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:
Energy-Efficiency
CPU vs. System Power
CPU Specs and Numbers
Test Setup, Benchmark Overview
Idle Power, Prime95
3D Rendering Power
Cinebench 2003
Memory Subsystem
Access Latencies
3DMark '05 and Gaming Power
FarCry, Call of Duty2
F.E.A.R., Prey
DOOM3
Final Thoughts

Give Us Some Feedback on this Review

 Low Power and Energy-Efficient CPUs from Intel and AMD
Core2 Duo E6300 vs. X2-3800+ (ADD) and X2-4600+ (ADO)
(Review by MS, August 20, 2006)

A New Breed, Some New Numbers

If there are new energy efficient processors, then they need to be offset from the standard versions by dedicated codes. The low energy processors are easily enough identified by the third letter in the model number, that is XXA stands for standard power rating, XXO means below 65 W and XXD would be 35 W or less.

In short, the current codes are:

To be true, according to our isolated CPU power measurements , even an X2-5000+ AM2 processor is hard to push beyond 65 W power consumption under full load and in so far, any lower speed grade should almost automatically qualify for the "O" in the AMD code. A "D" on the other hand is somewhat more difficult to achieve with clock speeds beyond 2 GHz, thus leaving the field essentially to the X2-3800+ AM2 or some Semprons.

To the best of our knowledge, Intel is currently not marketing any ultra-low power versions of the Core2 Duo family - not yet at least.

    

AMD "ADD" X2-3800+ and "ADO" X2-4600+. Note the core voltage of 0.976V and 1.088V, respectively!

Intel's E6300 is characterized by the 2 MB (total) shared L2 cache and running at 1867 MHz max core frequency.

Theoretical Considerations

Intel

Based on the raw numbers shown above in the CPUZ screen shots, it is possible to make a few predictions on the overall power consumption of each CPU. To start with Intel's E6300, we can assume that the idle power is going to be very similar to that of the E6700 or the X6800 for the simple reason that the core voltage at idle is going to be approximately the same - likewise, because of the Intel Enhanced Speedstep features, the frequency will also be the same, namely 1600 MHz. At the same time, the physical die size including the total amount of shared L2 cache is also going to be the same for all three processors. Even if 50% of the addresses are disabled in the case of the E6300, this may not result in any power savings since from a power distribution and thermal management standpoint it is more likely that individual micro blocks throughout the array are disabled rather than simply blasting the upper or lower half of the cache's SRAM array. Moreover, the full cache versions may not necessarily use up more power in the cache - for the simple reason that Intel's cache clock gating scheme on a micro-block basis turns off the clock input in areas that are not in current use. Consequently, the only power consumption stems from active areas (with a negligible standby current in the inactive portions).

As a result, even though it is a fact that larger caches draw more power, the cache management implemented in the Core2 Duo may more or less completely mitigate the higher power consumption of a doubling in size. By extension, the same goes the other way around, meaning that for many applications the number of active blocks will be the same and and only the number of unused cache cells will be lower. A little twist to that scheme is that a smaller cache also means that the access frequency of any given address will be twice that of a cache with twice the size, and by extension, since every new content will further require eviction of the old data, a smaller cache might even result in an increased power consumption compared to the larger counterpart.

AMD

In the case of the AMD pair of CPU, the situation is a bit easier. The two processors run at nominally different voltages, and therefore the idle power will be different regardless of the fact that Cool'n Quiet (C'n Q) will throttle both to the same frequency. In more detail, since the power consumption equals:

[Number of Switching Events] * voltage3 * Frequency,

where under idle conditions the [Number of Switching Events] and the core Frequency are constant; the only differentiating factor is the voltage. In this case, the idle power draw of the X2 4600+ (ADO) should be 1.288/0.928 or roughly 1.4 x that of the X2-3800+ (ADD). Under load, meaning that the multiplier is ramped up to the highest supported value, the frequency delta has to be factored in, which then results in a 1.4*1.2 = 1.67 times higher power consumption of the X2-4600+ compared to the X2-3800+. We will show some corresponding data in the following.

Athlon64 X2-3800+
(ADA3800DAA5CD)
Core2 Duo E6300
(HH80557PH0362M)

next page: => Test Configuration =>

All advice and educational articles on LostCircuits are free, but if you feel you can, please make a small donation to us!
Thank you!

General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site. All names and trademarks mentioned in this review are the exclusive property of the respective parent companies.
All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.
Copyright 2002 - 2008 LostCircuits