nVIDIA did not "fix" the single texture issue as modern games all take advantage of multi-texturing. The memory bus did get an overhaul however. With the NV35, nVIDIA doubled the memory bus from 128bit to 256bit, and despite having the memory run slower (1 GHz DDR2 on the NV30, 850 MHz DDR for NV35) the GeForceFX 5900 Ultra has more bandwidth (16GB/s vs 27.2 GB/s).

If you'd like to see some image quality comparisons between GeForceFX 5900 Ultra and Radeon 9800 Pro, please take a look at our Albatron GeForceFX 5900PV videocard review.

Overclocking the nBox:

Knowing the overclocking results we witnessed with the Albatron GeForceFX 5900PV we expected similar speeds from the MSI FX5900U-VTD256 Ultra, after all a GeForce is a GeForce is a GeForce.

Starting at 450 MHz core speed I slowly raised the core speed 5 MHz at a time, and easily passed 475 MHz. While overclocking the core alone we actually got up to 510 MHz! However, as with the Albatron GeForceFX 5900PV, when we overclocked the memory it decreased our max core overclock.

Because the MSI FX5900U-VTD256 Ultra uses the same 2.2ns Hynix BGA DRAM as the Albatron GeForceFX 5900PV we were expecting similar results from it on the memory side of things. Amazingly, the memory reached 1 GHz by itself.... however once we included a core overclock it severely slowed the memory down.

I know nVIDIA mean with their thermal management, but it makes finding the maximum speed of any given FX5900 much harder. After about an two hours of frustrating tinkering we found that if we clocked our card at 489 MHz core and 951 MHz memory we'd get optimal performance.