Graphical Benchmarks
With the graphics card in the equation, look for the P4 reference system to close the gap since we'll be card limited in some situations. But in general, the Platina should continue to perform strongly as shown already.
Only 5% difference between the pair of systems here. What's interesting is that the FX-51 and 5900 Ultra combination is also as fast 5950 Ultra and the 3.0C, as benchmarked recently for my article on the new NV38 powered reference board. Admirably high performance from both systems, the PAT BIOS for the EPoX Springdale motherboard is responsible for keeping the gap so small.
While Comanche 4 is definitely a graphical benchmark, making use of DirectX 8.1 and multiple vertex and pixel shader effects, it's more system bound than card board. Given that, the Platina should show the P4 system a clean pair of heels and indeed it does. That's the fastest Comanche 4 run I've personally witnessed, by far. Only Tarinder in some recent testing with an Athlon 64 @ 2.2GHz/DDR440 has seen higher at HEXUS.
Our custom Suntemple low detail test, used for system testing, is also focussed on subsystem performance, over the performance of any installed graphics card. It's the SETI difference all over again, nearly 17% faster than the Pentium 4 system.
Without repeating myself, it's 15% this time, for the same reasons we've seen before. The FX-51 offers mighty performance compared to the P4 setup. When the card is quick enough to allow full CPU power to take effect, the FX-51 in the Platina is on average anywhere from 10% to 20% faster than the 3.0C and optimised Springdale-PE setup. The last benchmark now.
We're card limited with X2, a nice deviation from the norm, showing that other system components have a valid effect on performance, as much as the CPU and memory subsystems.
