Testbed and Methods

The goal of our today’s testing session is checking out the performance level of the new chipset from NVIDIA, the nForce3 250. We compared Chaintech Zenith ZNF3-250 with mainboards on other Socket 754 chipsets: NVIDIA nForce3 150, VIA K8T800 and SiS755.

The testbed was configured as follows:

• Processor: AMD Athlon 64 3400+ (2.2GHz);
• Mainboards:
  • Chaintech Zenith ZNF3-250 (NVIDIA nForce3 250);
  • Chaintech Zenith ZNF3-150 (NVIDIA nForce3 150);
  • ABIT KV8-MAX3 (VIA K8T800);
  • ECS 755-A2 (SiS755);
• 1GB DDR400 SDRAM (2-3-2-6 timings);
• ATI RADEON 9800 XT graphics card (Catalyst 4.3 driver);
• Western Digital WD400JB HDD.

We ran our tests in Windows XP Professional SP1 with DirectX 9.0b installed. The BIOSes of the tested mainboards were set up for maximum performance.

The table below shows effective frequencies the processor was working at in the tested mainboards (as measured by the CPUZ diagnostic utility):

As you see, the difference is negligible, its influence on the test results can be dismissed.

Performance

Before discussing the test results, I think it is necessary to comment on the performance of the Athlon 64 processor in general. The architecture of Athlon 64 CPUs differs somewhat from other processors and diminishes the contribution of the chipset to the overall system performance.

Really, the memory controller constitutes the major part of any chipset for the Pentium 4 or Athlon XP processor. Traditionally, developing an efficient memory controller meant developing a fast chipset.

Athlon 64 architecture, however, puts the memory controller into the processor itself. Thus, all Athlon 64 platforms work with the memory at the same speed and show similar benchmark results in a majority of applications. Well, there are cases when bugs in the BIOS lead to an incorrect memory controller configuration, which results in a low performance, but this is in no way related to the chipset. Thus, there are other factors that determine the fastest platform for Athlon 64: efficient routing of data streams in the chipset, quality of HyperTransport and AGP buses and so on. However, these factors affect the resulting performance of the platform just slightly. So the fact that different chipsets for Athlon 64 show very similar level of performance shouldn’t surprise you.

There is also one more consequence to the things I’ve mentioned above: when evaluating a chipset for the Athlon 64 platform, you should always remember that the speed of the chipset is not the decisive factor as to its superiority over the competing products. The set of characteristics and overclocking opportunities are also two important factors for an Athlon 64 chipset.

The preliminaries over, let’s get to our tests.