Soyo SYKT400 Dragon Ultra vs. Shuttle AK37GTR

by HH Editor — Tuesday, November 05, 2002, 11:00 PM EDT

Page 4:
Soyo SYKT400 Dragon Ultra vs. Shuttle AK37GTR - Page 4

Soyo SY-KT400 Dragon Ultra vs. Shuttle AK37GTR
Showdown at the KT400 Corral!

By, Robert Maloney
November 6th, 2002

Performance Comparisons with 3DMark 2001SE

Synthetic Gaming

As an avid gamer, I was really interested in seeing the 3D performance, both in synthetic tests, and in real-world scores. For the synthetic tests, we chose MadOnion's 3DMark 2001 SE. It generates a score after rendering scenes and measuring performance using the MaxFX game engine, found in Remedy's popular game Max Payne. We ran two series of tests, once at 800x600 with 32-bit color and again at 1024x768x32, both times with all other program settings left at their defaults.

High scores all around, with the Asus board holding the slightest of leads over the Soyo KT400, at least at stock speeds. The scores really took a nosedive, however, when running the tests again while using DDR400 memory. Overclocking each board resulted in a 350 point increase for the Soyo Dragon, pushing it over the 14,000 mark, and a 520 point increase for the Shuttle board. Strangely, this increase still put in behind the score obtained at stock speeds with the Asus A7V333.

The differences we saw at 1024x768 were much less evident than at 800x600, although the order did not change. The Asus A7V333 maintained a 200 point lead over the Soyo board, which in turn was 500 points ahead of the Shuttle. Overclocking the two boards brought them closer together, with only a 270 point disparity.

Performance Comparisons with Comanche 4 Demo

Some hardcore gaming for you grunts

Another popular DirectX benchmarking program is NovaLogic's Comanche 4. Since this benchmark is more CPU dependant than other games, we can get a good feel for the overall system performance by comparing the benchmarked results.

The Asus A7V333 and the Soyo KT400 were in a tight race, with the Asus taking the lead by a mere 0.07 frames per second. The Shuttle board came in at a full frame behind the other two boards. When we switched over to DDR400 on the KT400-based boards, we also lost close to a full frame. When we raised the FSB, we pumped out almost 3 extra frames per second on the each board.

Performance Comparisons with Quake 3 Arena v1.17

And once again, the fan favorite!

We used the Quake 3 Timedemo with the display settings set to their minimums and the screen resolution at 640x480x16 for the Low Quality graphs and then chose 1024x768 with 32-bit color for the High Quality tests. This helped determine the CPU limitations of a motherboard by minimizing the impact the video card has on the performance of the game. With the display settings calibrated in this manner, the ability of the game to tax the video card is virtually eliminated, allowing the benchmark to focus almost solely on the motherboard's CPU performance.

At the lowest quality settings, the demo completed quickly, and we saw more of the same. While the Asus board still commanded a lead, the Soyo was close behind, but the Shuttle was a distant third. In this test, the Shuttle was a full 10 frames behind the other two boards. Again, the DDR400 scores were much lower than those obtained using DDR333. It is becoming quite apparent that DDR400's time has not yet arrived. Overclocking the Soyo Dragon gave us a great benchmark of 329.9 fps, 17% over the original score.

The scores were close again, but if you look closely, the Asus A7V333 score at 133MHz FSB actually managed to beat the Shuttle board, which had been overclocked to 141MHz. The Soyo board started out only 4 frames per second behind the Asus board, but easily leaped past it, posting up 307.4 fps when overclocked to 145MHz.

"Real World" Performance with the Stones

Simulated Application Performance

Business Winstone is an application-based benchmark, which runs through a series of scripts using popular business programs. It attempts to emulate a business system load, and then give a rating. We left the default setting so that these scripts were done five times and the final score given on the left.

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

Content Creation Winstone 2002 is another application-based benchmark, this time using popular content creation programs that are considered more "bandwidth hungry". It keeps these multiple applications open and switches among them while running scripts.

The Content Creation Winstone 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)

Each graph told us practically the same story. The Asus A7V333, while based on the KT333 chipset, managed to outperform the two KT400 based boards in almost every test. The Business Winstone 2001 benchmark was the only test in which the Soyo KT400 Dragon managed to usurp the title from the Asus board, whilst the Shuttle board managed third place finishes, consistently a step behind the others. When the same stick of RAM was set to 200MHz instead of 166MHz, the performance dropped across the board.

CONCLUSION:

It's hard for us to really shake the feeling that the KT400 just isn't all that it's cracked up to be. DDR400 was not officially supported, and it is very clear why, performance simply drops in every aspect when setting memory speeds to 200MHz. In fact, the Shuttle became quite flaky when using DDR400, and it took many attempts to complete some of these benchmarks. While we did not have a chance to try an AGP 8X capable graphics card in this round of testing, our current analysis is that there are negligible gains using AGP 8X Mode. What is especially bothersome is that two KT400 machines were beaten here on occasion, by a KT333 board using the same components installed in the same exact way. This could be due to a robust BIOS from Asus, while the early entries from Soyo and Shuttle still needed some tweaking. Perhaps later BIOS revisions will correct some of these issues.

Soyo SY-KT400 Dragon Ultra

Arguably one of the finest looking boards we have seen, and it comes with many features. The performance was top-notch although it couldn't beat the KT333 board we used as a reference. Although it doesn't officially support DDR400, it did not raise any stability issues when clocking the memory at 200MHz, but the performance level was quite poor. We did have good luck with overclocking, however, and managed to get an extra 10% in the FSB. A major detraction, however, has to be the price point. The cheapest we found the Dragon Ultra set for on Pricewatch, was $154.00, a little pricey in our books. The board itself without all the extras was listed in the $120-130 range.

Great looking board
6-channel on-board audio
Active heatsink on NB
Include Sigma breakout box

High price tag
DDR400 = poor performance
Was a real screamer
Long boot/reboot times

Shuttle AK37GTR

Another good looking board that would look great when paired with a Hercules' 3D Prophet 9700 and Fortissimo III, all based on blue PCBs. While the performance wasn't as great as the Soyo Dragon, it also doesn't cost as much. Then again, there weren't many items in the bundle either. We had a hard time finding it out there in the retail channels, but generally we would expect a sub-$100 price point. Again, DDR400 was not officially supported, but we suffered through such instability, we started to question whether or not we would complete the benchmarks. All in all, a good performer at a good price.