Related Articles

Overclocking

The overclocking performance of our X2 4200+ EE fell within our expectations of the lower volted chip and failed to suprise us with any outstanding numbers. The lower tolerances of these chips will in most cases not allow for massive headroom when tweaking the HT speed, but as our results show, there seems to be a decent enough amount of wiggle room to boost performance.

I used our excellent Asus M2N32-SLI Deluxe motherboard - a board I KNOW is solid up to 360MHz HT speed however I could not get the processor to run reliably above 310MHz HT speed with a multiplier of 9.

Don't get me wrong, that is still 55% over spec, and as you can see it allows the processor to sometimes outperform an FX-62 - but it is not as high of an overclock as I'd have expected due to the lower Vcore requirements.

The charts already gave away the maximum stable overclock we reached - 2.79 GHz - but what they don't tell you is what we had to do to get there.

Basically there were four factors to the successful overclock:

1. Figuring out the fastest stable memory speed  (we knew this from past reviews on this board)
2. Figuring out the fastest stable memory timing  (again, we already had this information)
3. Setting the HTT speed and multiplier to what a given processor can support
4. Getting the right Vcore

In order to run at 2.79GHz:

• used the excellent Asus M2N32-SLI Deluxe motherboard
• set Vcore to 1.500V
• set DDR voltage to 2.3V
• set DDR timing to 5-5-5-15-2T @ 1140MHz
• set HT multiplier at 4
• set the HT speed at 310MHz
• set the NB to SB multiplier at 4
• set the NB to SB HT speed at 310
• used a Noctua heatsink with two 63.5cfm Nikao 12cm fans

Energy Consumption

In order to test the power consumption advantages of the EE version of the 4200+ against the non-EE version, we chose to measure the actual watts drawn by the system (not including monitor) both when idling at the XP desktop, and when heavily loaded running the multi-threaded version of the SSE2 version of the POV-Ray rendering benchmark.  We left the Vcore setting on "AUTO", and used CPUZ to report it to us.

As you can see, there is an 11W difference in power consumption between the EE and non-EE parts running at the same speed, with default Vcore voltages; and the difference increases to 18W under heavy load.

But what does that mean in real world terms?

Well, if you are running one computer at home, idling most of the time, not much.

Let's say it was on 10 hours in a day, and was only loaded for one of those hours.

In one day, the difference in consumption would be (11W * 9) + (18W * 1) which is 117Wh. In one year, that would be 42.7kWhr; and if your utility charged you say 40c per kWh, you'd be saving $17.08 in electricity in a year. Not a huge savings, but any savings helps.

Now before you say "thats too high!" apparently "Tier 5" users now pay $0.482 per kWhr in California.

Now let's take a look at an ISP who might use 500 of these chips, running 24/7, heavily loaded:

24h/day * 18W * 365days/year = 157.68kWh per computer.

157.68kWh * 500 systems = 78,840kWh

Say you are in California paying $0.50 per kWh

That's a direct savings of $39,420 per year.

But wait! You are also generating a LOT less heat... and your air conditioning bill is likely to go down by a similar (but higher due to inefficiencies of air conditioning) amount! Basically, a large server farm could save around $80k/year by using EE chips.

Conclusion

We'd like to thank NCIX.com for helping us get a E6600 for this review. 

This chip definitely presents some interesting numbers for those intrigued by the low power concept. Here's a quick rundown of everything we noticed with this chip.

• We've tested the chip at stock speeds - and found that it performs pretty much how we would expect it to, compared to its slower (AM2 X2 3800+) and faster (AM2 X2 5000+) brothers.
• After finding its limits, we even got it to perform in the same ballpark as a stock FX-62 - for much less money. Granted, it did not overclock as high as we might have hoped, but 2.79GHz is nothing to sneeze at.
• It really does save some power - up to 18W under full load, and 11W when idling on the windows desktop.

Is the savings worth the price premium? For the average home user, the answer is 'NO'. It would take 2-3 years of "normal" use before the price differential was recouped.

For a server farm, the answer is "YES". The price differential could be saved back in little time, and the savings over longer terms can be quite worthwhile. This chip does foreshadow roughly the power consumption we should see when AMD moves to 65nm - as all their X2 5000 and lower rating 65nm processors are rated at 65W TDP - the same as this X2 4200EE. This should definitely make things more interesting as far as overclocking ability goes, and we wouldn't be suprised to again see an EE range of processors pop up sometime after the launch of the 65nm Athlon 64's. In the meantime, the chips that are available now do what they are intended to do very well and will be a good choice for those who deem the heat and energy savings an important buying decision factor.