System Builder Marathon, Sept. 2010: Value Compared

Power And Efficiency

Our power tests included an idle power measurement where the system was prevented from sleeping (active state), a full CPU load test with multi-threaded Prime 95 (one thread per core), a GPU load test with FurMark’s Stability Test at 1920x1200 and 4x AA, and a combination GPU plus CPU load with one CPU thread devoted to FurMark and the remaining threads devoted to multi-threaded Prime95. Whew!

Two GeForce GTX 480 graphics cards draw way too much power. Need we say more?

Efficiency is a comparison of work done to power used. To calculate that, we first begin with a chart that compares the performance of all real-world (non-synthetic) benchmarks. The cheapest PC is used as a baseline.

Again using the cheapest PC as a baseline, we compare the difference in power consumption to the difference in performance.

While overclocking increases the rate at which a PC finishes any given workload, it also increases power consumption. When performance is increased by a greater amount than power consumption, efficiency is increased. Low-voltage overclocking allows the $400 PC to get the biggest efficiency increase, but it never catches up to the efficiency-leading $1000 system.

Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
  • karma831
    oops double post...

    Anyways, I didn't really like the builds this SBM but I learned quite a bit. Thanks for the great read.
    Reply
  • Yeah the $2000 system did not make much sense. Only spending 10% of your build money on the CPU seems wrong. The two GTX480s and Nvidia mobo was bizarre. I guess they felt they had to throw the AMD CPU guys a bone. The lack of a solid state drive in a $2000 build was also odd to me. Which could be explained if your going after raw gaming power, where they did with the dual 480s, but then they gimped it with that AMD cpu. Why pair dual GTX 480s with a Phenom Hexacore; which are subpar for anything that uses 4 threads or less. For the same $2000, I think you would get a much better system with a core i7 950, 6gb of ddr3 1600, a 120gb SSD, and 2 GTX 460 1gb.
    Reply
  • HibyPrime
    I usually skip over the power and efficiency pages of the high-end SBM build, because the power usage is mostly irrelevant for such a high-end build... but when I saw it in the efficiency comparison...

    ONE KILOWATT? seriously!?
    Reply
  • Crashman
    stm1185For the same $2000, I think you would get a much better system with a core i7 950, 6gb of ddr3 1600, a 120gb SSD, and 2 GTX 460 1gb.-1 for the SSD comments since these have always hurt the system's overall score in the benchmark-based value analysis.
    Reply
  • avatar_raq
    One of the odd things encountered in the $2000 build is the results of Dirt2. This game bears the AMD logo, and in one benchmark the intel system scored almost double!! OMG!! AMD guys really need to do something about their CPUs and their relations to game developers.
    Reply
  • SLI does not seems to work in AMD system or it is throttling.
    Reply
  • Crashman
    MayPSLI does not seems to work in AMD system or it is throttling.The CPU is throttling the rest of the system. Most of the benchmarks show a CPU-capped pattern.
    Reply
  • I'd like to have seen CS5 tests rather than CS4, considering you're using a Win 7 x64 build.
    Reply
  • "The CPU is throttling the rest of the system. Most of the benchmarks show a CPU-capped pattern."

    There is noway i5 can be twice as fast as x6. Simply no way. Something is wrong. Unless there is an artificial limitation in the SLI board to prevent it running faster than that. Even 5670 is closer to SLI 480. Simply Dirt2 benchmark is wrong. And I sense SLI is not working. Better to try with Cross Fire setup.
    Reply
  • avatar_raq
    TheCapulet it's just plain unbelievable that the builder didn't do his homework. If you read the article thoroughly you would know the reasons behind the CPU choice! And Thomas was honest about the results and he clearly said 'we failed'. Besides it IS nice to see someone try that and inform us so that we don't repeat the same, or similar, mistakes!


    TheCapulet This will be the first month that people sign up hoping to win the 1k machine instead of the 2k.Free is always good! For me, I wish I win the $2000 build, simply because the 2 gfx cards alone worth almost as the mid-priced build ($920 vs $1000)!
    Reply