© vipin kumar ipdps-2011, may 18, 2011 1 25 th year panel – what’s ahead vipin kumar university...
TRANSCRIPT
© Vipin Kumar IPDPS-2011, May 18, 2011 1
25th Year Panel – WHAT’S AHEAD
Vipin Kumar
University of Minnesota
[email protected] http://www.cs.umn.edu/~kumar
© Vipin Kumar IPDPS-2011, May 18, 2011 2
Applications
What will be the next wave of grand challenge problems to focus on in the next 10 years and beyond ?
© Vipin Kumar IPDPS-2011, May 18, 2011 3
Compute centric to Data Centric
Compute Intensive
Data Intensive
Transition:
© Vipin Kumar IPDPS-2011, May 18, 2011 4
Compute centric to Data Centric
Compute Intensive
Data Intensive
Transition:
Enabled by 6 decades of exponential growth in computing power, storage capacity, networking and more recent development of the Internet technology, data and compute clouds
© Vipin Kumar IPDPS-2011, May 18, 2011 5
Compute centric to Data Centric
Compute Intensive
Data Intensive
Transition:
© Vipin Kumar IPDPS-2011, May 18, 2011 6
Great Challenges Facing the Society
Improving health care and reducing costs
Finding alternative/ green energy sources
Predicting the impact of climate change
Reducing hunger and poverty by increasing agriculture production
© Vipin Kumar IPDPS-2011, May 18, 2011 7
Scalable Data Analysis
General Circulation Models: Mathematical models with physical equations based on fluid dynamics
Figure Courtesy: ORNL
An
om
alie
s fr
om
188
0-19
19 (
K)
Parameterization and non-linearity of differential equations are sources for uncertainty!
Cell
Clouds
LandOcean
Example: Understanding Climate Change
Detection of Global Dipole Structure
© Vipin Kumar IPDPS-2011, May 18, 2011 8
Most challenging algorithmic problems
Dense vs. Sparse
Structured versus Unstructured
Static vs. Dynamic
Data intensive computations tend to be unstructured, sparse and dynamic
Restructuring algorithms for locality key to scalability Crucial in the context of emerging architectures based on multi-core,
GPUs,…
© Vipin Kumar IPDPS-2011, May 18, 2011 9
Software
How will we ever be able to hide parallelism obstacles for the masses and what is the road forward towards that ?
© Vipin Kumar IPDPS-2011, May 18, 2011 10
Computing Platforms
How will we be able to keep improving the performance growth of the past and what will be the major challenges in the next 10 years and beyond that ? What technical barriers are anticipated and what disruptive technologies are behind the corner ?
© Vipin Kumar IPDPS-2011, May 18, 2011 11
Do We Need Benchmarks Specific to Data Intensive
Computing ?
Performance metrics of several benchmarks gathered from Vtune• Cache miss ratios, Bus usage, Page faults etc.
Benchmark applications were grouped using Kohenen clustering to spot trends:
012
345
8
9
11
ap
riori
bayesia
nb
irch
ecla
th
op
scalp
arc
kM
ean
sfu
zzy
rsearc
hsem
ph
ysn
pg
en
en
et
svm
-rfe
MineBench
67
10
Clu
ste
r N
um
ber
gcc
bzi
p2
gzi
pm
cf
twolf
vort
ex
vp
rp
ars
er
ap
si
art
eq
uake
lucas
mesa
mg
rid
sw
imw
up
wis
era
wcau
dio
ep
icen
cod
ecjp
eg
mp
eg
2p
eg
wit
gs
toast
Q1
7Q
3Q
4Q
6
SPEC FP MediaBenchTPC-HSPEC INT
Reference: [Pisharath J., Zambreno J., Ozisikyilmaz B., Choudhary A., 2006]