Menu
Intel Pentium 4 HyperThreading-Spezial: Benchmarks in multitasking scenarios

Intel Pentium 4 HyperThreading-Spezial: Benchmarks in multitasking scenarios

Multitasking with Seti

In order to be able to use Seti @ Home as a fair background application, we had to make sure that the system m it and the system without Hyper-Threading were loaded identically during the test. However, since Seti almost never goes into idle mode on a multi-processor system, it has its own processor, even if it is only of a virtual nature, we had to set the task priority from Seti in this test set to normal. Ultimately, the Hyper-Threading system managed more after the benchmarks were completed than the system with deactivated Hyper-Threading. So there are not only the disadvantages and advantages that will become clear in the following .

and Quake 3 Arena

Seti and Quake 3 Arena
  • Normal:
    • Without load
      353.6
    • P4 3.06 HT an
      250.9
    • P4 3.06 HT off
      237,9
  • Max:
    • Without load
      301.6
    • P4 3.06 HT an
      196,3
    • P4 3.06 HT from
      179,9

Certainly the frame rates achieved with Seti in the background are only a shadow of the 350fps achieved in the individual test, but 13 frames per second can be achieved with Hyper-Threading save more to the goal. That makes 5.5 percent (9.1 for max details) more performance in Quake 3 alone, and Seti managed even more of the WU on the virtual dual processor system. Not bad for the beginning.

and 3DMark2001

Seti and 3DMark2001 SE
Unit: Points
    • Without load
      13.099
    • P4 3.06 HT on
      10.234
    • P4 3.06 HT off
      7.351

The loss due to Seti without Hyper-Threading is more than 40 percent. With Hyper-Threading, this loss is just over 20 percent. Hyper-Threading can significantly increase the performance when using Seti and the 3DMark2001 SE at the same time.

and UT2003

Seti and Unreal Tournament2003
  • Flyby:
    • Without load
      162,3
    • P4 3.06 HT an
      139,7
    • P4 3.06 HT off
      78,5
  • Botmatch:
    • Without Load
      75.9
    • P4 3.06 HT an
      50,8
    • P4 3.06 HT from
      31,4

Without Hyper-Threading, you lose more than 50 percent of the original performance in UT2003 due to the simultaneous operation of Seti. Hyper-threading can at least partially improve this relationship. You only lose 14 percent of your actual performance in the flyby and only 33 percent in the botmatch. Hyper-Threading increases the performance by 60 to 78 percent compared to a processor with the same clock speed without Hyper-Threading. If you consider the values ​​as 100% without Hyper-Threading, you achieve 178 percent of this performance in the flyby and 160 percent in the botmatch.

and WinACE

Seti and WinACE
Unit: minutes, seconds
    • Without load
      2:49
    • P4 3.06 HT an
      3: 54
    • P4 3.06 HT off
      4:00

WinACE still loses around 30 percent due to the simultaneous operation of Seti, even with activated Hyper-Threading its original performance. The performance with Hyper-Threading is still 2.5 percent higher than the performance without Hyper-Threading. WinACE and Seti do not harmonize optimally, so that the increase through Hyper-Threading is rather small.

On the next page: Multitasking with Media Player