facebook rss twitter

Review: ABIT Siluro FX 5600 DT 256MB & FX 5600 Ultra DT128MB

by Tarinder Sandhu on 29 October 2003, 00:00

Tags: Abit Siluro FX 5600 DT 256MB, abit

Quick Link: HEXUS.net/qatn

Add to My Vault: x

System setup and notes

  • Intel Pentium 4 3.2GHz 800FSB CPU
  • EPoX 4PDA2+ i865PE Springdale motherboard
  • 2 x 256MB Corsair XMS3500 memory run at 2-6-2-2 @ DDR400
  • Intel reference cooler
  • IBM 41.5GB Hard drive
  • Pioneer 105 DVD/RW
  • 420w Samcheer PSU
  • Samsung 181T 18.1" TFT
  • Hansol 920D 19" flat CRT for 1600x1200x32 tests

Video Cards used

  • ABIT Siluro FX 5600 Ultra DT 128MB (350/700)
  • ABIT Siluro FX 5600 DT 256MB (325/400)
  • Tyan Tachyon Radeon G9600 Pro 128MB (400/600)
  • NVIDIA reference GeForce FX 5600 Ultra Rev. 2 128MB (400/800)
  • ATi Radeon 9800 Pro 128MB (380/680)

Software

  • Windows XP Professional Build 2600.xpclient.010817-1148
  • Intel 5.00.1012 chipset drivers
  • ATi CATALYST 3.7 drivers for the 9600 Pro and 9800 Pro cards
  • NVIDIA Detonator 45.23 for the ABIT FX 5600 / Ultra
  • 3DMark 2001SE v330
  • Quake III v1.30 Demo Four
  • UT2003 Demo build 2206
  • Comanche 4 benchmark
  • X2: The Threat - Rolling Demo
  • Gun Metal 2 Benchmark 1

Notes

The two ABIT cards will be compared to their equivalent ATI accelerator, the Radeon 9600 Pro. And to see where they all fit into the big scheme of things, a Radeon 9800 Pro 128MB card will be used as a performance yardstick. The two FX 5600s were benchmarked using NVIDIA's latest, official driver set, the 45.23s. At the time of writing, ATi's newest set, the Catalyst 3.7, was used to re-benchmark the 9600 PRO and 9800 PRO cards. We've now added X2: The Threat and Gun Metal 2 Benchmark 1 to bring the testing suite a little up to date.

One of the reference FX 5600 Ultra Rev. 2's charms is its ability to become almost silent when pottering around in 2D mode. However, neither ABIT card exhibited this ability, even though both cards had differing core clocks for 2D and 3D work. That feature would have been appreciated by the occasional gamer who tends to spend much of their time in a 2D environment; I could be describing myself here. Both cards exhibited excellent 2D ability via the preferred DVI connection, and both were up to par with the standard D-SUB interface. Having been an ATi user for some time, NVIDIA's 2D and excellent multi-display support is as good as anyone else's in the consumer-level sector. We now wish that more companies would offer a dual DVI option on midrange cards. Sales of TFT screens have rocketed in the last year, and it's not uncommon for users to be running two DVI panels in tandem.

As a preface to the benchmarks that will follow, let's talk about numbers and power, for its these qualities that still separate one card from another. Both the big boys of the graphics world like to push new API-busting features, and we're sure they'll have a larger part to play in '04. At this time, though, brute strength is usually a decent determinant of performance. That's as long as you don't try to run a DX9 demo on, say, a GeForce 4 Ti 4600.

Card GPU Clock Pipelines Fillrate (MP/s) Process Memory interface Memory Speed Bandwidth Features
ABIT FX 5600 256MB 325 4 1300 0.13 128-bit 400 6.4GB/s 0.13, CineFX DX9+
ABIT FX 5600 Ultra 128MB 350 4 1400 0.13 128-bit 700 11.2GB/s 0.13, CineFX DX9+
GeForce FX 5600 Ultra Rev. 2 400 4 1600 0.13 128-bit 800 12.8GB/s 0.13, CineFX DX9+
Radeon 9600 PRO 400 4 1600 0.13 128-bit 600 9.6GB/s .13u, DX9+
Radeon 9800 PRO 380 8 3040 0.15 256-bit 680 21.7GB/s F-buffer, 256-Bit bus

The standard FX 5600 256MB becomes incredibly stunted by having only 6.4GB/s of bandwidth. Sure, it applies all kinds of special bandwidth-saving techniques such as colour and Z-compression, but so do all the other cards listed above. The Radeon 9600 PRO, on the other hand, uses a faster core and decent memory bandwidth. It should be an interesting battle between the midrange cards, as both sets us slightly different methods for antialiasing and anisotropic filtering. The high-end comparison card should wipe the floor with these midrange offerings, especially when high levels of AA and AF are used. Note that benchmarks will be carried out at 1024x768x32 and 1600x1200x32. Both resolutions will also be subject to 4x AA and 8x AF torture.

It's really a question of maximising the potential of the hardware through robust and efficient drivers. ATi's Catalyst 3.7s and NVIDIA's 45.23s carry both of those qualities, although ATI's 16-bit OGL texture bug is a concern. Anyway, let's get down to the benchmarks.