The first thing that catches your eye is the fact that in the new NV35 the company has finally implemented 256bit memory controller having given up DDR II in favor of DDR standard. With this stone NVIDIA has killed the whole flock of birds. Let’s count them now.
Firstly, the doubling of the memory bus width increased its bandwidth. Although the bandwidth didn’t double, but grew up by “only” 70%, because there are still no DDR chips working at 1000MHz (500MHz DDR).
Secondly, replacing DDR II chips with less expensive DDR SDRAM will reduce the future production cost of the new graphics cards.
Thirdly, the common DDR SDRAM chips dissipate less heat, which will allow giving up huge active and passive cooling solutions. For instance, the graphics cards based on the newest ATI chips do not have any graphics memory cooling at all.
And finally, the changes made to the memory controller resulted into changes of the caching schemes, which definitely got even more fine tuned this way, and texture compression, frame buffer and Z-buffer algorithms closely connected with them got significantly enhanced. By the way, NVIDIA claims that the updated compression algorithms, which have been called Intellisample HCT (High Resolution Compression Technology), provide up to 50% performance increase in “hard” modes using full-screen anti-aliasing and anisotropic filtering.
Other differences between NV30 and NV35 are marked with [notes] symbol in the table above. Let’s discuss them here:
- : New NVIDIA chip inherited flexible NV30 architecture, which pixel processor functional units are not clearly split into separate pixel pipelines (see our NVIDIA GeForce FX 5800 Ultra Review for more details). Like its predecessor, the new NV35 forms either 4 or 8 pixel “pipelines” depending on the situation. This idea hasn’t been very nicely implemented in NVIDIA GeForce FX 5800/5800 Ultra (NV30): “8-pipeline” configuration was enabled only when nothing is written into the frame buffer. In other words, when Z-buffer or stencil buffer are calculated, the graphics processor works twice as fast as in regular frame drawing mode, where NV30 performs as a classical chip with 4 pixel pipelines with 2 TMUs on each.
Everybody expected that the situation will be improved in NV35, especially since NVIDIA has already managed to implement completely the flexible architecture peculiarities in its GeForce FX 5600/5600 Ultra (NV31) working correctly with “4x1” and “2x2” schemes (4 pixel pipelines with 1 texturing unit in each and 2 pixel pipelines with 2 texturing units in each respectively).
However, despite all expectations, NV35 didn’t acquire any advantages over its predecessor: like in NV30, the scheme including 8 pixel “pipelines” is used only for Z-buffer and stencil buffer calculation. It is characteristic that NVIDIA doesn’t regard it as a drawback of their NV35. on the contrary, in all press-releases about NVIDIA GeForce FX 5900 Ultra this feature is described as an architectural advantage, because the chips performs twice as fast during shadows calculation compared with the speed it shows when drawing the frame buffer. Moreover, to stress this difference of the new NV35 from the traditional chips, NVIDIA’s marketing people introduced a new loud name...
- : ...Ultra Shadow Technology. This technology includes one very interesting enhancement acquired by the new NV35: you now can set the edge Z values (depth bounds) for shadows calculation, beyond which no calculations will be carried out. This way you will be able to save time and hence increase the performance in those scenes where all shadows are calculated in real time.
The scheme below depicts the depth bounds as zmin and zmax. Beyond these two values no calculations will take place.
To be able to calculate the shadows faster, gaming engines should certainly be “aware” of this feature, because they will need to calculate and transfer to the accelerator the depth bound values. So far, the existing 3D games and benchmarks with dynamic shadows based on the stencil buffer do not use this feature of the new NV35.
- : From the functional point of view pixel processors of the new NV35 remained the same as by the predecessor, i.e. the company doesn’t offer anything new for its pixel shaders.
However, it is not true for the performance: NVIDIA claims that the floating point calculations are now performed twice as fast. Pixel shaders version 2.0 using floating point data representation used to be a weak spot of NV30, so this improvement has been expected to take place in NV35.
As a result, CineFX architecture “speeded up” in NV35 is now called CineFX 2.0, which was obvious from the very beginning.
- : As we have already mentioned above, the new graphics processor received updated frame buffer, Z-buffer and texture compression algorithms, which is promised to improve its performance in “hard” modes with full-sreen anti-aliasing and anisotropic filtering.
Now please meet the new graphics card from NVIDIA.