|
The SGI NUMAlfex shared-memory architecture enables Altix 3000 servers and superclusters to distribute calculations over an array of nodes.
Marathon Oil is involved with the first two of three steps in the well management process, all related to estimating oil and gas reserves in the subsurface and predicting how quickly the company can produce those reserves:
- the geocellular model, a static model describing the rocks and fluid properties of the field,
- reservoir simulation, a fluid dynamic model characterizing how fluids flow through and from those rocks, and
- the economic model, enabling the company to efficiently manage the monetary resources required to produce the hydrocarbons.
"The first two steps allow Marathon to look at a number of possible production scenarios - how many platforms do you build, when do you build them, how much do you spend, etc. - and selecting the best one to produce the optimum rates of oil", explained Mark Petersen. The third step models the cash flow from the project based on each of the scenarios under consideration.
Scalable SGI Altix 3000 systems are available today in server configurations of 4 to 64 processors, and supercluster configurations of 4 to 512 processors. For customers demanding even larger Altix superclusters, SGI plans to support configurations of 1,024 processors in May 2004 and larger over time.
With global shared memory across cluster nodes, Altix 3000 superclusters will scale to up to thousands of processors. Such supercluster capabilities leverage the built-in SGI NUMAlink interconnect fabric, delivering data across nodes up to 200 times faster than conventional clustering interconnects. For the oil and gas customers this means that larger, more complex reservoir models than ever before can be run directly out of one main memory. This, in turn, will enable faster and better solutions for maximizing oil field yields. |
