AMD Stream SDK is the software development kit for AMD's Stream computing technology.
AMD Pioneers Stream Computing on Graphics Processor Units
Stream computing allows AMD's graphics processors to accelerate complex computations working in concert with today's high-performance, low-latency CPUs.
Along with leading companies and academic institutions worldwide, AMD is working to build a complete stream computing ecosystem, one that delivers the performance, applications, software and tools necessary to turn AMD's low-cost stream computing vision into reality.
Fortune 1000 companies, leading software developers, and academic institutions are demonstrating today that stream computing is the key to tremendous performance in a number of applications.
What is Stream Computing
Stream computing harnesses the tremendous processing power of graphic processors for high-performance, data-intensive computing in a wide range of scientific, business and consumer applications, providing organizations the ability to process massive amounts of information in significantly less time.
Stream Computing Applications
The accelerated processing associated with stream computing has implications for a number of fields now and in the future, as the ecosystem around stream computing matures:
Life Sciences & Scientific Research
Life sciences & scientific researchAMD's stream computing efforts are contributing to the life sciences by enabling faster results in areas such as disease research, giving organizations the opportunity to do more granular studies in the same amount of time as in the past.
Stanford University uses the distributed computing application, Folding@Home, taking advantage of AMD graphics processors for disease research. Learn more about Folding@Home.
Climate research is another area of interest with stream computing. Analysis of large data sets for storm and hurricane forecasting can be done faster or in more detail, potentially resulting in the issuing of severe weather warnings further in advance, and ultimately resulting in a better understanding of the world's climate.
Enterprise Applications
Enterprise ApplicationsStream Computing is being used in a wide range of compute-intensive enterprise applications to deliver increased return on R&D investment and to ultimately improve profitability for the organization.
Enterprises in financial services, manufacturing, energy (oil and gas), content creation, mining, construction, and defense, and governments will greatly benefit from the power of AMD's stream computing products.
Major institutions have been using server farms to do risk assessment using Monte Carlo simulations, and for derivatives pricing using models like Black-Scholes.
Simulations conducted by PeakStream, Inc. (now owned by Google) using AMD hardware show that stream computing can provide these companies with more detailed answers in significantly less time, letting them make faster business decisions, giving them a competitive advantage.
Simulated processing of risk assessment models similar to those used by financial institutions were completed 16 times faster than traditional methods (1).
Oil and gas companies are using stream computing to analyze more data in shorter periods of time to more quickly and reliably discover where resources lie, speeding discoveries of crude oil deposits. AMD graphic processors on PeakStream's software platform are allowing oil and gas companies to achieve 20 times faster seismic data modeling (1).
In homeland security, communications analysis and facial recognition can be drastically improved using stream computing, with implications for airport security, as well as photograph and video analysis.
For web search companies with incredibly large databases to organize and sort through, stream computing may offer a compelling business case providing increased processing power in less space.
AMD's Stream Computing Software Stack
AMD has developed a comprehensive software stack for stream computing.
(1) Based on simulations performed by PeakStream, Inc. using traditional processing models vs. using AMD's ATI processors.
Requirements:
· ATI Radeon Linux Display Drivers 8.5 or better
Installation:
Install ATI driver
1. As root, run the script "ati-driver-installer-8.49.4-x86.x86_64.run"
2. Run "aticonfig" (or "aticonfig --initial" if it is the first time.)
3. reboot
Install CAL
1.Run CAL installer script "amdcal-1.01.1_beta.x86_64.run" (default cal location is /usr/local/amdcal)
2. CAL needs libstdc++.so.5. If that is not present (eg, in Redhat Enterpise Linux 5.1), you will need to install lib compat-libstdc++-33-3.2.3-61.i386.rpm, and also compat-libstdc++-33-3.2.3-61.x86_64.rpm
Install Brook
1. run Brook installer script "amdbrook-1.01.0_beta.x86_64.run" (default Brook location is /usr/local/amdbrook)
Setup environment variables
1. Add the CAL and brook library paths
Eg, in a bash shell,
export LD_LIBRARY_PATH=/usr/local/amdcal/lib64/:/usr/local/amdbrook/sdk/lib:$LD_LIBRARY_PATH
Notes:
So far we have tested on Suse Enterprise Server SP1 (no issues), and on Redhat Enterprise Linux 5.1 (need to install extra rpm's to get libstdc++.so.5).
Product's homepage
Find us on Google+
