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: A new Platform Quad-FX Idiosyncrasies The Motherboard Test Config, Benchmark Overview Memory Performance The 64-bit Question Idle Power by OS Max Power Render Power (Cinebench) trueSpace POV-Ray Media Encoding 3DMark'06 F.E.A.R., DOOM3 FarCry 1.4 COD2, Prey Final Words Give Us Some Feedback on this Review
AMD's Quad FX Platform What's in a 4x4?
(Author: MS, January 21, 2007)

3D Render Power

Let's take a moment and compare the rendering efficiency in Cinebench run as 32-bit executable in WindowsXP-32 and the 64-bit version in Vista-64. For the comparison across the spectrum of processors, we only list the 32-bit results in the chart.

Higher is better: The 64-bit version scored 1562 points, equivalent to 16 seconds render time as opposed to 19.5 seconds in the 32-bit binaries.

We then took power measurements on both the single-threaded and multithreaded benchmarks with and without C'nQ.

Power draw during rendering, lower is better (light columns: WinXP-32, no C'nQ; dark columns: Vista-64)

With respect to energy efficiency of both scenarios, the 64-bit version comes out on top. In the single threaded version, the render time decreased from 62 seconds to 50 seconds - along with a lower power consumption, which leaves nothing to argue, in the multithreaded version, the shortened render time still factors in enough for a clear victory.

What is somewhat confusing at first glance is that even in a nominally high load, as long as it is a single thread, CNQ makes a substantial difference, so we were running AMD's power monitoring tool to see what is ging on across the cores. As it turns out, as the load is toggled across the four cores, the other three cores are throttling back. This type of fast throttling affects both core frequencies and voltages - as far as we can tell, the voltages are switched on a per CPU basis, which makes sense since it is the VRM that suppplies common voltage to both cores but the frequencies are actually changed on a per core basis. The other "confusing" issue here is that the cores show up at any frequency between 1000 and 3000 MHz. Keep in mind that those are "caught in the act of switching" screenshots that show the transitioning of core frequency over several steps from high to low or vice versa (it took a few trials to get this tpe of screenshot).

Power consumption per render pass, lower is better (light columns: WinXP-32; dark columns: Vista-64)

Lesson learned, even with a nominally higher power draw, the reduced processing time in a multithreaded application results in lower power consumption per workload or better efficiency rating. Once again, keep in mind that a poorly tuned VRM may skew the load power consumption and that the idle draw is off by an order of magnitude without C'nQ anyway.

next page: => Render Power (Efficiency): TrueSpace =>