System Builder Marathon, Q1 2014: The $2400 People’s Choice PC

Deals
By Thomas Soderstrom
published

Power, Heat, And Efficiency

We already knew that the previous build’s six-core processor and Hawaii-based graphics cards were energy hogs. But at least that machine idled down to reasonably low power consumption. My latest effort is even more miserly, though not to the degree I was expecting.

The chart below shows global power consumption, which includes losses inside the power supply. At 85% efficiency, the new system’s 802 W peak input power turns into 682 W of output for its Corsair HX750 power supply.

Noticing that the new machine draws 116 W less under full CPU load, our temperature chart might make a few readers do a double-take.

Both setups use the same CPU cooler, yet the current system's four-core CPU runs far hotter than the previous build’s six-core processor. The Haswell architecture's issues with heat are a glaring flaw in an otherwise excellent product.

Tom's Hardware readers who want a multi-purpose machine with an emphasis on gaming will be happy with the change to GeForce GTX 780 cards and a Core i7-4770K. Is the new build's advantage in games strong enough to counter the previous machine’s productivity lead, at least giving us a tie at stock clock rates?

Truth be told, one reason last quarter's box sunk to the new machine's performance levels at stock settings was that its workstation-oriented SATA driver sacrificed performance in the name of data integrity. After reverting back to Windows 8’s stock AHCI driver for its overclocked configuration, storage scores jumped by 15%. Thus, the new machine is not as strong as my combined scores would have us believe.

Thomas Soderstrom
Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
41 Comments Comment from the forums
  • Darkerson
    Interesting move, showing the nicest build 1st instead of last. Cant wait to see all the builds compared and see what you all come up with as the budget goes down.
    Reply
  • captain_jonno
    Looks good. Surprised only went with a 750w PSU though. Considering 2x 780 ti's and overlocking
    Reply
  • Crashman
    12951919 said:
    Looks good. Surprised only went with a 750w PSU though. Considering 2x 780 ti's and overlocking
    Yessir, two 780s and a bit of experience in part picking lead me to expect around 700W of required system power. And, it came out just a little less than 700W.

    Power supplies of greater capacity and similar reliability at this price tend to be lower-efficiency units. And we like efficiency too.

    Reply
  • YellowBee
    I dont get the "W" usage?680+237 = 917w. Not 802w as meation above?
    Reply
  • Crashman
    12952008 said:
    I dont get the "W" usage?680+237 = 917w. Not 802w as meation above?
    It's not calculated power, it's measured power for the entire system (at the power plug). No addition or subtraction was used.

    1.) Start the system, wait for all processes to load, take a measurement (Active, but idle)
    2.) Load the CPU using eight thread of AVX-optimized Prime95, take a reading (CPU Load).
    3.) Load GPUs with 3DMark 11 Test 1 in loop, take max reading as it heats up (GPU Load).
    4.) Load both applications (CPU+GPU Load).

    The "math problem" is that any program used to fully load the GPU also partly loads the CPU. So when test 4 is Prime95+3DMark, Prime95 can only use whatever CPU resources are left with 3DMark running.

    So the most accurate system power reading is with "CPU+GPU Load" applied. The system measurement for "CPU Load" still includes the power of an idle GPU. And the system power measurement for "GPU Load" still includes the amount of CPU energy it takes to run the GPU's test application.
    Reply
  • YellowBee
    12952046 said:
    12952008 said:
    I dont get the "W" usage?680+237 = 917w. Not 802w as meation above?
    It's not a calculation, it's a reading for the entire system (at the power plug). Load the CPU using eight thread of AVX-optimized Prime95, take reading one. Load GPUs with 3DMark 11 Test 1 in loop, take max reading as it heats up.

    The "math problem" is that any program used to fully load the GPU also partly loads the CPU. So when test 3 is Prime95+3DMark, Prime95 can only use whatever CPU resources are left with 3DMark running.

    So the most accurate system power reading is with "CPU+GPU Load" applied. The system measurement for "CPU Load" still includes the reading of an idle GPU. And the system power measurement for "GPU Load" still includes the amount of CPU power it takes to run the GPU.

    Very much appreciated and satisfying answer.
    Thanks Crashman :)
    Reply
  • bemused_fred
    12952008 said:
    I dont get the "W" usage?680+237 = 917w. Not 802w as meation above?
    It's not calculated power, it's measured power for the entire system (at the power plug). No addition or subtraction was used.1.) Start the system, wait for all processes to load, take a measurement (Active, but idle)2.) Load the CPU using eight thread of AVX-optimized Prime95, take a reading (CPU Load).3.) Load GPUs with 3DMark 11 Test 1 in loop, take max reading as it heats up (GPU Load).4.) Load both applications (CPU+GPU Load).The "math problem" is that any program used to fully load the GPU also partly loads the CPU. So when test 4 is Prime95+3DMark, Prime95 can only use whatever CPU resources are left with 3DMark running.So the most accurate system power reading is with "CPU+GPU Load" applied. The system measurement for "CPU Load" still includes the power of an idle GPU. And the system power measurement for "GPU Load" still includes the amount of CPU energy it takes to run the GPU's test application.
    Any chance of including these calculations in all future articles, so that we know exactly how the power graph is calculated? Ta.
    Reply
  • Crashman
    12952271 said:
    12952046 said:
    It's not calculated power, it's measured power for the entire system
    Any chance of including these calculations in all future articles, so that we know exactly how the power graph is calculated? Ta.
    Which calculations?

    Reply
  • jabuscus
    wow. such performance. many ram. they should've put in 16gb of ram for real high-end specs. ;)
    Reply
  • Versutia
    As I'm into quiet enclosures, I'd go along this route:

    http://pcpartpicker.com/p/3fuGw

    Wondering how much of a difference would non-reference cards make. Obviously, CPU cooler and RAM could be different, BR drive optional, storage drive as well.
    Reply