|
Comments
Did you read today's front page stories & breaking news?
SYS-CON.TV
|
From the Wires
Fast GPUs Accelerate Performance of In-Memory OLAP Server
Open-Source Approach Enables Successful and Efficient Collaboration Between Software Producer and Universities for Open-Source Jedox's Palo OLAP Server
By: Marketwire .
Feb. 3, 2009 11:00 AM
FREIBURG, GERMANY -- (Marketwire) -- 02/03/09 -- As Microsoft exits the OLAP software for business planning market with an announcement to discontinue Performance Point Servers, Jedox announces a breakthrough with regard to planning performance of OLAP servers. The research project by Jedox and the Albert-Ludwig's University in Freiburg -- sponsored by DFG, the German Research Foundation -- was performed to optimize access methods for multidimensional data in memory-based OLAP databases. The development is based on the open-source Jedox's Palo OLAP Server. By utilizing fast GPUs originally designed for computer games, the team of researchers and developers was able to make the breakthrough and accelerate the performance of OLAP analyses with Palo by up to 40 times of its current performance. A prototype of this GPU version of Palo will be introduced by Jedox at this year's CeBIT at their stand in hall 4, D 53. Additionally, Jedox will also present the new release of its multiprocessor version of Palo 3.0. The Research Team, an International Cooperation Presently, Professor Amitava Datta of the University of Western Australia in Perth is in Freiburg as a guest researcher in connection with the DFG project. The renowned computer science professor is a guest of Jedox and is collaborating with their developer team and the researchers of the Albert-Ludwigs-University. Central subject of the Palo project is the development of efficient algorithms for memory-based OLAP applications. Professor Datta is an expert in the fields of algorithmic and parallel computing and for quite some time, has been exploring additional application possibilities utilizing the high-performance GPUs. With his expertise and involvement in the first-hand and international exchange, he is bringing optimal qualifications to the team of researchers and developers in Freiburg. Today, the performance of databases is a central factor in successful business solutions which, for fast decision-making, need to be able to analyse huge amounts of data ad hoc. Professor Datta and Jedox take advantage of the latest developments in the hardware field: instead of conventional processors (CPU), the scientific field has started utilizing GPUs to deal with complex computations. Modern graphics cards consist of many small processor cores which parallelize and consequently speed up computations. Chip manufacturers such as Nvidia and AMD introduced graphic processors featuring higher processing power than conventional CPUs in 2008. What makes the GPU so fast? To illustrate the difference in computing and the resulting speed, here is a practical example: Using the conventional processor architecture, an Easter bunny planning the distribution of Easter presents would first determine which houses are occupied by children. Then, he would calculate the optimal route in a relatively elaborate manner and subsequently, he would carry his gifts from house to house along the route. In the "new world" of GPUs, tens of thousands of bunnies would be sent off to all houses simultaneously without prior route calculation. Arriving at a house, each bunny would check if children are present and if so, drop off the goodies. This approach is less efficient for each of the Easter bunnies, but, due to the huge amount of available bunnies (i.e. GPU processors), much faster overall. Graphic Cards in the OLAP field: Effects in practice The Palo project utilizes graphic cards by Nvidia featuring an excellent programming interface. Palo is the ideal basis to pursue new methods by using graphic cards in the OLAP field: on one hand because the open source code of the Open Source solution accommodates the broad international research and development endeavours; on the other hand because of Palo's in-memory architecture, which is ideal for the use of the new GPUs with increasingly bigger graphics memory. "In-Memory OLAP servers are ideal for the utilization of the fast graphic cards, and we are glad to have found one of the first commercial application fields for GPUs in the area of Business Intelligence," Prof. Datta of the University of Perth explains. So far, unknown speed in accomplishing complex calculations or the improved performance for decentralized business processes with many parties involved, e.g. in corporate planning, will have a tremendous effect in day to day business. About Jedox Jedox, founded by Kristian Raue in 2002, is one of the world's leading suppliers of open source-based Business Intelligence and corporate performance management solutions. Jedox's core product, Palo, accommodates the entire range of BI requirements including planning, reporting and analysis. Twenty per cent of DAX 30 companies and many other organizations in Germany and abroad have implemented Jedox solutions including well-known companies such as Quelle, Clariant Switzerland and Nycomed Germany. Internationalization and diversification in the field of Commercial Open Source Software are main objectives for continued growth. The company employs 60 people at its headquarters in Freiburg. PRESS CONTACT: Latest Cloud Developer Stories
Subscribe to the World's Most Powerful Newsletters
Subscribe to Our Rss Feeds & Get Your SYS-CON News Live!
|
SYS-CON Featured Whitepapers
Most Read This Week
Breaking Cloud Computing News
|
|||||||||||||||||||||||||||||||||||||||||||||||||