Tyan S2495 Trinity KT400
More KT400 Madness

By, Tom Laverriere
February 17, 2003

TESTING METHODOLOGY:

We're sure you have seen significant variations in benchmark scores from one site to the next.  As such, we feel it is necessary to explain exactly how we configure our test systems, before running any benchmarks. When testing these boards, the first thing we did was enter the system BIOSes and set each board to "Load Optimized Defaults" and configured our RAM to run at 166MHz (DDR333).  The other memory timings were set by the SPD.  The hard drive was then formatted, and Windows XP Professional (SP1) was installed. After the Windows installation was complete, we hit the Windows Update site and downloaded all of the available updates with the exception of the ones related to Windows Messenger. Then we installed all of the necessary drivers, and disabled then removed Windows Messenger from the system.  Auto-Updating and System Restore were also disabled, and we setup a 768MB permanent page file on the same partition and drive as the Windows installation. Lastly we set the Visual Effects to "best performance", installed all of the benchmarking software, defragged the hard drive and ran all of the tests at the CPU's default and overclocked speeds.  Now, for our results...

OVERCLOCKING EXPERIENCE:

Despite some of the BIOS's shortcomings I was able to hit a very respectable FSB setting.  I achieved the highest memory setting with the FSB set to 192MHz giving us a speed of 2208MHz (11.5x192MHz) and the CPU multiplier to 11.5X.  This gave me 384MHz DDR and some very good numbers.  This was also pulled off by setting the Vcore to 1.80V and the DIMM voltage to +0.050V.  I was able to complete all benchmarks at this setting and the board remained extremely stable.  I will note that I was able to boot into Windows at 200MHz FSB with the 11x multiplier, but we were unable to run any benchmarks reliably.  Keep in mind that this was all done with simple air cooling and the PC2700 sticks that we used in our nForce2 reviews recently, just to have a level playing field.

There are a couple other findings from overclocking this board I feel are noteworthy.  First, as I mentioned above I was able to boot into Windows with the FSB set to 200MHz and the CPU multiplier set to 11x.  I was only able to achieve this with the memory modules in DIMM's 1 and 3.  The system would not boot into Windows any higher than 196MHz FSB with the memory modules in DIMM's 1 and 2.  I also found that setting the Vcore to 1.80V was the most stable setting.  I'm sure some of you are asking why we didn't give the CPU more juice to get a better overclock.  Anytime I set the Vcore higher than 1.80V, the system seemed unstable and I could not run any benchmarking software successfully. 

Just to see if the two sticks of PC2700 memory were holding back the system, we popped a stick of 256MB Geil PC3500 memory on the Tyan S2495 board.  We were able to boot into Windows at 2200 MHz (11x200MHz) and run a couple benchmarks.  We also did a set of benchmarks in Unreal Tournament 2003 and achieved some higher scores than what are posted later in this review.  The board was perfectly stable during these benchmarks with the Geil stick at DDR400.  Even though I was able to hit DDR400 on the nForce2 boards, using the same PC2700 memory sticks that I am using here, we all know one of the shortcomings of the nForce2, is DDR400 mode.  The benchmarks show that it is actually slower than when the nForce2 is running at DDR333.  This is due to the chipset severely relaxing the timings when at 400MHz, for stability.  It's no wonder they could indeed boot at 400MHz but with negative effects in performance.  However, we're talking Dual Channel DDR333 on the nForce2 versus Single Channel DDR 333 or 400 on KT400s.  So, it's hard to compare apples to apples in either case.

Finally, I would like to note that despite being able to hit a 192MHz FSB on the Tyan KT400 board, we are going to run all our overclocked benchmarking tests at 12.5x182MHz (2275MHz).  We are doing this to give us a good comparison to our previous nForce2 motherboard reviews.  For those that didn't notice, the 12.5x182MHz actually gives a higher clocked CPU (2275Mhz compared to 2208MHz).

The Hot Hardware Test Systems AMD showcase...

Motherboard:

Tyan S2495 Trinity KT400

Bios v1.02 (1/9/2003)

Common Hardware and Software:

AMD 2600+ Athlon Processor 333MHz FSB

2 X 256MB Kingston PC2700 memory

NVIDIA GeForce4 Ti 4200 (Drivers - v.41.09 WHQL)

Seagate 40GB ATA-100 7200RPM Hard Drive

On board sound

WinXP Professional w/ SP1

DirectX 8.1

VIA Hyperion 4in1 Drivers v4.45
 

SiSoft Sandra and Overclocking Tests Simple performance metrics

Sandra Testing: Default Settings DDR333
 

For the default and overclocked settings, the memory was set at 333MHz with 6-3-3-2.5 timings.  These are not the most aggressive timings, but we are only using PC2700 sticks of memory.  Comparing these results to nForce2 boards we can see that the numbers are not as good.  While the CPU and Multimedia scores are nearly the same, the biggest setback is in the memory scores, where the Tyan S2495 board only managed a little above 2100. The nForce2 motherboards came in a little above 2500 with these same settings.  The obvious advantage with the nForce2 boards is being able to run in Dual DDR mode.

Sandra Testing: Overclocked DDR364

Here we notice that the Tyan S2495's score comes in at a little over 2300 which is still less than the Dual DDR nForce2 boards at default settings.  Again, the Dual Channel DDR is the big advantage for the nForce2 board.  The other Sandra benchmarks are about the same for both the KT400 chipset and nForce2 chipset.  The Tyan S2495 is putting out some very competitive numbers.  I would like to remind you that even at this overclock, the board was perfectly stable and I had no problems running any of the benchmarks.  So, for those of you who felt this board may not be overclocker friendly, I think we all stand to be corrected somewhat.

ZD eTesting Labs Business and Content Creation Winstones Desktop Application Performance

The Business Winstone tests include:

• Five Microsoft Office 2000 applications (Access, Excel, FrontPage, PowerPoint, and Word)

• Microsoft Project 98

• Lotus Notes R5

• NicoMak WinZip

• Norton Antivirus

• Netscape Communicator

In the Winstones, we can see the comparison charts.  Business Winstone 2002 shows the advantages of having a much higher memory bandwidth when it comes to these benchmarks.  The nForce2 seems to pull ahead easily when compared to the KT400 Tyan board.  To put this in perspective, we're talking about a 14% gain in performance at default settings.  This is fairly significant, but I would like to note that even this type of gain is not easily noticeable to the average user.  The Tyan KT400 board can still hold its own and performed well in this benchmark.

Content Creation Winstone tests include:

• Adobe Photoshop 7.0
• Adobe Premiere 6.0
• Macromedia Director 8.5
• Macromedia Dreamweaver UltraDev 4
• Microsoft Windows Media Encoder 7.01.00.3055
• Netscape Navigator 6/6.01
• Sonic Foundry Sound Forge 6.0

We have pretty much the same story here when it comes to Content Creation.  The nForce2 motherboards keep about the same margin of performance gains here at default settings.  The numbers to note here are the overclock from the Tyan S2495 board.  The KT400 board still did not manage to reach the default setting numbers of the nForce2 boards.  These benchmarks show the real benefits of memory bandwidth.  That is the obvious reason for the KT400 falling behind.  Let's see if memory bandwidth plays the same role in 3D gaming performance.

Quake 3 and Comanche 4