Windows 98 VGA Charts
Introduction
This review compares the performance of sixteen different video cards under Windows 98 SE, using a combination of games and synthetic benchmarks. The cards span seven generations of hardware, from the TNT2 to the GeForce 6 series.
Most of the video cards compared in this review are well-represented in benchmark results published during their respective eras. However, older reviews often have a number of problems:
This review seeks to fill in these gaps by testing seven generations of graphics hardware in the same relatively high-performance Windows 98 system, using the latest versions of games, benchmarks, and drivers.
While there are no strict criteria for the cards tested in this review, they do share several characteristics:
Most of the video cards compared in this review are well-represented in benchmark results published during their respective eras. However, older reviews often have a number of problems:
- Most compare only two or three generations of hardware.
- Many use older versions of games, synthetic benchmarks, and drivers, and so are less relevant today.
- Many use slower CPUs which can easily be upgraded today.
This review seeks to fill in these gaps by testing seven generations of graphics hardware in the same relatively high-performance Windows 98 system, using the latest versions of games, benchmarks, and drivers.
While there are no strict criteria for the cards tested in this review, they do share several characteristics:
- AGP interface
- modest power requirements
- not generally bottlenecked by the CPU used
Test Setup
System
Chipset CPU RAM HDD Sound PSU |
Dell Dimension 4100
Intel 815E Pentium III 1000 256 MB PC-133 20 GB Western Digital Caviar 205AA Sound Blaster Audio PCI 64 (CT5807) 200 W Dell PS-5201-7D |
Test Candidates
Card GeForce 6200 GeForce FX 5200 GeForce4 MX440 GeForce4 MX420 GeForce3 GeForce2 Ultra GeForce2 GTS GeForce 256 DDR Riva TNT2 Ultra Radeon 9550 Radeon 8500 Radeon 7500 Rage Fury MAXX PowerVR Kyro II Matrox G400 MAX Voodoo3 3000 |
Drivers ForceWare 81.98 ForceWare 45.23 ForceWare 45.23 ForceWare 45.23 ForceWare 45.23 ForceWare 45.23 ForceWare 45.23 ForceWare 45.23 ForceWare 30.82 Catalyst 6.2 Catalyst 6.2 Catalyst 6.2 4.12.7942 22.0033 6.83.017 1.07.00 |
Direct3D 9.0c 9.0a 7.0 7.0 8.0 7.0 7.0 7.0 6.0 9.0 8.1 7.0 6.0 7.0 6.0 6.0 |
Memory
(MB) 256 128 64 64 64 64 32 32 32 256 128 64 32 64 32 16 |
Bus
Type DDR2 DDR DDR SDR DDR DDR DDR DDR SDR DDR DDR DDR SDR SDR SDR SDR |
Fill Rate
(MTexel/s) 1400 1000 1100 1000 1600 2000 1600 480 300 1000 2200 1740 500 350 333 333 |
Bandwidth
(GB/s) 13.3 6.4 6.4 2.656 7.36 7.36 5.312 4.8 2.928 6.4 8.8 7.360 4.576 2.8 3.2 2.66 |
Brand EVGA Generic EVGA Generic Visontek Hercules Generic Creative Diamond ATI ATI ATI ATI Hercules Matrox STB |
Software
- Windows 98 SE
- DirectX 9.0c (December 2006)
- 3DMark99 MAX
- 3DMark2000 v1.1
- Quake II 3.20
- Quake III Arena 1.32
- Unreal Gold
- Expendable
Methodology
Each card was tested on a fresh install of Windows 98 SE. The system was restarted between different benchmarks, but not between consecutive runs of the same benchmark. All benchmarks were run at 640x480, 800x600, and 1024x768, at 16-bit and 32-bit color. V-Sync was disabled for all tests.
Shorter benchmarks (Quake II, Quake III Arena timedemos) were run multiple times, with the highest repeatable score being chosen. Longer benchmarks (3DMark99 MAX, 3DMark2000, Expendable) were run only once. Unreal Gold's looping timedemo was run four times for each screen mode, with the first score being discarded and the highest remaining score being chosen.
Shorter benchmarks (Quake II, Quake III Arena timedemos) were run multiple times, with the highest repeatable score being chosen. Longer benchmarks (3DMark99 MAX, 3DMark2000, Expendable) were run only once. Unreal Gold's looping timedemo was run four times for each screen mode, with the first score being discarded and the highest remaining score being chosen.