The document discusses the Blue Gene supercomputer project by IBM. It describes Blue Gene as a massively parallel supercomputer using thousands of PowerPC processors to support large memory. The name "Blue Gene" refers to IBM's corporate color blue and the intended use of computational biology and protein folding. The project began in 1999 with a $100M effort to build a petaflop supercomputer. There have been four Blue Gene projects including BlueGene/L, the first system which achieved over 280 teraflops and set performance records.

1. Seminar
On
Blue Gene

2. Content
 What is Blue Gene
 Why the name “Blue Gene”?
 History
 Results
 Blue Gene Projects
 References

3. What is Blue Gene
 A massively parallel supercomputer using tens of
thousands of embedded PowerPC processors
supporting a large memory space
 With standard compilers
and message passing
environment

4. Why the name “Blue Gene”?
 “Blue”: The corporate color of IBM
 “Gene”: The intended use of the Blue Gene
clusters – Computational biology, specifically,
protein folding

5. History
 Dec’99, IBM Research announced $100M US effort to build
a Petaflop scale supercomputer.
 Two goals of The Blue Gene project :
 Massively parallel machine architecture and software
 Bio-Molecular Simulation – advance orders of magnitude
 November 2001, Partnership with Lawrence Livermore
National Laboratory (LLNL)

6. Results
 Linpack Top 500 Supercomputers

7. Blue Gene Projects
 Four Blue Gene projects :
 BlueGene/L
 BlueGene/C
 BlueGene/P
 BlueGene/Q

8. Blue Gene/L
 The first computer in the Blue Gene
series
 IBM first announced the Blue Gene/L
project, Sept. 29, 2004
 Final configuration was launched in
October 2005

9. Blue Gene/L - Unsurpassed Performance
 Designed to deliver the most performance per
kilowatt of power consumed
 Theoretical peak performance of 360 TFLOPS
 Final Configuration (Oct. ‘05) scores over 280
TFLOPS sustained on the Linpack benchmark.
 Nov 14, ‘06, at Supercomputing 2006, Blue
Gene/L was awarded the winning prize in all
HPC Challenge Classes of awards.

10. Blue Gene/L Architecture
 Can be scaled up to 65,536 compute or I/O
nodes, with 131,072 processors
 Each node is a single ASIC with associated
DRAM memory chips
 Each ASIC has 2 700 MHz IBM PowerPC
processors
 PowerPC processors
 Low-frequency, low-power embedded processors,
superior to today's high-frequency, high-power
microprocessors by a factor of 2 or more

11. Blue Gene/L Architecture contd…
 Double-pipeline-double-precision Floating Point Unit
 A cache sub-system with built-in DRAM controller
 Node CPUs are not cache coherent with one another
 FPUs and CPUs are designed for low power consumption
 Using transistors with low leakage current
 Local clock gating
 Putting the FPU or CPU/FPU pair to sleep

12. Blue Gene/L Architecture contd…
1024 nodes
System Overview

13. Blue Gene/L Architecture contd…
 1 rack holds 1024 nodes or 2048 processors
 Nodes optimized for low power consumption
 ASIC based on System-on-a-chip technology
 Large numbers of low-power system-on-a-chip technology
allows it to outperform commodity clusters while saving on
power
 Aggressive packaging of processors, memory and
interconnect
 Power Efficient & Space Efficient
 Allows for latencies and bandwidths that are significantly
better than those for nodes typically used in ASC scale
supercomputers

14. Blue Gene/L Networks
 Each node is attached to 3 main parallel
communication networks
 3D Torus network - peer-2-peer between compute nodes
 Collective network – collective & global communication
 Ethernet network - I/O and management (such as access to
any node for configuration, booting and diagnostics )

15. Blue Gene/L System Software
 System software supports efficient execution of
parallel applications
 Compiler support for DFPU (C, C++, Fortran)
 Compute nodes use a minimal operating system
called “BlueGene/L compute node kernel”
 A lightweight, single-user operating system
 Supports execution of a single dual-threaded application
compute process
 Kernel provides a single and static virtual address space to
one running compute process
 Because of single-process nature, no context switching
required

16. Blue Gene/L System Software contd…
 To allow multiple programs to run concurrently
 Blue Gene/L system can be partitioned into electronically
isolated sets of nodes
 The number of nodes in a partition must be a positive integer
power of 2
 To run program – reserve this partition
 No other program can use till partition is done with current
program
 With so many nodes, component failures are inevitable. The
system is able to electrically isolate faulty hardware to allow
the machine to continue to run

17. Blue Gene/L System Software contd…
 Parallel Programming model
 Message Passing – supported through an implementation of
MPI
 Only a subset of POSIX calls are supported
 Green threads are also used to simulate local concurrency

18. Thanks