Gigabyte's 7VAXP-A Ultra KT400A Motherboard
Competition For The nForce 2

By, Tom Laverriere
May 27, 2003

Let's start the testing with Sandra and the Winstones both at Default settings and overclocked, but first let's take a look at our setup and testing methodology.

Motherboard:

Gigabyte K7 Triton 7VAXP-A Ultra

Common Hardware and Software:

AMD 2600+ Athlon Processor 333MHz FSB

2 x 256MB Kingston HyperX PC3500 Memory

NVIDIA GeForce4 Ti 4200 (Drivers - v.44.01 WHQL)

Seagate 40GB ATA-100 7200RPM Hard Drive

On-board 6-channel Sound

WinXP Professional w/ SP1

DirectX 9.0a

Via Hyperion 4-in-1 v.4.46 Drivers

TESTING METHODOLOGY:

Before we run any benchmarks, we feel it necessary to make clear exactly what it is we do to get our test systems up and running.  When testing the Gigabyte motherboard, the first thing we did was enter the system BIOS and set the board to "Load Optimized Defaults". We then configured the Memory CAS Latency, and other related memory timings, to be set by the SPD (Serial Presence Detect). The hard drive was then formatted, and Windows XP Professional with SP1 was installed. After the Windows XP Professional installation had completed, we hit the Windows Update site and downloaded all of the available updates. Then we installed all of the necessary drivers, disabled and removed Windows Messenger, disabled Auto-Updating, disabled System Restore and set a 768MB permanent page file. Lastly, we set the Visual Effects to "best performance", installed all of our benchmarking software, defragged the hard drive and ran all of the tests at the CPU's default and overclocked speeds.

THERMALTAKE VOLCANO 11+ XASER EDITION:

Also worth noting about our setup is the cooling solution we used for the CPU.  A big thank you to Newegg for sending us the latest and greatest from Thermaltake: The Volcano 11+ Xaser Edition heatsink and fan.  As you can see from the pictures above, this is one major piece of cooling equipment.  The Volcano 11+ is an all copper heatsink which is polished to a nice shine on the bottom.  Although the finish is nice, I would have to recommend a bit of lapping in order to get the surface perfectly flat, as there are still some noticeable edges on the bottom.  The heat sink comes equipped with a fan shroud to place an 80mm fan on top of the heat sink.  There are three ways to set up this heat sink.  One way is to just let the fan run at full speed which provides the best cooling, but is especially noisy.  A rear bracket with an external knob is provided which allows the user to control the fans speed and is also the option we chose.  The third and final option is to place a temperature sensor under the processor and the fan speed will adjust in reaction to the processors temperature, faster for higher temps and slower for idle temps.  While this is nice, it's not completely trustworthy and we felt like we did not have the control we wanted when it came to noise/performance options.  This heat sink/fan from Thermaltake definitely performs when compared to the stock AMD heat sink/fan combination.  When using the stock cooler, we were seeing temps around 61C at load.  When using the Thermaltake cooling solution, load temps were right around 48C which is a very nice difference.  We'd have to recommend aftermarket cooling solutions when using an Athlon chip and overclocking.  A great place to start is with Thermaltake's Volcano 11+ Xaser Edition.  Thanks again to Newegg for providing us with the equipment.

OVERCLOCKING EXPERIENCE:

Overclocking with this motherboard was simple, although the results weren't as great as I had hoped when starting out.  The highest, stable FSB I could reach was 182MHz x 12.5 ( 2275MHz ).  This isn't too shabby but is only a 200MHz overclock from default settings ( 166MHz x 12.5 = 2075MHz ).  I was able to boot into Windows and run a few benchmarks at 185MHz FSB giving us 2312MHz, but not all benchmarks would complete.  We managed to hit the 182MHz FSB with the CPU vcore set at +7.5%.  We bumped the vcore all the way up to +10% to hit the higher FSB frequencies, but the board became unstable.  While we were able to hit 182MHz FSB, it was at the expense of more aggressive memory timings.  Despite us using Kingston's HyperX PC3500 DIMM's (known around here for great DDR400 support at aggressive timings), timings had to be set at 2-3-3-7 in order to get to 182MHz FSB.  While slower timings may not affect overall performance immensely, it will cause some degree of performance degradation.  Overclocking results will differ from user to user, so it's not always wise to draw too many conclusions from our results, but we do think what you're seeing here should be fairly typical.  Now that we have seen the overclocked and default speeds, let's look at some benchmarks to see what kind of performance these speeds provide.

SiSoft Sandra and Overclocking Tests Simple performance metrics

Sandra Testing: Default Settings DDR333
 

Sandra Testing: DDR400
 

Sandra Testing: Overclocked DDR364

Sandra scores look to be right up there compared to an nForce2 motherboard at default settings.  The Gigabyte motherboard actually manages to outperform some of the nForce2 scores at default settings in every Sandra category shown here.  While the difference is small, maybe VIA's Faststream64 technology is really making the difference as it seems to be outperforming a dual channel DDR motherboard.  When looking at the scores for both DDR400 and DDR364, they are not as high as an nForce2 solution.  Now is a good time to mention some quirks we found with this motherboard.  First and foremost, this board does not support memory timings of 2-2-2-5.  6T is the most aggressive timing option for Active to Precharge timings.  We were able to run the Kingston memory at 2-2-2-5 on the nForce2 boards which will improve benchmark scores slightly compared to 2-2-2-6.  Secondly, this board seemed a bit finicky with PC3500 DIMM's.  When using both the Kingston and Geil PC3500 rated DIMM's, the board was very unstable when setting the memory timings manually to the most aggressive timings ( 2-2-2-6 ) and when using the "By SPD" option.  Only when we set the timings manually to 2-3-3-6 we were able to reach perfect stability with the PC3500 DIMM's.  We were able to try some older PC2700 DIMM's in this board and ran them without a hiccup at both the 2-2-2-6 and the "By SPD" timings.  While this is not a major performance hindrance, it definitely effects performance and begs to ask the question, why can't we run the most aggressive timings with the latest PC3500 modules?  We tend to think that Gigabyte could possibly correct this problem with a future BIOS update, but for now we're not sure if it's a chipset or motherboard level issue.

ZD eTesting Labs Business and Content Creation Winstones Desktop Application Performance

The Business Winstone 2002 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

This graph shows the benefits of having a motherboard equipped with dual channel DDR architecture.  While the Gigabyte motherboard's scores are still respectable, they are no match for a similar board using the nForce2 chipset.  This is obviously not Gigabyte's fault since they are manufacturing their "best" solution with what VIA provided them.  These scores further prove the significant performance increase when using a motherboard that supports dual channel DDR memory.

Content Creation Winstone 2003 tests include:

• Adobe Photoshop 6.0.1
• 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 5.0c (build 184)

We see the same story here as we did with the Business Winstone benchmark, the nForce2 motherboard leads by a sizeable margin.  Default settings of the Business Winstone benchmark showed a 14% performance gap in favor of the nForce2 motherboard, when compared to the Gigabyte KT400A motherboard.  That gap is widened a bit more here with a 22% lead for the nForce2 motherboard.  Perhaps VIA's Faststream64 technology cannot keep up with dual channel DDR solutions?  Or perhaps it's VIA's Southbridge that is the bottleneck for drive performance?  Winstones are fairly easily influenced by Drive Subsystem performance, so that is a distinct possibility.  Let's look at some gaming benchmarks to see if any of these theories hold true.

Quake 3 and Comanche 4