Template 1ccarial Preview

1
One step ahead
2
The Challenges of Architectures that Grow to
Petascale and can be Sustained Economically

• Steve Reinhardt
• Principal Engineer, SGI
• spr at sgi.com

3

• SGIs systems are evolving to enable ultrascale
  versions of todays applications and enable a new
  type of computational science, while remaining
  economically sustainable.

4
Agenda

• Besides Architecture
• Enabling Ultra-scale Applications
• Enabling New Computational Science
• Sustaining Economically

5
Besides Hardware Architecture ...

• Efficient execution environment
• RAS
• OS architecture
• Linux scaled aggressively, with multiples in
  ultrascale configurations
• Robust scheduling
• RAS
• Packaging density / heat dissipation
• RAS

6
Agenda

• Besides Architecture
• Enabling Ultra-scale Applications
• Enabling New Computational Science
• Sustaining Economically

7
Local PerformanceNeeded Flexibility of Memory
Access
Absolute Performance
Note Original (Jan2003) models used for both X1
and Altix
8
Ideal Machine (Technical/Economic Balance)

• High, cost-effective cache bandwidth of mass
  market parts
• Highest cost-effective memory bandwidth
• Design focus on gather/scatter

Price Performance
Absolute Performance
Note For O(100KP) petascale machines, value of
O(5X) processor performance advantage is less
than today
9
Local Performance Multi-Paradigm
Low Compute
high Intensity
Low Data locality
High
10
Ultraviolet Concept Architecture

APU
UV Petascale GAM . Globally Addressable
. Low Latency . High Bandwidth .
O(100K) Ports
I/O
GPU
11
Global Performance

• Communications
• grids becoming more dynamic -gt low latency
  essential
• processor counts growing -gt low latency
  essential
• low latency -gt global address space
• in clock periods, remote memory getting further
  away
• bandwidth-conserving operations needed
• high absolute link performance
• Synchronization
• current mechanisms insufficient for ultrascale
• optimizations will help, but maybe not enough
• new mechanisms needed
• Dynamic load balancing
• mechanisms need to mature, and interfaces become
  standard

12
Challenges

• Clear virtual machine and performance models for
  these new mechanisms
• Compilers/tools that exploit these mechanisms
  mostly automatically and accept user hints
• Appropriate performance balance for typical uses
• Need to gain successful experience at very large
  scale (10-30KP) before going to ultrascale (100KP)

13
Agenda

• Besides Architecture
• Enabling Ultra-scale Applications
• Enabling New Computational Science
• Sustaining Economically

14
Scientific Process
Observe existing data for patterns
Hypothesize models that match the data
Test those models to understand accuracy (i.e.,
add new data)
Believed first coined by Scott Studham et al.,
PNNL
15
Scientific Process
Observe existing data for patterns
Hypothesize models that match the data
Test those models to understand accuracy (i.e.,
add new data)
Believed first coined by Scott Studham et al.,
PNNL
16
Example Post-Genomic Biology

• lt10 of the human genome is known to code for
  proteins
• Selective pressure generally removes unused
  genetic material
• What is the other 90 of the genome doing?
• Have the raw data (genome)
• Need to add other types of data (e.g., protein
  association info)
• Multi-petabytes of data all told
• Probably not a purely computational problem

17
Differences from First Principles

• Data access patterns impossible to predict a
  priori -gt low latency / global
  address space
• New tools for data exploration needed
• need to automatically search for new,
  perhaps-vaguely-defined, patterns (that foster
  new theory)
• highly interactive/coupled with the scientists
  thought process
• but beware difficulty of launching new languages
• Contents of memory much more valuable
• RAS

18
and now for something completely different
Star-P

• Developed by Alan Edelman and colleagues at MIT,
  etc.
• Simple extensions to the MATLAB language
• data parallel, MIMD, and mixed
• Builds on the existing base of MATLAB programs
• broadening the market for HPC systems
• New back-end server implemented for parallel
  execution
• Preserves key MATLAB strengths
• very high level language
• interactivity / exploration
• easy visualization

Put the fun back in supercomputing
19
Agenda

• Besides Architecture
• Enabling Ultra-scale Applications
• Enabling New Computational Science
• Sustaining Economically

20
Key Points

• SGI retains system focus
• but uses commodity components wherever practical
• Exploit best mass-market processors (Itanium)
• augment to make suitable for wider range of HPC
  apps
• Use Linux fully
• reap the cost benefits of reduced support of
  proprietary Unix variant
• IFB cables, EFI firmware
• Innovations for ultrascale must be relevant for
  wider markets
• e.g., multi-paradigm computing must accelerate
  ISV apps
• Use new technologies to broaden the market
• e.g., Star-P

21

• SGIs systems are evolving to enable ultrascale
  versions of todays applications and enable a new
  type of computational science, while remaining
  economically sustainable.

22
One step ahead
23

• There are no technology-independent lessons in
  computer science.
• Butler Lampson, Xerox PARC