Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

LOSTCIRCUITS	SHORTCUTS: Top Page Die Shrinks and Leakage Currents Test Setup Amorphium3 Cinebench 2003 I Cinebench 2003 II MDK2 3DMark2001SE Comanche4 Unreal Tournament 2003 X2 Sysmark 2004 Sysmark2004 continued Conclusions Prescott 3.0 Pricing Give Us Some Feedback to Improve our Reviews
Intel Pentium4 "Prescott" To Scale or not to Scale, that is our question.
(Review by MS, May. 23, 2004)

The Cinebench 2003 RayTracing benchmark is heavily optimized for HyperThreading and, therefore, it is not surprising that the Pentium4 has a definite edge over any of the AMD competition. What is interesting about most Ray Tracing methods is that only a relatively small amount of data is loaded into memory (< 1 scanline at the time) and then rendered which implies heavy traffic of data from memory to the CPU and back. This would favor low latency turnaround times for data, which, on the CPU level means short pipelines. A large cache helps but does not appear to weigh in as much as short access latencies.

Cinebench 2003 raytracing score, higher is better.

The graph references a variety of different processors and speedgrades, but we are also interested in a side-by-side comparision of the P4 family members only for clarity:

Prescott (navy blue), Northwood (periwinkle) and ExtremeEdition (dark blue). As it turns out, the Northwood is not only faster but actually also scales better than the Prescott in this particular benchmark. The important factor for performance appears to be raw clock speed as well as data access by the instruction units, that is pipeline length and thread level parallelism, that is, HyperThreading. Since the data are fed to the CPU on a line by line basis and are non-recurrent, the different cache levels are hardly utilized and, therefore, it does not matter much whether it is a Northwood or a Galatin-based ExtremeEdition.

next page:    => Cinebench 2003 Continued =>