Most Read Latest News Blog Resources

IBM emphasizes speed with new Milepost GCC compiler




July 1, 2009 — 
IBM has created what it is calling the world’s first open-source machine learning compiler.

The Milepost GCC (GNU Compiler Collection) compiler, released yesterday, attempts to significantly reduce software development times, according to IBM executives. The objective is to develop compiler technology that can automatically learn how to best optimize programs for reconfigurable embedded processors.

Initial IBM experiments with the Milepost GCC conducted on IBM System p servers achieved an average 18% performance improvement on embedded application benchmarks, according to IBM.

With Milepost GCC, applications can be more quickly tuned for a specific architecture. For example, if a company wants to develop a new mobile phone, it normally takes application developers months to get their software running properly.

“Our technology automatically learns how to get the best performance from the hardware, whether mobile phones, desktops or entire systems; the software will run faster and use less energy,” said Bilha Mendelson, manager of code optimization technologies at IBM Research.

“We opened the compiler environment so it can access artificial intelligence and machine learning guidance to automatically determine exactly what specific optimizations should be used and when to apply them to ramp up performance.”

IBM created Milepost GCC in collaboration with the Milepost consortium, which includes the IBM Haifa Research Lab, the University of Edinburgh, audio provider ARC International and embedded application company CAPS Enterprises.

In addition to the compiler, a code-tuning website called Collective Tuning Center is available, letting developers upload software code and get input on how to configure the code for faster operation.

In explaining how the code-tuning site works, Mendelson noted that the Tuning Center website is connected to the Milepost GCC to allow remote tuning of applications.

“The compiler extracts some characteristics of the compiled application that are collected and fed to the Machine Learning (ML) engine,” Mendelson said.

“Also, as part of the training of the ML, performance data of a set of related benchmarks is collected with the set of features associated with it. This data is passed to the code-tuning part. It includes a database that captures this data and the ML engine and models that advised the compiler with the right set of optimizations to employ.”


Related Search Term(s): IBMopen source


Share this link: http://www.sdtimes.com/link/33588
 

Add comment


Name*
Email*  
Country     


  • Comment
  • Preview
Loading



 
 
 
 
News on Monday
more>>
SharePoint Tech Report
more>>


   

 
 
Download Current Issue
ISSUE 3/15/2010 PDF

Need Back Issues?
DOWNLOAD HERE

Receive the print Edition?


 
blogs tab
Google Code turns 5
Google Code Turns 5, and adds a Paxos Algorithm to make the system more stable and reliable.
03/17/2010 11:16 AM EST

Test your Visual Studio 2010 know-how
Microsoft is offering free beta certification exams for Visual Studio 2010.
03/17/2010 11:08 AM EST

Microsoft lifts the hood on IE9
Microsoft is previewing IE9.
03/16/2010 01:10 PM EST

 

Events calendar tab
3/22/2010 to 3/25/2010
Santa Clara, Calif.
The Eclipse Foundation

4/12/2010 to 4/14/2010
Las Vegas
Penton Media

4/12/2010 to 4/15/2010
Santa Clara, Calif.
O'Reilly Media

4/19/2010
New York City
Flagg Management

4/25/2010 to 4/28/2010
Overland Park, Kans.
IIUG