Modern and Future Processors

Download Report

Transcript Modern and Future Processors

Dr. Gheith Abandah, Chair
Computer Engineering Department
The University of Jordan
20/4/2009
1




Multicore Trend
Current Performance Improvement
The Parallel Programming Challenge
The Future
20/4/2009
2


Modern processor chips contain processing
cores and levels of memory caches.
A multicore processor is an integrated
circuit (IC) that has two or more processors
for
◦ enhanced performance,
◦ reduced power consumption,
◦ and more efficient simultaneous processing of
multiple tasks.
20/4/2009
3
Company
Processor
Year
Cores
IBM
Power 4
2001
2
Power 5
2003
2
Power 6
2008
2
Cell
2006
9
Sun
UltraSPARC T2
2007
8
Intel
Core 2 Duo
2006
2
Itanium 2 Duo
2006
2
Core 2 Quad
2007
4
Athlon
2005
2
Opteron Quad
2006
4
AMD
20/4/2009
4
276 million
transistors
20/4/2009
5
20/4/2009
6
234 million transistors
20/4/2009
7
20/4/2009
8
20/4/2009
9
20/4/2009
10
20/4/2009
11




Multicore Trend
Current Performance Improvement
The Parallel Programming Challenge
The Future
20/4/2009
12
20/4/2009
13
20/4/2009
14




Multicore Trend
Current Performance Improvement
The Parallel Programming Challenge
The Future
20/4/2009
15


All modern
computers are
parallel computers,
often with 2 or
more processors.
Most user
applications are
serial applications
that use one
processor.
20/4/2009
16


If one core is going to do all your work, why
would you buy new processors with more
cores?
Processor manufacturing companies are:
◦ Promoting parallel programming by providing
tools and training.
◦ Supporting research for innovative ways to utilize
the multiple cores.
20/4/2009
17

Parallel programming is harder than normal
programming because it involves:
◦
◦
◦
◦
◦

Additional techniques
Problem partitioning
Synchronization
Access control
…
90% of programmers don’t do parallel
programming.
20/4/2009
18




Multicore Trend
Current Performance Improvement
The Parallel Programming Challenge
The Future
20/4/2009
19






Research on better semiconductor materials
to build smaller, faster, and cooler
transistors.
Fine power management.
Keep tweaking the cores for more
performance optimizations.
Multiple cores are here to stay.
More and larger caches.
Compilers that generate parallel threads
automatically.
20/4/2009
20