The much maligned and misunderstood 3DMark 2003, is fortunately for us here, mostly without controversy. Since we're using only one graphics card in this test, the Radeon 9700 Pro, there should be less muck to cloud the picture. The only concern you could remotely point to here, is whether or not the test's Athlon or Pentium 4 compilations favor one host CPU over another. However, this can be argued for nearly every test you could throw at a processor, either real application or synthetic. As the old saying goings, let's "get over it" and just run the numbers.
| || |
| 3DMark 2003 Standard and CPU Test |
| No controversy here - One graphics card fits all |
The 3.2GHz P4 only takes the lead by 5% versus the Athlon XP 3200+ here. The graphics pipeline is the limiting factor and as you can see, at 1024X768, the performance gain from 3GHz to 3.2GHz in the Pentium 4 is negligible. However, the CPU test portion of a standard 3DMark 03 run, will drop the quality settings down, taking the burden off the graphics subsystem.
Things scale more in line with what is expected here with nearly an 8% advantage for the 3.2G P4. Additionally, the 3.06GHz P4's 60MHz clock speed advantage actually gives it a small edge over the 3GHz P4 with 800MHz FSB.
| || |
| Novalogic's Comanche 4 |
| Frame rates on the battle field |
Flight Sim fans in our midst will attest to the fact that one of the best things you can do to improve performance in rendering all that terrain data, is to drop in a faster processor. Novalogic's Comanche 4 is no exception and the test is very CPU and bandwidth intensive. The 3.2GHz P4 has a 21% lead over the Athlon XP 3200+ and approximately a 7% lead over its 3GHz siblings. This again all scales appropriately, since a 200MHz speed advantage versus a 3GHz P4, translates to roughly a 6.6% clock speed boost.
Quake 3 Four Demo, SPEC ViewPerf And The Wrap-up