X-bit labs

Albatron Gigi GeForce FX5700 Ultra Graphics Card Review (page 3)

Category: Video

by Alexey Stepin

[ 03/05/2004 | 09:55 PM ]

Noise, Overclocking and 2D Image Quality

The new Wise Fan is much louder than the first version, at least you can hear the noise from its fans even against the sound from other system components. However, the noise is well-tuned, and the human ear bears it better than the high-frequency squealing from the Volari Duo V8 Ultra, for example.

The fan redundancy scheme works correctly, launching the third fan when any of the two main fans stops. I had no chance to check out the situation with overheating: the temperature remained normal, never exceeding the threshold of the Wise Fan II thermal diode.

Unfortunately, we revealed some issues with 2D image quality: the image was fuzzy in 1600x1200@75Hz. The fuzziness was even more visible when the refresh rate was set to 85Hz. You can also see shadows cast by letters and graphics on white background. This problem usually occurs when the monitor cable is poorly screened, but this time it is the graphics card itself that is to blame as RADEON 9800 XT reference board produces a crystal sharp image on our monitor. The quality of 2D image was perfect in 1280x1024@75Hz, but if you own a CRT monitor with a 21” or larger diagonal, you may find this card unsuitable.

Our Gigi GeForce FX 5700 Ultra was surprisingly good during overclocking, allowing to increase the GPU frequency to 600MHz (a very satisfactory result!), and the memory frequency to 1025MHz. Having raised the memory clock rate further, we got all kinds of visual artifacts. Still, the extra 125MHz is a nice speed boost for 2.2ns memory with a rated frequency of 900MHz. Of course we used an additional fan, so you shouldn’t repeat this test without taking care of proper cooling. Considering that the backside of the PCB has just a couple of small heatsinks, such overclocking may be a disaster in a closed system case.

Testbed and Methods

Let’s find out how fast the card is and what advantages our overclocking brings to it. The testbed was configured as follows:

• CPU: AMD Athlon 64 3400+  (2.20GHz, 1MB L2);
• Mainboard: ASUS K8V Deluxe;
• RAM: Corsair XMS PC3200 512MB (2x256MB, 2-3-3-6);
• HDD: Seagate 7200.7, Serial ATA-150, 8MB buffer;
• Audio: Creative Sound Blaster Live!;
• Software: Microsoft Windows XP Pro SP1, DirectX 9.0b;
• Drivers:  ATI CATALYST 4.1, 3.9, NVIDIA ForceWare 53.03 and 52.16.

The two versions of ForceWare need a comment of ours. Version 53.03 is not yet approved by FutureMark, probably because of certain “optimizations” implemented by NVIDIA in this driver. 3DMark03 build 340 blocks such optimizations in ForceWare 52.16, but cannot do this with ForceWare 53.03. So we are going to use only the approved drivers with this benchmarking suite.

We used the following games and applications to check out the performance of the card:

First-person 3D shooters:

• RTCW: Enemy Territory;
• StarTrek Elite Force 2;
• Unreal Tournament 2003;
• Halo: Combat Evolved;
• Tron 2.0;
• Highly Anticipated DX9 Game 1;
• Highly Anticipated DX9 Game 2.

Third-person 3D shooters:

• Tom Clancy’s Splinter Cell;
• Tomb Raider: Angel of Darkness;
• Prince of Persia: Sands of Time;
• Max Payne 2: The Fall of Max Payne.

Simulators:

• IL-2 Sturmovik: Forgotten Battles;
• X2: The Threat;
• F1 Challenge 99-2002.

Real-time strategies:

• C&C Generals: Zero Hour.

Semi-synthetic benchmarks:

• Final Fantasy XI Official Benchmark 2;
• Aquamark3.

Synthetic benchmarks:

• Futuremark 3DMark03 build 340.

We removed Deus Ex 2: Invisible War from our list as it doesn’t meet our requirements for producing adequate results. Right now we are considering replacing some other games from our benchmarks list with more advanced applications. We chose the highest settings in each game during the tests to provide the maximum image quality and level of detail.

3D shooters start the testing session – these games account for the biggest share of all 3D games, actually.