Excellent article! It would be nice to see similar article for Intel processors because even if they are less power hungry, the Intel chipsets are not the "greenest". I am just considering a 24/7 home server and this information is very useful for me. Thank you!
Good article! You have the wrong processor name listed for the 2.10 GHz G1 Brisbane as "Athlon 64 X2 4800+ EE". It should be 4000+. I was a bit confused when I read the performance charts and noticed two 4800+ listed until I realized the mistake.
besides that intel still has the memory controller as a separate chip on the motherboard, where as amd has that included on the chip.. therefor a higher chip power use might be offset by the absence of the external memory controller, which would become visible when idling ..
The extra power consumption on the Phenom is due to the fact that the NB/IMC voltage stays at 1.250v even when the rest of the processor is running in standby. Kinda of annoying that they put it that high, since with a bios that still has the p-states section you can easily under volt the IMC without losing stability, especially at stock speeds. That will cut down on the idle and load power usages drastically.
This article, which btw is realy usefull for me atm - thanx, brought me an idea. What about to compare all pc hardware (e.g. motherboards, harddrives etc)in conjuction to power consumption. It would be realy advantage (at least for me coz i speculate what hardware use to build up server which will run 24/7/365 considering energy cost to be as less as possible.
For the strange power consumption figures for the Phenom, try using the downcore option (it's on the Asus M3A32-MVP) and limited it to 2 cores, and disable the L3 if possible, and retest to see if the figures are comparable to the X2s.
IMO the dynamic power used by the processor is a factor of frequency, voltage and probably the number of transistors. The Phenom has 2 more cores, and a much larger L3 cache. The Phenom core itself is fairly similar to the K8s, and should not exhibit that much power increase if you can match the setup of a typical X2...
I would be interested to see a chart that showed how many watts it takes to do a multi threaded task. For example, a faster core(s) finishes the job quicker then can drop to its low power state. And also on how much work can be done in a distributed computing task. This would be similar to "it takes X-watts to complete a task.”
I´m happy with my X2 4000 Brisbane. With the progam CrystalCPUID I can automaticly manage the processor to consume just 5w per core until cpu usage reaches 60%, than changes to 2700Mhz 70w in less than 100ms.
Its grate and more economyc than cool n´quite.
Creating charts like these are very useful to system builders as well. System power consumption is not only limited to to power that system uses alone, with all the power used there is a good amount of heat generated.
HTPCs is probably hit the hardest by this, and any other system that is placed in so called computer desks that have a closed cabinet for the PC. Hot running system increases heat generation and affects home/server room temperatures which also affects the power bill. Cooler running systems also lead to quieter system cause of the decrease of cooling fans.
One Item I would also like to see tested are power consumption among motherboards. CPUs and GPUs are obvious that the higher the performance the more the power will be used, but how does it compare with motherboards.
[citation][nom]mlmiller1[/nom]Nice article! I would be interested to see a chart that showed how many watts it takes to do a multi threaded task. For example, a faster core(s) finishes the job quicker then can drop to its low power state. And also on how much work can be done in a distributed computing task. This would be similar to "it takes X-watts to complete a task.?[/citation]
YES! Specifically, I'd like to see how much energy is used by each system if they have to do the same task for a given about of time ... like play a dvd for 2 hours.
Dayanidhi is coming this February to India's first and independent annual summit for the game development ecosystem - India Game Developer Summit (gamedevelopersummit dot com) to share the experience he gained while meeting the challenges in building a team (from scratch to the current size of 150), setting up processes, building and maintaining automation tools to achieve best possible productivity giving benefit of optimised time-to-market to the organisation operating in the worldwide market.