The Leadtek WinFast A350 TDH MyViVo
&
The Abit Siluro FX 5900 OTES
Two GeForce FX 5900s Square-Off...

By - Marco Chiappetta
August 19, 2003

The WinFast A350 TDH MyViVo looks just like the NV30 based A300 it's replacing.  The Twin-Turbo II cooling system, and aluminum enclosure give this card a very distinctive look; it sort of reminds us of a silver boom-box from the mid-80's!  We took the A350 apart to get a closer look at the innards and found an aluminum and copper heatsink, mounted over the GPU and 128MB of RAM.  The center of the heatsink is comprised of a thick copper block, with thin fins running horizontally from top to bottom.  When fully assembled, the fans on the front of the enclosure work in tandem to push / pull air over the fins.  One fan blows air into the enclosure, while the other exhausts the hot air out.  The Twin-Turbo II cooling system seems to do its job very well.  When we overclocked this card, core temperatures never got out of hand (as you'll see a bit later).  However, there is one drawback to this system.  It seems to get clogged up with dust fairly quickly.  Potential buyers should be sure to check the grilles occasionally to ensure they are not being obstructed.  With the card disassembled, we also found a thermal pad on the on rear of the enclosure that lined up with the back of the GPU.  This should help pull a bit of heat away from the core as well.  The external plate houses a single DVI and a single DB15 connector, mounted on each side of the Video-In / Out connector.  The inherent capabilities of the GeForce FX 5900 GPU give this card the ability to run dual monitors independently.  The rest of the PCB follows NVIDIA's reference design, with one exception.  The 4-pin power connector at the rear of the board is mounted flat to the PCB.  On most other 5900s, this connector is mounted vertically.  We prefer Leadtek's approach, as it makes it easier to route cables neatly inside the case.

Screenshots From Ulead DVD MovieFactory 2 Nice and Simple

The "My ViVo" version of the WinFast A350 TDH, as the name implies, has the ability to capture and output video to and from an external source.  An on-board Philips SAA7108AE video encoder / decoder handles all of the ViVo functionality.  The hardware, when used in conjunction with the included software offer the following features...

Ulead's DVD MovieFactory 2 SE is one of the programs Leadtek included with the A350 TDH My ViVo.  This program gives uses the ability to capture video directly from an external source and burn it directly to a DVD or CD, so you won't have to gobble up a ton of hard drive space to archive your personal videos.  DVD MovieFactory also helps in the creation of customer CD / DVD labels and lets you edit video directly from DVD -RW and +RW disks.  It uses a very simple. menu driven interface that we found very easy to use.

Screenshots From Ulead Video Studio 7 SE For Budding Movie Moguls

Leadtek also includes a copy of Ulead Video Studio 7 SE with this card.  Video Studio is a nice compliment to DVD MovieFactory 2.  With this program, users can also capture and edit video, but with Video Studio 7 users can also add transition effects, or voice-overs to their video.  There is library of transitions, filters and customizable titles included with this program that should keep newbies to video editing busy for quite a while.

A Closer Look At Abit's Siluro FX 5900 OTES

The Leadtek WinFast A350 TDH MyViVo
&
The Abit Siluro FX 5900 OTES
Two GeForce FX 5900s Square-Off...

By - Marco Chiappetta
August 19, 2003

HOW WE CONFIGURED THE TEST SYSTEM:

The Leadtek WinFast A350 TDH MyViVo and Abit Siluro FX 5900 OTES were tested on an i875P "Canterwood" based MSI 875P Neo-FIS2R motherboard, powered by a Pentium 4 3.0CGHz processor (800MHz Bus).  The first thing we did when configuring this test system was enter the BIOS and loaded "High Performance Defaults".  Then we set the memory to operate at 200MHz (Dual DDR400), with the CAS Latency and other memory timings set by SPD.  The AGP aperture size was then set to 256MB.  The hard drive was then formatted, and Windows XP Professional with SP1 was installed.  When the Windows installation was complete, we installed the Intel chipset drivers and then hit the Windows Update site to download and install all of the available updates, with the exception of the ones related to Windows Messenger.  Then we installed all of the necessary drivers for the rest of our components and Windows Messenger was disabled and removed from the system.  Then Auto-Updating and System Restore were disabled, the hard drive was de-fragmented and a 768MB permanent page file was created.  Lastly, we set Windows XP's Visual Effects to "best performance", installed all of the benchmarking software and ran the tests at our CPU's default clock speed.  All of the benchmarking was done with ATi's and NVIDIA's drivers configured for maximum visual quality.  ATi's "Quality" Antialiasing and Anisotropic filtering methods were employed throughout our testing, while the Performance slider available on NVIDIA's "Performance and Quality" driver tab was set to "Quality".  For the "4X AA + Aniso" tests listed in our graphs, we enabled 4X AA and 8X Anisotropic filtering in both NVIDIA's and ATi's driver panels.

HotHardware's Test Setup Not Quite Top of the Line!

Common Hardware:
Intel Pentium 4 Processor 3.0GHz / 800MHz System Bus
MSI 875P Neo-FIS2R
512MB (256MB x2) Corsair XMS3200C2
Seagate Barracuda V 7200 RPM SATA 120GB Hard Drive

Common Software:
Windows XP with SP1
DirectX 9.0a
Intel Chipset Software v5.00.1012
Intel Application Accelerator RAID Edition v3.0

Video Cards Tested:
Leadtek WinFast A350 TDH MyVIVO (128MB)
Abit Siluro FX 5900 OTES (128MB)
ATI Radeon 9800 Pro (128MB)

Video Drivers Used:
ATI Catalyst Drivers v3.5 - WHQL Certified
NVIDIA DetonatorFX Drivers v44.67 & v45.23

Performance Comparisons With Gun Metal DirectX 9.0 Gaming

For our first set of tests, we used the relatively new Gun Metal benchmark developed by Yeti Studios.  Like all of the other benchmarks we used in this review, Gun Metal is based on an actual game.  Gun Metal uses Vertex Shader 2.0 and Pixel Shader 1.1 ops in the creation of its game world.  Due to this fact, the Gun Metal benchmark is heavily GPU & system bandwidth limited, and because Yeti's intent was to stress all modern 3D accelerators, 2X Anti-aliasing and Anisotropic filtering are enabled by default and cannot be disabled.

In this test, the NVIDIA powered cards held onto a small lead over the ATi Radeon 9800 Pro when we used the brand new v45.34 Detonator FX drivers.  However, when the v44.67 drivers were used with the 5900 had a much larger margin of victory; in the neighborhood of 20%.  As expected, the performance of the Leadtek and Abit cards are virtually identical when using the same drivers.  Only a fraction of frame per second separate the Siluro and A350 at either resolution.

Next Up: More DirectX Testing
  

The Leadtek WinFast A350 TDH MyViVo
&
The Abit Siluro FX 5900 OTES
Two GeForce FX 5900s Square-Off...

By - Marco Chiappetta
August 19, 2003

We continued testing the Leadtek and Abit FX 5900s with Ubi-Soft's Splinter Cell, using the Oil Rig demo created by the folks at Beyond 3D.  This is another test that is heavily dependant on Pixel Shader performance.  Throughout the custom demo, pixel shaders are used to render realistic looking ocean water surrounding a geometrically complex Oil Rig.  As we've mentioned in the past, in its current form, anti-aliasing is "broken" with Splinter Cell.  Due to this limitation, we do not have any AA scores listed in the graphs below.

Benchmarks With Splinter Cell More DX Shaders

At 1024x768, the A350 and Siluro posted the exact same score of 41.1 FPS.  When we raised the resolution to 1600x1200, the Leadtek card nudged slightly ahead of Abit's Siluro, but the .1 FPS advantage falls well within the margin of error.  Also note that the v45.23 Detonators gave the 5900s about a 3% performance boost in this test.  A 3% increase in performance may not be earth shattering, but we'll take it.  Splinter Cell favors the Radeon 9800 Pro's architecture, as is evident by the 10% lead it held onto at both resolutions.

Head-to-Head Performance With Comanche 4 Semper Fi

For our next test, we used Novalogic's combat helicopter simulator Comanche 4.  Comanche 4 uses DX8 class pixel and vertex shaders to produce some of the realistic visuals used throughout the game.  Unlike Splinter Cell and Gun Metal, this benchmark is heavily influenced by CPU and system memory performance, especially at lower resolutions.  At higher resolutions when AA and Aniso are enabled, however, the video card becomes much more of a limiting factor.

Once again, we saw the Leadtek WinFast A350 and the Abit Siluro FX 5900 OTES performing at essentially the same levels at both resolutions, regardless of what settings were used.  Switching from the v44.67s to the new v45.23s again yielded another small performance increase when anti-aliasing and anisotropic filtering were enabled.  The Radeon 9800 Pro excels in this game, besting the 5900s in all but one configuration (1600x1200 - 6X AA).

UT2003 & Quake 3 Benchmarks 

The Leadtek WinFast A350 TDH MyViVo
&
The Abit Siluro FX 5900 OTES
Two GeForce FX 5900s Square-Off...

By - Marco Chiappetta
August 19, 2003

Next up, we have some more DirectX testing with Epic's Unreal Tournament 2003.  When testing with UT2003, we use a special ini file that is supposed to ensure that all of the cards are being benchmarked using the exact same in-game settings and "High-Quality" graphical options.  Unfortunately, the validity of the GeForce FX's scores have been called into question because NVIDIA's drivers cause texture filtering level changes to something between trilinear and bilinear, when Anisotropic filtering is enabled (click here for more detailed information on this topic).  Please keep that in mind when viewing these results.

Performances Comparisons With UT:2003 Head Shot!

Continuing the trend, the Abit and Leadtek cards performed at virtually the same levels at both resolutions, in all test configurations.  With two cards based on the same GPU, clocked at the same speeds, similar performance is to be expected.  What's interesting to note is the 10% jump in performance at 1600x1200 when anti-aliasing is enabled with the new v45.23 drivers.  NVIDIA has been optimizing for more efficient bandwidth use it seems.  Once again, the 9800 Pro took the top spot in every test until we enabled Anisotropic filtering.  However, until the issue mentioned above is cleared up, these numbers will be questioned.

Benchmarks / Comparison With Quake 3 Arena v1.32 I Cannot Wait For Doom III...Q3 Needs To Be Retired...

Next, we moved on to some OpenGL testing with Quake 3 Arena.  We installed the v1.32 point release and ran some tests using the built-in timedemo, "demo four".  Before running these tests, we set Quake 3 to its "High Quality" graphics option with Tri-Linear filtering enabled, and then we maxed out the texture quality and geometric detail options.

The GeForce FX 5900 doesn't support 6X anti-aliasing in OpenGL applications, which is why we have "N/A"s listed in our graphs for this, and the next set of tests.  At both resolutions, regardless of what drivers were used, or whether or not anti-aliasing or anisotropic filtering were enabled the GeForce FX 5900s performed at identical levels.  Without any anti-aliasing at 1024x768, the Leadtek and Abit cards outperformed the Radeon 9800 Pro, but that changed when we raised the resolution to 1600x1200.  With AA enabled, the FXs held onto a slight lead at both resolutions, but that lead disappeared when AA and Aniso were enabled simultaneously.   

Some Serious Sam, Overclocking & The Conclusion... 

The Leadtek WinFast A350 TDH MyViVo
&
The Abit Siluro FX 5900 OTES
Two GeForce FX 5900s Square-Off...

By - Marco Chiappetta
August 19, 2003

We used Croteam's Serious Sam: The Second Encounter for our last set of benchmarks.  We configured the game to use OpenGL (this is one of the few games that can use either DirectX or OpenGL) and ran a series of tests using the built-in "Little Trouble" demo.   To ensure the playing field was level, we used Beyond3D's "Extreme Quality" script which maxes out the texture and filtering quality with all of the cards being tested.

Head-to-Head / Performance With Serious Sam: TSE Lots of Guns, Action and Explosions!

The Serious Sam results were a mixed bag.  At 1024x768, the GeForce FX cards performed within 1% of each other, outpacing the Radeon 9800 Pro by about 9%.  The scales tipped in favor of the Radeon 9800 Pro at 1600x1200, however.  At the higher resolution,  the GeForce FX cards again performed similarly, but the Radeon 9800 Pro pulled ahead by about 8%.  So, what do all the numbers tell us?  Well, the Radeon 9800 Pro is clearly the victor.  It won 17 of the head-to-head tests versus 6 for the GeForce FX 5900s.  The Radeon 9800 Pro was also able to complete 4 more tests (OpenGL with 6X AA) than the FX.  Choosing between the Leadtek and Abit cards is impossible though.  They both had their share of victories, but none of the performance deltas were significant.  Underneath the fancy coolers, they are essentially the same card.

Overclocking With The Leadtek & Abit GeForce FX 5900s Fast = Good  |  Faster = Better!

We installed CooBits and raised the core and memory clock speeds on both of the cards until we started to see visual anomalies on-screen during our benchmarks.  With its extra large heatsink and exhaust system, we expected the Siluro to really shine in the overclocking department.  It didn't let us down, but the Leadtek A350 wasn't about to lay down and hand a victory to Abit.  Both of the cards we're looking at today had the same default core and memory clock speeds (400MHz / 850MHz).  We were able to push the Siluro FX 5900 OTES up to 486MHz for the core, with the memory at 964MHz.  The A350 went a step further, hitting overclocked core and memory speeds of 491MHz and 970MHz.  The performance increases realized by overclocking these cards are represented in the graph above.

Leadtek WinFast A350 TDH MyViVo

We intended to compare GPU temperatures while these cards were overclocked, but unfortunately we could not.  For some reason, the temperature tab within NVIDIA's drivers was not available with the Abit Siluro FX 5900 OTES.  The screen capture above was taken after approximately 30 minutes of benchmarking with the overclocked Leadtek A350.  As you can see, the GPU core temperature never got even close the 140°C slowdown threshold.

The Leadtek WinFast A350 TDH MyViVo and the Abit FX 5900 OTES performed at identical levels and had the same image quality. So, to pick a winner we have to focus on their prices, features, overclockability and bundles.  Unfortunately for Abit, the Leadtek card sweeps all four categories.  The A350 TDH MyViVo is available on-line for $355 US.  The Abit FX 5900 OTES is currently selling for $375.  Both of these cards have similar features and have custom cooling solutions, but the A350 has the added benefit of ViVo capabilities.  The Leadtek card also has the more complete bundle, that included two full version games, video editing software and it overclocked a bit higher too.  When all was said and done, the Leadtek A350 TDH MyViVo was clearly the victor of this shoot-out.  Abit makes one heck of a video card, but the Siluro FX 5900 OTES needs a price drop and a beefed up bundle to hang with Leadtek's offering.  Both of these cards play second fiddle to the Radeon 9800 Pro, however.  The Radeon 9800 Pro won the majority of benchmarks and is available for about $320.  If I was spending my money today, it'd be on a Radeon 9800 Pro, but the NVIDIA fans out there will be happy with either of the 5900s we looked at today.  Based on its competitive price and feature set, we're giving the Leadtek WinFast A350 TDH MyViVo a HotHardware Heat Meter rating of 8.  The Abit Siluro FX 5900 OTES has innovative cooling but was overshadowed by the A350.  We're giving it a 7.5 on the Heat Meter...

Leadtek WinFast A350 TDH MyViVo

 

Abit Siluro FX 5900 OTES

There are a ton of folks just like you in HotHardware's PC Hardware Forum...Go There Now!