Keynote : Introduction : Xeon!
The third day of IDF was opened with a Keynote which was aimed at the Enterprise and communication market. This was held with 2 core spokes persons, Mike Fister (Senior VP), and Sean Maloney (Exec VP). There was a strong present for promotion of WiFi, PCI-Express, Itanium, Hyper Threading and AdvancedTCA. Intel touted some of these technologies as essential, since failure to implement could lead to issues within the Internet backbone. Since revenue streams are pretty much stagnant, and the bandwidth usage is doubling each month.
Mike Fister was our host for today
Intel claimed that the speed of sales within the Xeon platform will be higher in Quarter 1 of 03, than the entire of 2002. This is a pretty good result since the market is still within a lull over the entire .com over investment. The Xeon CPU operates within the high-end desktop, workstation, and server markets - it is down to different implementations to suit each need.
Intel is planning to increase the cache within the Xeon to 1MB towards the end of the year, and bring out a 90nm unit. The reason for Intel wanting to scale the memory architecture is that they wanted to balance this towards the CPU speed.
To show that the implementation of the Xeon CPUs within certain areas makes a large difference to the performance of the CPU, Intel did a demonstration which they ran 2 Xeon based systems but in 2 different configuration.
2 Node - 2 CPU DP Processor (2GHz Speed) - Cluster Environment.
4 Way - 4 CPU SMP Processor (2.8GHz Speed) - SMP Environment.
The benchmark which was run evolved doing database manipulation.
When the benchmark was run, it showed that the Cluster system took 39 Seconds to finish the test, and the SMP system took 16 seconds. This is all down to the implementation of the system. Different configurations are faster for different things, the database style environment benefits from the close proximity of the CPU within the system bus, and of course of the higher clock speeds. Thus the SMP system won. The drawback within the cluster environment is that it has several software layers which are there to retain data integrity.
