems.

178
00:11:14,700 --> 00:11:17,880
>>Dr. McGovern: Okay, so what is the current
cycle? And what are you expecting to go up

179
00:11:17,880 --> 00:11:18,120
by?

180
00:11:18,120 --> 00:11:22,140
>>Dr. Neeman: It’s always hard to say for
sure, because it depends on market conditions

181
00:11:22,140 --> 00:11:26,670
and the particulars of the decisions we make.
We would expect a factor of two to three improvement

182
00:11:26,670 --> 00:11:27,670
over the current system.

183
00:11:27,670 --> 00:11:31,640
>>Dr. McGovern: Okay. So I know your current
naming scheme is that you name the supercomputers

184
00:11:31,640 --> 00:11:34,900
Boomer and Sooner. Which one are
you on now and what is the plan for the new one?

185
00:11:34,900 --> 00:11:39,460
>>Dr. Neeman: So as it turns out, we actually
have three names. So the current system is

186
00:11:39,460 --> 00:11:45,190
Boomer; its predecessor was Sooner. Some years
ago we got a grant from the National Science

187
00:11:45,190 --> 00:11:49,529
Foundation to buy a somewhat different flavor
of supercomputer, on a smaller scale. And

188
00:11:49,529 --> 00:11:50,779
we named that one Schooner –

189
00:11:50,779 --> 00:11:51,220
>>Dr. McGovern: Okay.

190
00:11:51,220 --> 00:11:52,520
>>Dr. Neeman: – after, of course, the Sooner
Schooner.

191
00:11:52,520 --> 00:11:52,920
>>Dr. McGovern: Right.

192
00:11:52,920 --> 00:11:56,810
>>Dr. Neeman: So we haven’t used that name
for a while, so we’re bringing it back into

193
00:11:56,810 --> 00:11:59,600
the rotation. So the next system will be known
as Schooner.

194
00:11:59,600 --> 00:12:03,450
>>Dr. McGovern: Okay. So how does OSCER help
researchers across campus? You said that part

195
00:12:03,450 --> 00:12:05,100
of your mission is education. Can you tell
us more?

196
00:12:05,100 --> 00:12:09,209
>>Dr. Neeman: Sure. On the education side,
we do a workshop series, we’re actually

197
00:12:09,209 --> 00:12:13,940
doing it right now – did it yesterday – called
“Supercomputing in Plain English.” Now this

198
00:12:13,940 --> 00:12:22,000
is 11 sessions where we cover not deep technical
content, but the concepts that underlie supercomputing

199
00:12:22,040 --> 00:12:27,760
and related technologies. So this past week
we did shared memory parallelism where you

200
00:12:27,760 --> 00:12:32,770
have a bunch of cores in your CPU, and they’re
all working together to solve some larger

201
00:12:32,770 --> 00:12:37,550
problem so that they can speed up the time
to solution. Next week we’re going to do

202
00:12:37,550 --> 00:12:42,390
distributed parallelism, where you have a
bunch of computers and they cooperate together

203
00:12:42,390 --> 00:12:49,360
over a network to solve a problem even bigger
than you could do inside an individual computer.

204
00:12:49,360 --> 00:12:52,620
But this workshop series – we’ve been
doing it literally since the day we opened

205
00:12:52,620 --> 00:12:57,520
our doors. We are the OU Supercomputing Center
for Education Research. We didn’t just put

206
00:12:57,520 --> 00:13:01,950
the “E” first because it made a better
acronym; we also put it first because we really

207
00:13:01,950 --> 00:13:08,200
do believe that if you aren’t helped to
learn how to use these technologies effectively,

208
00:13:08,200 --> 00:13:13,060
then giving you access to the technologies
is a waste of your time. So what we do – we

209
00:13:13,060 --> 00:13:18,510
focus very strongly, and the very first thing
we did the very first day we opened our doors

210
00:13:18,510 --> 00:13:22,670
was to teach the very first of these workshops.
And we’ve been doing it about every other

211
00:13:22,670 --> 00:13:28,870
year ever since then. Starting in 2007, we
started putting it out over video conferencing.

212
00:13:28,870 --> 00:13:35,370
And so we’ve served over 2,000 people in
every U.S. state – except Vermont and Rhode

213
00:13:35,370 --> 00:13:42,170
Island for some reason – and three U.S. territories
plus 17 other countries on every continent

214
00:13:42,170 --> 00:13:43,959
except Australia and Antarctica.

215
00:13:43,959 --> 00:13:47,410
>>Dr. McGovern: That impressive! You should
get on Janux, and then you can serve even

216
00:13:47,410 --> 00:13:47,850
more people.

217
00:13:47,850 --> 00:13:53,670
>>Dr. Neeman: That is true. And in fact, we
use this information when we write grant proposals

218
00:13:53,670 --> 00:13:58,190
to federal research funding agencies and say,
“Look at the amazing stuff we’ve been

219
00:13:58,190 --> 00:14:02,880
doing teaching the world about supercomputing,”
and that helps us to get funding.

220
00:14:02,880 --> 00:14:06,390
>>Dr. McGovern: So if the students get really
interested in using a supercomputer, how do

221
00:14:06,390 --> 00:14:08,500
they start? Do they come find you? What do
they do?

222
00:14:08,500 --> 00:14:12,560
>>Dr. Neeman: So they come find me, and they
come find someone like you. So in order for

223
00:14:12,560 --> 00:14:18,220
a student to get an account, all they need
is a faculty or staff sponsor, and then they

224
00:14:18,220 --> 00:14:23,290
just send an email – and we have a webpage
that lists the information that we need. They

225
00:14:23,290 --> 00:14:29,250
send an email with that information, we have
to verify it a little bit, and then we'll put together

226
00:14:29,250 --> 00:14:32,870
an account for them. We can usually turn that
around in about a business day or less.

227
00:14:32,870 --> 00:14:36,160
>>Dr. McGovern: Okay, and I know from past
experience that you really help them get started too.

228
00:14:36,160 --> 00:14:36,250
>>Dr. Neeman: Yes.

229
00:14:36,250 --> 00:14:37,529
>>Dr. McGovern: Because you helped our research
group get going.

230
00:14:37,529 --> 00:14:40,860
>>Dr. Neeman: So we’re very hands on. If
you’re starting in supercomputing and you

231
00:14:40,860 --> 00:14:46,970
don’t yet know A) how to do supercomputing at
all, and/or B) how to use our supercomputer,

232
00:14:46,970 --> 00:14:52,580
then either you can come visit one of us for
an hour or two or we’ll come visit you.

233
00:14:52,580 --> 00:14:55,589
And our general rule of thumb is we don’t
leave that meeting until you’re productive.

234
00:14:55,589 --> 00:14:56,220
>>Dr. McGovern: Yeah.

235
00:14:56,220 --> 00:14:59,589
>>Dr. Neeman: Until you can do supercomputing
without us being there.

236
00:14:59,589 --> 00:15:03,290
>>Dr. McGovern: And something I talked about
in some of my other interviews was research opportunities

237
00:15:03,290 --> 00:15:05,600
for undergraduates. Do you hire undergraduates
in your group at all?

238
00:15:05,600 --> 00:15:10,400
>>Dr. Neeman: We do a little bit. We don’t
have a huge budget for that. We currently

239
00:15:10,400 --> 00:15:17,410
have…I think it’s two student positions?
Maybe a third now – not all of them are

240
00:15:17,410 --> 00:15:24,410
filled right now. But we actually do hire
students, and we have them help us on a lot

241
00:15:24,440 --> 00:15:29,980
of the core tasks that we need to get done,
but that a person could learn pretty quickly

242
00:15:29,980 --> 00:15:35,320
how to do them effectively and correctly.
And then once a student starts down that road,

243
00:15:35,320 --> 00:15:39,980
over time they learn more and more, so that
they can come out of whatever their degree

244
00:15:39,980 --> 00:15:45,839
program is ready to take on a position similar
to the work that they’ve been doing. And

245
00:15:45,839 --> 00:15:52,000
the demand for those kinds of positions is
huge. I was at a big supercomputing conference

246
00:15:52,000 --> 00:15:58,170
back in November and there was a booth from one of the oil companies. And I went up to them and
I said, “So are you in the supercomputing

247
00:15:58,170 --> 00:16:02,029
business?” He said, “No, I’m here to
recruit.” And I said, “Oh. Well, so how’s

248
00:16:02,029 --> 00:16:04,950
that going?” And he said, “Well it’s going
pretty well, but it’s hard to find people.”

249
00:16:04,950 --> 00:16:09,000
And I said, “Well if you could find all
the people you wanted, how many more of them

250
00:16:09,000 --> 00:16:11,500
would you hire?” He said, “We would hire
three times as many –

251
00:16:11,500 --> 00:16:12,000
>>Dr. McGovern: Wow.

252
00:16:12,000 --> 00:16:13,149
>>Dr. Neeman: – if we could only find them.”

253
00:16:13,149 --> 00:16:13,930
>>Dr. McGovern: Wow.

254
00:16:13,930 --> 00:16:19,370
>>Dr. Neeman: Yeah. So the demand for people
who understand supercomputing is enormous.

255
00:16:19,370 --> 00:16:24,200
The other great thing about learning supercomputing
is it’s one of the only ways we know to

256
00:16:24,200 --> 00:16:29,000
actually be able to predict the future. And
I don’t mean that in the sense of running

257
00:16:29,000 --> 00:16:32,529
simulations, I mean in the sense of the future
of technology.

258
00:16:32,529 --> 00:16:33,300
>>Dr. McGovern: Ah, okay.

259
00:16:33,300 --> 00:16:35,540
>>Dr. Neeman: So you’ve heard of Moore’s
law: computing speed doubles every couple

260
00:16:35,540 --> 00:16:39,620
of years. Well I mentioned earlier, one of
the wonderful things about supercomputers

261
00:16:39,620 --> 00:16:44,380
is if it’s a supercomputer today, in 10
to 15 years it’s going to be a laptop.

262
00:16:44,380 --> 00:16:45,050
>>Dr. McGovern: Right.

263
00:16:45,050 --> 00:16:50,620
>>Dr. Neeman: And in 25 to 30, it’s going
to be a cell phone. So what you know once

264
00:16:50,620 --> 00:16:57,260
you’ve used supercomputing is you know 10
to 15 years ahead of what the world of technology

265
00:16:57,260 --> 00:17:03,079
is going to look like. Because what we’ve
seen again and again perfectly consistently is whatever happens in

266
00:17:03,079 --> 00:17:08,139
supercomputing today will be on your desk
or lap in 10 to 15 years; that always happens. So

267
00:17:08,139 --> 00:17:14,159
it’s one of the few ways we have – you
know, Yogi Berra is claimed to have said

268
00:17:14,159 --> 00:17:19,149
and so are others have claimed to have said, “Predictions are hard, especially about the future.”

269
00:17:19,149 --> 00:17:19,419
>>Dr. McGovern: Right.

270
00:17:19,419 --> 00:17:23,489
>>Dr. Neeman: But this is one prediction we
can teach you to make. Because all you've got

271
00:17:23,489 --> 00:17:28,600
to do is play with today’s supercomputers,
and you can pretty much guarantee what a laptop is

272
00:17:28,600 --> 00:17:30,619
going to look like in 10 to 15 years.

273
00:17:30,619 --> 00:17:33,239
>>Dr. McGovern: Well one of the things that
they’ve been talking about – we’ve taught them

274
00:17:33,239 --> 00:17:34,300
about transistors.

275
00:17:34,300 --> 00:17:34,800
>>Dr. Neeman: Mhmm.

276
00:17:34,800 --> 00:17:38,500
>>Dr. McGovern: And about one
of the technologies – carbon nanotube transistors?

277
00:17:38,500 --> 00:17:40,879
>>Dr. McGovern: And this looks like they might
fundamentally revolutionize everything.

278
00:17:40,879 --> 00:17:45,619
>>Dr. Neeman: Yeah, there’s a lot of new
exciting technologies that the particular

279
00:17:45,619 --> 00:17:51,259
technologies we’re using today in hardware
are maybe coming toward the end of their life

280
00:17:51,259 --> 00:17:54,600
in the sense that we can’t keep shrinking
them forever.

281
00:17:54,600 --> 00:17:55,100
>>Dr. McGovern: Right.

282
00:17:55,100 --> 00:17:58,019
>>Dr. Neeman: We can shrink them down to a
certain level, but we’re not too many years

283
00:17:58,019 --> 00:18:02,340
away from that being as far as we can shrink
them. And that means we’re coming to the

284
00:18:02,340 --> 00:18:07,840
end of Moore’s law for that technology,
but we’re not coming to the end of Moore’s

285
00:18:07,840 --> 00:18:12,919
law because as you say, there are new technologies
coming. So carbon nanotubes is one of them.

286
00:18:12,919 --> 00:18:20,000
Another is DNA computing, another is quantum
computing. These are all areas that are...well, rich

287
00:18:20,080 --> 00:18:22,850
areas of research, which means we don’t
know what we’re doing yet.

288
00:18:22,850 --> 00:18:24,460
>>Dr. McGovern: [laughs] Right.

289
00:18:24,460 --> 00:18:30,090
>>Dr. Neeman: Where within probably 10 to 15 years,
one or more of these technologies will be

290
00:18:30,090 --> 00:18:36,859
in common use around the world. There’s
already one company that sells what they describe

291
00:18:36,859 --> 00:18:44,900
as a quantum computer, and there are explorations
– and you see in, in the newspapers articles

292
00:18:44,999 --> 00:18:50,309
about using DNA to store data. In some ways
it’s superior to the technologies we have

293
00:18:50,309 --> 00:18:56,649
today, including, especially, the amount of capacity you
can squeeze onto it. And carbon nanotubes have

294
00:18:56,649 --> 00:19:00,070
a lot of advantages with respect to physical
properties that make it possible to get a

295
00:19:00,070 --> 00:19:05,820
lot more out of them and to keep that shrink
rate going much farther than the current technologies

296
00:19:05,820 --> 00:19:06,200
could do.

297
00:19:06,200 --> 00:19:10,000
>>Dr. McGovern: Right. I think the current
nano…I think they’re 14 nanometers for a transistor –

298
00:19:10,000 --> 00:19:10,500
>>Dr. Neeman: Right.

299
00:19:10,500 --> 00:19:11,059
>>Dr. McGovern: – right now?

300
00:19:11,059 --> 00:19:16,409
>>Dr. Neeman: Right. We’ve seen officially
announced down to at least eight nanometers,

301
00:19:16,409 --> 00:19:22,690
maybe – I can’t remember if I’ve seen
six or five actually announced. I know I’ve

302
00:19:22,690 --> 00:19:27,340
heard discussions about that. You can look
on Wikipedia and they’ll talk about it.

303
00:19:27,340 --> 00:19:32,340
But yes, they’ll keep going for a while,
but there isn’t that much farther they can

304
00:19:32,340 --> 00:19:32,559
go.

305
00:19:32,559 --> 00:19:35,799
>>Dr. McGovern: Right. So part of the reason
we’re doing this series of interviews is

306
00:19:35,799 --> 00:19:39,019
to have the students learn about the impact
of computing and the impact of the Internet

307
00:19:39,019 --> 00:19:42,450
on science and engineering research. So can
we ask those two questions? We’ll start

308
00:19:42,450 --> 00:19:43,200
with the impact of computing.

309
00:19:43,200 --> 00:19:43,700
>>Dr. Neeman: Sure.

310
00:19:43,700 --> 00:19:46,730
>>Dr. McGovern: How have your experience shown
you – how has computers and supercomputers

311
00:19:46,730 --> 00:19:47,629
changed the world?

312
00:19:47,629 --> 00:19:52,570
>>Dr. Neeman: In the specific context of science
and engineering research, computers have changed

313
00:19:52,570 --> 00:19:57,210
everything. The way that we do science and
engineering research today is fundamentally

314
00:19:57,210 --> 00:20:01,980
different from how it was done before there
were computers. And a lot of that is we now

315
00:20:01,980 --> 00:20:06,460
have the ability, in addition to analyzing
mountains of data – which is incredibly

316
00:20:06,460 --> 00:20:13,009
important – we also have the ability to
simulate, to, to pretend a particular physical

317
00:20:13,009 --> 00:20:18,080
phenomenon in order to discover not just the
effects of that phenomenon, which is usually

318
00:20:18,080 --> 00:20:22,940
what you can get from laboratory experiments
and observation, but to actually examine the

319
00:20:22,940 --> 00:20:30,200
mechanisms. The, the great thing about computational
science and engineering is that you can tackle

320
00:20:30,229 --> 00:20:37,500
problems that are either too big or too small
or too fast or too slow or too dangerous or

321
00:20:37,580 --> 00:20:44,600
too expensive to do in real life. So if you
think about too big, my advisor for my PhD

322
00:20:44,600 --> 00:20:49,419
was actually an astronomer – a cosmologist.
So he studies the entire universe over its

323
00:20:49,419 --> 00:20:54,340
entire lifetime, from an infinitesimally after
the Big Bang until today, and in fact, on

324
00:20:54,340 --> 00:21:00,129
into the future. The only way they can figure
out what’s going on in the universe is to

325
00:21:00,129 --> 00:21:05,169
guess what the rules are and then simulate
it and see whether the result looks like what we

326
00:21:05,169 --> 00:21:10,159
see when we look through a telescope. So what
they’ve come up with now is they have a

327
00:21:10,159 --> 00:21:15,509
very clear sense of how old the universe is
relative to the Big Bang. They have a pretty

328
00:21:15,509 --> 00:21:20,600
good sense of what the mix of ingredients
in the universe are. Much of what they did

329
00:21:20,600 --> 00:21:25,600
in the ‘80s and ‘90s was like trying to
bake a cake where you knew the list of ingredients,

330
00:21:25,600 --> 00:21:29,840
but had no idea the proportions. So you
know, one day you would bake a cake that was,

331
00:21:29,840 --> 00:21:35,570
you know, 10 pounds of butter and a pinch
of flour and an ocean of sugar. And you know,

332
00:21:35,570 --> 00:21:39,720
as much as everybody wants to have that heart
attack, you’re not going to get a very good

333
00:21:39,720 --> 00:21:45,279
cake. Now they’ve done enough of these that
they have a pretty good sense of the propor—how

334
00:21:45,279 --> 00:21:51,359
much flour, how much sugar, how many eggs.
Only it’s dark matter and dark energy and

335
00:21:51,359 --> 00:21:56,619
visible matter and visible energy that are
the components. Helium and hydrogen are the, the

336
00:21:56,619 --> 00:21:59,190
primary constituents of the whole universe.

337
00:21:59,190 --> 00:21:59,489
>>Dr. McGovern: Right.

338
00:21:59,489 --> 00:22:04,149
>>Dr. Neeman: They wouldn’t have known any
of that without computing. With weather forecasting,

339
00:22:04,149 --> 00:22:08,500
in the olden days – well, you know the joke is you
have a rope you hang out the window, and when

340
00:22:08,500 --> 00:22:13,710
it’s wet, it must be raining and when it’s
dry, it’s sunny out. You know…it wasn’t

341
00:22:13,710 --> 00:22:18,259
quite that bad, but before computers it was
very, very difficult to predict the weather.

342
00:22:18,259 --> 00:22:23,070
Now we can predict the weather with a pretty
good degree of reliability. What they call

343
00:22:23,070 --> 00:22:29,149
the skill scores are vastly better today than
they were even a decade or two ago, let alone

344
00:22:29,149 --> 00:22:33,700
50 years ago. These are huge changes. But
more than that, if I want to understand how

345
00:22:33,700 --> 00:22:37,729
does a tornado work? What’s really going
on inside a tornado? Well here’s a really

346
00:22:37,729 --> 00:22:40,690
bad way to go about it; I could walk around
inside that tornado.

347
00:22:40,690 --> 00:22:41,479
>>Dr. McGovern: Right.

348
00:22:41,479 --> 00:22:45,950
>>Dr. Neeman: That’s not the best plan – that’s
an example of too dangerous. But on the other

349
00:22:45,950 --> 00:22:53,700
hand, if I simulate it, I can virtually walk
around inside my virtual tornado at no risk.

350
00:22:53,720 --> 00:22:59,590
And if I think about expense, you know the,
the airplane manufacturers – they used to

351
00:22:59,590 --> 00:23:03,919
have to build dozens of prototypes. It’s
incredibly expensive to do that. It’s very

352
00:23:03,919 --> 00:23:10,090
time consuming, and people die when they fly
them. Now they’ve gone from dozens of prototypes

353
00:23:10,090 --> 00:23:14,639
down to typically for a new, newly released
kind of air vehicle, typically they’ll do

354
00:23:14,639 --> 00:23:15,080
two or three.

355
00:23:15,080 --> 00:23:15,720
>>Dr. McGovern: Because the—

356
00:23:15,720 --> 00:23:20,249
>>Dr. Neeman: And the savings – but they’ll
do zillions of simulations. So they actually

357
00:23:20,249 --> 00:23:25,710
understand the vehicle better than they used
to, but vastly cheaper.

358
00:23:25,710 --> 00:23:30,259
>>Dr. McGovern: Right. That’s a great answer
of how it’s fundamentally transforming everything.

359
00:23:30,259 --> 00:23:34,720
And I was going to add for the weather one
thought. They have all these new weather radars

360
00:23:34,720 --> 00:23:35,309
>>Dr. Neeman: Mhmm.

361
00:23:35,309 --> 00:23:38,639
>>Dr. McGovern: –that exist. And I don’t
think that they could ingest and understand that

362
00:23:38,639 --> 00:23:40,320
data without the supercomputers as well.

363
00:23:40,320 --> 00:23:44,259
>>Dr. Neeman: It’s fascinating. When those
weather radars first came online, the, the

364
00:23:44,259 --> 00:23:50,419
only technology that was affordable at the
time for moving data from the weather radar

365
00:23:50,419 --> 00:23:55,970
to the supercomputers that were going to use
it to do the simulations – they had phone

366
00:23:55,970 --> 00:23:58,359
modems running at 56 kilobits per second.

367
00:23:58,359 --> 00:23:58,840
>>Dr. McGovern: Wow.

368
00:23:58,840 --> 00:24:03,580
>>Dr. Neeman: Right? And they would completely
max those out with the data that was coming

369
00:24:03,580 --> 00:24:09,729
off the radars. Now, of course, they can wire
them up with one or 10 or 100 megabit lines

370
00:24:09,729 --> 00:24:14,849
affordably, because network technology has
improved – in fact, even faster than computing

371
00:24:14,849 --> 00:24:20,729
technology has improved. But none of that
would have been conceivable – all those

372
00:24:20,729 --> 00:24:26,720
radars would have been sitting there producing
very little value if not for the ability to

373
00:24:26,720 --> 00:24:32,849
first gather that data and then send it out
to the supercomputers at very high speeds.

374
00:24:32,849 --> 00:24:38,669
So that’s been hugely important to our ability
to, to forecast the weather, and so have satellites.

375
00:24:38,669 --> 00:24:44,090
And all of that satellite technology has been
deeply dependent on improvements in computing.

376
00:24:44,090 --> 00:24:47,440
>>Dr. McGovern: Right. So that brings me to
my next question, which is that they’re

377
00:24:47,440 --> 00:24:51,869
learning about why or how computers have changed
the world and then how has the Internet changed

378
00:24:51,869 --> 00:24:52,600
the world. So –

379
00:24:52,600 --> 00:24:55,139
>>Dr. Neeman: Right. And, and so the answer
to the question about the Internet, there’s

380
00:24:55,139 --> 00:24:59,799
really two pieces to it: there’s the things
piece and there’s the people piece. So on

381
00:24:59,799 --> 00:25:05,269
the things side, we went from – I remember
when I was an undergrad, I had a little – it

382
00:25:05,269 --> 00:25:09,619
wasn’t even a PC; we didn’t have that
term back then. But I had a little computer,

383
00:25:09,619 --> 00:25:15,849
and it had a little phone modem that would
do 300 bits per second. And I felt myself

384
00:25:15,849 --> 00:25:21,840
fortunate because I didn’t have to go into
the computer lab to work on a project. And

385
00:25:21,840 --> 00:25:25,899
that worked great until I ran out of – we were given
allocations for how much time we could have,

386
00:25:25,899 --> 00:25:30,220
and I ran out. And then I had to call the
professor at home during dinner. He was way

387
00:25:30,220 --> 00:25:34,070
better about it than I would’ve been. [Dr.
McGovern laughs] And he actually got me an

388
00:25:34,070 --> 00:25:38,739
additional allocation, I was able to finish
the project. But nowadays networks are typically

389
00:25:38,739 --> 00:25:45,239
running a bill—a million to a billion times
that fast. And so the amount of data that we can

390
00:25:45,239 --> 00:25:46,409
move is phenomenal, –

391
00:25:46,409 --> 00:25:47,369
>>Dr. McGovern: Right.

392
00:25:47,369 --> 00:25:53,529
>>Dr. Neeman: – but it’s also always improving.
So we’re having to invent new words for

393
00:25:53,529 --> 00:25:58,639
how big things are, right? We know words like
kilobyte and megabyte and gigabyte; now we

394
00:25:58,639 --> 00:26:07,200
know people are typically familiar with terabyte.
Coming along is petabyte, exabyte, zettabyte

395
00:26:07,200 --> 00:26:12,700
and yottabyte, and they’re already working
on inventing words for amounts of data bigger

396
00:26:12,789 --> 00:26:14,000
than yottabyte.

397
00:26:14,000 --> 00:26:14,900
>>Dr. McGovern: That’s hard to imagine.

398
00:26:14,900 --> 00:26:19,460
>>Dr. Neeman: It, it, it’s hard to imagine,
but soon enough it’ll be on your desk. And

399
00:26:19,460 --> 00:26:25,419
that’s really the amazing thing – you
know famously Bill Gates was quoted many,

400
00:26:25,419 --> 00:26:29,749
many years ago saying, “Well I can’t imagine
that somebody’s going to need more than

401
00:26:29,749 --> 00:26:32,000
640 kilobytes of RAM.”

402
00:26:32,000 --> 00:26:32,500
>>Dr. McGovern: Right.

403
00:26:32,500 --> 00:26:40,000
>>Dr. Neeman: Now we can’t imagine 640 kilobytes
of CPU cache, right? That, the, the change that we’ve

404
00:26:40,000 --> 00:26:44,340
experienced is unbelievable, and, and really there’s
a beautiful thought experiment underlying

405
00:26:44,340 --> 00:26:47,279
this – and then I’ll get back to the whole
Internet, people part. But there’s a beautiful

406
00:26:47,279 --> 00:26:52,269
thought experiment you can do. So a typical
undergrad, a, a, not a returning student,

407
00:26:52,269 --> 00:26:55,899
but a, an undergrad coming out of high school
to one significant figure is about 20 years

408
00:26:55,899 --> 00:27:01,440
old, and life expectancy in the U.S. to one
significant figure is about 80. So a typical

409
00:27:01,440 --> 00:27:06,340
undergrad has about 60 years left. And I know
you think that’s a long time. Believe me,

410
00:27:06,340 --> 00:27:11,169
it goes like that. [snaps] But that 60 years
– well computing speed is doubling every

411
00:27:11,169 --> 00:27:16,909
two years, so that 60 years there’s going
to be 30 doublings. Two to the thirtieth is

412
00:27:16,909 --> 00:27:21,369
a billion. So what you’ve got on your desk
today compared to what you’ll have on your

413
00:27:21,369 --> 00:27:25,179
desk the day you die – well, the day before
you die. The day you die it’s not that useful.

414
00:27:25,179 --> 00:27:28,690
[Dr. McGovern laughs] But the day before you
die, the machine on your desk is going to

415
00:27:28,690 --> 00:27:34,500
be a billion times bigger and faster than
what you’ve got on your desk today. We can’t

416
00:27:34,500 --> 00:27:38,669
possibly imagine what we’re going to do
with all that computing power; we do know

417
00:27:38,669 --> 00:27:44,039
it’s going to be amazing. So – now getting
back to your question about the Internet.

418
00:27:44,039 --> 00:27:49,559
What’s really changed is not the technology.
That’s marched on, and that’s wonderful,

419
00:27:49,559 --> 00:27:53,919
and I’m in the technology business, so all
of that stuff jazzes me up. But the really

420
00:27:53,919 --> 00:27:58,090
amazing thing that’s happened is our ability
to communicate with one another; it has fundamentally

421
00:27:58,090 --> 00:28:04,100
transformed. You know, the way you and I communicated when we were in grade school, right? We didn’t

422
00:28:04,100 --> 00:28:09,019
have any of these newfangled cell phones and
all that sort of thing. But the way that people

423
00:28:09,019 --> 00:28:15,889
can communicate now – it’s changed how
we run our lives. So it used to be you couldn’t

424
00:28:15,889 --> 00:28:20,039
run right up to the deadline on some assignment,
because if something went wrong you couldn’t

425
00:28:20,039 --> 00:28:23,799
pick up the phone and expect the other person
to answer; now they’re carrying their phone

426
00:28:23,799 --> 00:28:27,489
on their hip, and you can get them 24/7. And
in fact if they don’t answer their phone,

427
00:28:27,489 --> 00:28:31,869
you’re annoyed at them, because you’ve
got an expectation that people can be reached

428
00:28:31,869 --> 00:28:38,460
on demand. That’s fundamentally different
from how people lived even 30 years ago, and

429
00:28:38,460 --> 00:28:45,090
that change has made it possible to do our
work in ways that were unimaginable back then.

430
00:28:45,090 --> 00:28:48,849
You know, if you’ve seen pictures of people
carrying around the first cell phones, and

431
00:28:48,849 --> 00:28:52,989
they were the size of like, a stack of bricks
and they weighed about as much as a stack

432
00:28:52,989 --> 00:28:56,859
of bricks. And if you were lucky, there were
maybe three or four cities in the country

433
00:28:56,859 --> 00:29:01,989
where you could use them, and the sound quality
was awful – but it was better than nothing.

434
00:29:01,989 --> 00:29:04,710
And now you’ve got these tiny little things
– actually they’ve gotten bigger now,

435
00:29:04,710 --> 00:29:07,669
because now we’re all playing games on them
and watching videos – but you’ve got these

436
00:29:07,669 --> 00:29:12,950
tiny little things that can do more than the
fastest supercomputers of 30 years ago. It’s

437
00:29:12,950 --> 00:29:18,200
just incredible. And that’s lead to – we
can tell each other not only, you know, here’s a

438
00:29:18,200 --> 00:29:23,190
picture of what I had for lunch today, which
may or may not be of interest to your friends.

439
00:29:23,190 --> 00:29:27,719
But you know, “I’m going to go over to see
such-and-such show or I’m going to go to

440
00:29:27,719 --> 00:29:32,379
such-and-so park and hang out. Anybody want
to come with?” People just show up. That

441
00:29:32,379 --> 00:29:34,000
was unimaginable 30 years ago.

442
00:29:34,000 --> 00:29:37,219
>>Dr. McGovern: Well and in the research land,
you can collaborate with people in different

443
00:29:37,219 --> 00:29:40,359
countries that you possibly – you couldn’t
even imagine at all.

444
00:29:40,359 --> 00:29:40,590
>>Dr. Neeman: Right.

445
00:29:40,590 --> 00:29:41,330
>>Dr. McGovern: And now you can –

446
00:29:41,330 --> 00:29:45,460
>>Dr. Neeman: And, and, I’ve got colleagues
all over the world now. And if I’ve got

447
00:29:45,460 --> 00:29:49,389
a question, I can just shoot an email over
to them and in five minutes I’ve got an

448
00:29:49,389 --> 00:29:55,099
answer, assuming they’re not asleep. But
they’ve got an answer for me. Again, not

449
00:29:55,099 --> 00:29:56,929
something you could’ve imagined 30 years
ago.

450
00:29:56,929 --> 00:29:59,489
>>Dr. McGovern: Well and I was thinking along
the lines of the Internet, another thing you

451
00:29:59,489 --> 00:30:00,580
can do is video conferences – >>Dr. Neeman: Mhmm.

452
00:30:00,580 --> 00:30:02,950
>>Dr. McGovern: – with people around the
world. We’re constantly scheduling – I

453
00:30:02,950 --> 00:30:05,879
have a group of colleagues that are scattered
across like five different time zones. And

454
00:30:05,879 --> 00:30:08,629
we’re trying to make our meetings, we’re
all saying, “Oh this time zone. No, this

455
00:30:08,629 --> 00:30:08,789
time,” you know? 

456
00:30:08,789 --> 00:30:11,359
>>Dr. Neeman: You’ve got to pick a time
when everybody can be awake, and there’s

457
00:30:11,359 --> 00:30:11,719
always one –

458
00:30:11,719 --> 00:30:12,049
>>Dr. McGovern: Right.

459
00:30:12,049 --> 00:30:13,849
>>Dr. Neeman: – poor schmo who has to do
it in the middle of the night.

460
00:30:13,849 --> 00:30:16,000
>>Dr. McGovern: There, there is. But you couldn’t
even have done that before. You couldn’t have had 

461
00:30:16,000 --> 00:30:16,600
>>Dr. Neeman: That’s absolutely correct.

462
00:30:16,649 --> 00:30:17,149
>>Dr. McGovern: – those
meetings.

463
00:30:17,149 --> 00:30:17,900
>>Dr. Neeman: Right.

464
00:30:17,900 --> 00:30:22,320
>>Dr. McGovern: So I know that, you know, another thing that the Internet can do – when soldiers

465
00:30:22,320 --> 00:30:26,099
deploy now, they can actually keep in contact with their families. When my dad deployed

466
00:30:26,099 --> 00:30:30,809
I was 10, we could call him once a week
because a long-distance call to a very far away

467
00:30:30,809 --> 00:30:34,789
country didn't sound so good. We could hardly hear him. It was super expensive, which you can’t

468
00:30:34,789 --> 00:30:39,179
imagine now. And sometimes we called him on
ham radio, because that was another way that

469
00:30:39,179 --> 00:30:42,599
we could find him. But there was no Skype
or anything else. And it’s a completely

470
00:30:42,599 --> 00:30:44,710
different world now, and I think that’s
great.

471
00:30:44,710 --> 00:30:50,440
>>Dr. Neeman: Yeah. It, it’s changed the
way that we…relate to one another, and you

472
00:30:50,440 --> 00:30:55,809
know, it’s always a mixed blessing. There
are pluses and minuses, but we’re not going

473
00:30:55,809 --> 00:31:02,379
back from that. We’re moving forward to
be able to communicate with each other using

474
00:31:02,379 --> 00:31:07,469
these technologies. And another thing that
I find absolutely fascinating is now especially

475
00:31:07,469 --> 00:31:13,929
with tablets – they’re becoming so cheap
that even in the poorest of countries, for

476
00:31:13,929 --> 00:31:17,789
many people they’re starting to be affordable.
In a way that, you know, when we had desktop

477
00:31:17,789 --> 00:31:24,129
PCs it was not imaginable that, that the poorest
of the poor could be able to have their own

478
00:31:24,129 --> 00:31:27,239
technology. Now these things are becoming
accessible. You know, if you want to buy a

479
00:31:27,239 --> 00:31:33,809
tablet for 35 bucks, you can go on the Wal-Mart
website or wherever and you can pick one up.

480
00:31:33,809 --> 00:31:37,299
It’s not going to be the world’s best
tablet by any means, but it’ll do the job.

481
00:31:37,299 --> 00:31:41,089
>>Dr. McGovern: I’ve heard some stories
about how smart phones are going everywhere

482
00:31:41,089 --> 00:31:41,210
>>Dr. Neeman: Mhmm.

483
00:31:41,210 --> 00:31:43,229
>>Dr. McGovern: – and places that you wouldn’t
have imagined either; places that don’t

484
00:31:43,229 --> 00:31:46,869
even have electricity, but they’re using
and finding new ways to charge them just so

485
00:31:46,869 --> 00:31:47,529
they can be connected.

486
00:31:47,529 --> 00:31:48,820
>>Dr. Neeman: That’s exactly correct.

487
00:31:48,820 --> 00:31:51,769
>>Dr. McGovern: So is there any other information
that we should tell our students, particularly

488
00:31:51,769 --> 00:31:52,820
if they’re interested in OSCER?

489
00:31:52,820 --> 00:31:57,039
>>Dr. Neeman: Right. So if you’re interested
in supercomputing, go to www.oscer – with

490
00:31:57,039 --> 00:32:04,300
an E, O-S-C-E-R – ou.edu. And you can learn
all about what we’re doing now – and if

491
00:32:04,309 --> 00:32:09,729
you want, you can contact us and we can help
you get started learning to use supercomputing.

492
00:32:09,729 --> 00:32:11,529
>>Dr. McGovern: Well thank you very much.

493
00:32:11,529 --> 00:32:12,800
>>Dr. Neeman: Thanks for having me!