VIA KT333 in the test: EPoX 8K3A + against MSI KT3 Ultra ARU

VIA KT333 in the test: EPoX 8K3A + against MSI KT3 Ultra ARU

Table of contents
  1. 1 Introduction
  2. Addendum: Elitegroup K7S6A
  3. Are you spoiled for choice?
  4. 2 EPoX 8K3A +
    1. Scope of delivery & documentation
    2. Equipment/layout
    3. Bios/overclocking
  5. 3 MSI KT3 Ultra ARU
    1. Scope of delivery & documentation
    2. Equipment/Layout
    3. Bios/Overclocking
  6. 4 Test system
    1. SiSoft Sandra 2002 - Memory Test
    2. WinACE 2.11
    3. CineBench 2000
    4. 5 Quake3Arena
    5. Vulpine GLMark v1.1
    6. 6 3DMark 2000 v1.1
    7. CPUMark
    8. 3DMark 2001
  7. 7 Conclusion VIA KT333
  8. Rating EPoX 8K3A +
  9. Rating MSI KT3 Ultra (ARU)


It has been some time since we started with the K7S6A and the SiS745 We took a close look at the first chipset with DDR333 (PC2700) support and we still remember the somewhat ambiguous results. Above all, the RAM seemed to have thwarted our plans at the time. Armed with three different RAM modules of very good quality, we now wanted to take a look at the latest offspring from VIA, the KT333. In our review, we wanted to find out whether the chipset is worthwhile for DDR RAM beginners or owners of a KT266 (A) despite its otherwise identical properties to the predecessor. At this point, many thanks to EpoX and MSI , who made the mainboards 8K3A + and KT3 Ultra available to us. Although it is firmly planned, the Asus A7V333 no longer has itPassed our test.

8K3A + & MSI KT3 Ultra ARU

Addendum: Elitegroup K7S6A

As announced in the review at the time, we also sent the K7S6A with the latest BIOS and high-quality memory (Corsair) to the Test course to straighten out the partially desolate performance of the board with questionable bios and relatively picky Nanya RAM. But although this time we were able to activate the absolutely sharpest timing in the BIOS, the values ​​achieved with the PC2700 remained practically on the same, poor level. Since this fact has to be attributed to the board, we decided to devalue the board. The complete review can be found here .

The agony of memory choice?

Much has been written about the advantages and disadvantages of the latest generation of chipsets with PC2700 support. Hailed by the manufacturers as a further milestone in computer development, we were very skeptical even before our review of the K7S6A with SiS745 chipset. At that time, the RAM used promptly became our undoing, which made a perfect evaluation of the board impossible.

In order to be able to get the most complete overview of the maximum performance of the KT333 this time, we have no fewer than three Various RAM brands were used, which were able to prove their good to very good performance in numerous tests. We would like to take this opportunity to thank the companies OCZ , Crucial and Corsair for providing the modules.

Crucial DDR333 CL2.5
Corsair DDR333 CL2.0
OCZ DDR400 CL2.5

However, this time it did not go without complications either. First of all, the circuit boards had to make do with the well-known Crucial memory. The RAM is specified to CL2.5 and conforms to JEDEC. Both the EpoX 8K3A + and the MSI KT3 Ultra ARU therefore promptly refused to work with CL2.0. However, if the 8K3A + was satisfied with the setting 2.5 with otherwise fastest timing, the MSI KT3 Ultra wanted to start with an 'Active to Precharge' of 6. The board was therefore unable to fully use the memory, which in various tests with optimal timings ran far above the actual clock frequency>

The RAM provided to us by Corsair should help here. Officially specified up to 180MHz on CL2.0, the memory showed its capabilities at up to 200MHz in some reviews. The first tests with the 8K3A + were positive in any case, so that nothing stood in the way of a test with perfect memory timings, i.e. the maximum possible chipset performance. The KT3 behaved differently, which did not want to drive a CL 2.0 with the Corsair RAM either. Windows 2000 booted, but was extremely unstable and at the latest the first benchmark brought the involuntary blue screen/reset. In order to be able to ascribe this fact to the board and not to the memory, the MSI board was quickly equipped with a DDR400 module (CL2.5) from OCZ. But even here the board said goodbye after a few seconds with a reset. After all, the equally important Active to Precharge Time could be set to 5 with both RAMs.

Should our duel therefore have different requirements in the memory timingsoccur? A look at the results seems to invalidate this and at the same time explain the sobering behavior of the MSI KT3 Ultra with a CAS latency of 2.0. So much in advance: The board running on 2.5 should prove to be only slightly slower than the 8K3A + running on 2.0, which suggests that the chip on the MSI board is already clocked with sharper timing than the BIOS want to reveal. The EpoX 8K3A +, on the other hand, is starting out rather conservatively, but for this reason it can be tuned a lot more.

But let's take a look at the test candidates for now.

On the next page: EPoX 8K3A +