Accueil > Produits et services > Composantes matérielles > 

About AMD dual core processor (pdf)

About Intel dual-core processor and Hyper-Threading (HT) Technology (Flash Demo)

RT-LAB Support for AMD's New Dual-Core Opteron Processors

RT-LAB has been fully tested on the recently launched dual core AMD Opteron processor. The AMD Opteron processor is an x86 dual core processor for 32-and 64-bit computing.

The below benchmark shows the comparison between AMD and Intel.

Benchmark: RT-LAB Model with Two Subsystems Running on RedHawk Linux

The results in the graph indicate the minimum step size of the RT-LAB model running without overruns and exchanging data through a shared memory. This test was executed on:

  • Intel Xeon 3.0 Ghz, 1Go RAM
  • AMD dual core Opteron 2.2 Ghz, 1 Go RAM

RT-LAB models run three times faster on the AMD Opteron than on the Intel Xeon. The incremental bandwidth is about 3.2 Gbits for the data exchange between the two subsystems.

Application Example: Kundur Network

The Kundur power system consists of two fully symmetrical areas linked together by two 230 kV lines of 220 km length. It was specifically designed in [1,2] to study low frequency electromechanical oscillations in large interconnected power systems. Despite its small size, it mimics very closely the behavior of typical systems in actual operation.

The electromagnetic transient type of simulation made in RT-LAB enables the study of fast and detailed phenomena like single-phase faults in the Kundur network and observe their effects on a larger time scale, i.e. on the electromechanical scale, like inter-area power oscillations.

[1] P. Kundur, Power System Stability and Control, McGraw-Hill, 1994, Example 12.6, p. 813

[2] Klein, Rogers, Moorty and Kundur: "Analytical investigation of factors influencing PSS performance," IEEE Trans. on EC, Vol. 7 , No 3, September 1992

CPU Assignment on the AMD Dual Core Opteron 2.2Ghz

CPU 0: Operating system
CPU 1: 2 Machines + local network
CPU 2: 2 machines + local network ( i.e. electrical part is splitted equally over 2 cpus)
CPU 3: Control subsystem for all the machines

Calculation time of CPU1: 14 µs
Calculation time of CPU2: 13 µs
Calculation time of CPU3: 12 µs

The model is well distributed over the CPUs in terms of computing power.

CPU Assignment on Intel Dual-Xeon 3.0 Ghz Hyperthreaded

CPU 0: Operating system
CPU 1: 2 machines + control
CPU 2: 2 machines + control
CPU 3: down

Kundur network running on RedHawk 2.2

Intel Xeon 3.0Ghz
AMD Opteron Dual Core 2.2 Ghz
50 µs
18 µs

The timing above is the minimum time step of simulation without overruns.