![[H]ardOCP](/images/blank.gif){kind=small}
- Date:
- Tuesday , April 27, 2004
- Author:
- Brent Justice
- Editor:
- Kyle Bennett
- Google +1
GeForceFX 5700U GDDR3
GDDR3 graphics memory is ready for the masses, and NVIDIA shows us that it can easily be adapted to even their current video cards. Are there any benefits to it? Many.
FarCry
(DirectX 9)
We are using the full version game of FarCry with patch 1.1 applied. We used the default game detected quality settings for the GeForceFX 5700Ultra. Above you can see what the in-game settings used. Note that FarCry is running in Pixel Shader and Vertex Shader 1.1 mode only. We performed a manual run using FRAPS to record the framerate.
In FarCry the GeForceFX 5700Ultra with GDDR3 did score 1FPS lower than the GeForceFX 5700Ultra with GDDR2 in the minimum FPS category. At only 1FPS difference this can easily fall within the margin of error. You can see that the maximum FPS was exactly the same and the average FPS is also the same, with only a .2 FPS variance. Therefore we can conclude that clock for clock there will not be any noticeable differences in FarCry with GDDR3 memory versus GDDR2 memory. With the capability to overclock the GDDR3 memory to 1100MHz on the GeForceFX 5700Ultra we can see how that affects performance. We were able to gain 2 FPS in the minimum FPS category and 1.4 FPS in the average FPS category. The differences are so small that in actual gameplay one would not notice the difference of 2 FPS while playing.
Flight Simulator 2004
(DirectX 9)
We are using the full version game of Flight Sim 2004 from Microsoft. Under the hardware tab we have the resolution set for each card with Trilinear filtering and textures to max. Under the weather, aircraft, and scenery settings we chose the “Ultra High” level.
We performed a manual run using FRAPS to record the framerate.
The method is simple; we hopped in a Bombardier Learjet 45 taking off from Seattle-Tacoma International. We chose this airport because it has ground textures, the rendering of water, and mountains on the horizon, which are the three main types of landscapes in the game. The weather was set to Fair Weather and we disabled the Weather rate of change so that it was static. The selected date and time was set for summer during the Day at 1:00PM. Flight Sim 2004 has a feature where you can save this scenario so that we could start it off with the exact same settings each time. Once the game was loaded we set the Auto Pilot to take us on a straight line out from the airport up to an altitude of 10,000 feet at maximum power.
We pressed the “S” key until we had an outside view of the plane and panned around to the rear of the plane. We then enabled the on screen data with “Shift Z” so that we could see the altitude indicator. We then pressed “F4” on the keyboard at the same time as we started the FRAPS counter. With the plane on autopilot all we had to do was bring the gear up with “G” and watch it rise up to 10,000 feet while the camera had an outside view of the rear of the plane. When the indicator hit 10,000 feet we stopped the FRAPS recording. Doing this with a little practice gives an accurate set of repeatable results we can use for performance comparisons.
In Flight Sim 2004 the GeForceFX 5700Ultra with GDDR3 did score 1FPS higher in the minimum FPS category and 1.3 FPS higher in the average FPS category. With such a small difference gameplay performance is not distinguishable in-game between the two. Overclocking the memory only to 1100MHz on the GeForceFX 5700Ultra with GDDR3 did give us a noticeable rise in performance. As you can see we only gained 1 FPS in the minimum category but the average framerates rose by 5 FPS. Looking at the graph we can clearly see that the yellow line has a consistent rise not dipping down in performance where the card did at default clock speeds. Therefore we can conclude in Flight Sim 2004 that performance between a 5700Ultra with GDDR2 and GDDR3 is the same, and by overclocking the memory as high as we can we do see a noticeable rise in performance.