TGM: Enhanced Functionality
We already described the hardware monitoring system from Tyan in our Tyan Tachyon G9500 PRO review. Since then, its hardware part hasn’t changed. It can still keep track of all voltages, temperatures and can control the fan rotation speed to reduce the noise it produces. The software part did change, however, and the new version of Tyan Graphics Monitor works correctly with third-party cooling solutions. For example, it won’t raise an alarm not hearing the signal from the tachometer when you use a passive heatsink from Zalman.
Overclocking and 2D Image Quality
I overclocked the card with the help of Rage3D Tweaker version 3.8 as the frequency range offered by TGM seemed too narrow. The graphics chip sped up to amazing 525MHz without a screech, while the memory could do nothing better than +10MHz. Otherwise, there were horrible artefacts in the picture. The main reason for this is the slow memory chips, and also the insufficient cooling (a bad contact between the chips and the heatsink). So, I had to stop at 305MHz (610MHz DDR) mark, which can hardly be even called overclocking. So, the 0.13micron process provided the graphics chip with huge overclocking potential, but the memory of this card doesn’t match the GPU. By the way, recalling our overclocking experience with Tyan Tachyon G9500 PRO, I can say that the memory chips on that card also were a bottleneck.
As for 2D quality, it’s all right, just like by Tachyon G9500 PRO. A slight blur of the image appeared in 1600x1200@85Hz only.
Testbed and Methods
So, we are going to pit our RADEON 9600 PRO based card against two other mainstream products – Tyan Tachyon G9500 PRO and ABIT Siluro FX5600 Ultra (350MHz/700MHz). For the sake of comparison, we also include the results shown by the top-end RADEON 9800 PRO based card. Let’s see which architecture is better.
The testbed was configured as follows:
- AMD Athlon XP 2600+ “Thoroughbred” (2.083GHz, 333MHz FSB) CPU;
- EPoX EP-8K3A+ mainboard;
- 512MB Corsair XMS3200 (2-2-5 1T, 333MHz) memory;
- Maxtor DiamondMax Plus D740X (2MB buffer), 40GB HDD;
- Creative SoundBlaster Live! 1024 sound card;
- OS: Microsoft Windows XP SP1;
- Drivers: VIA Hyperion 4-in-1 v.4.47, ATI Catalyst 3.5 (for Tyan cards), NVIDIA Detonator 44.03 (for GeForce FX 5600 Ultra).
The benchmarks I used are listed below:
- Futuremark 3DMark 2001SE Build 330
- Futuremark 3DMark03
- Codecult CodeCreatures Benchmark Pro v1.0
- Unreal Tournament 2003 v2225, Antalus Flyby
- Quake 3: Arena v1.32, Demo four
- Return to Castle Wolfenstein v1.4, Demo checkpoint
- Serious Sam: Second Encounter v1.05, The Grand Catherdral
- Jedi Knight 2: Jedi Outcast v1.04, Massasi Temple – Lightsaber Test
- Splinter Cell v1.2b, 1_1_1Tbilisi Demo
All the games ran with maximum graphics quality settings in three resolutions: 1024x768, 1280x1024 and 1600x1200. There were two work modes: “light” (without full-screen anti-aliasing and anisotropic filtering) and “hard” (4x FSAA and 8x AF).