Page 1:Better? Faster? Cheaper?
Page 2:Taking It To The Limit, One More Time
Page 3:Test Settings
Page 4:Benchmark Results: PCMark And Sandra
Page 5:Benchmark Results: Non-Gaming Applications
Page 6:Benchmark Results: Crysis
Page 7:Benchmark Results: F1 2010
Page 8:Benchmark Results: Just Cause 2
Page 9:Benchmark Results: S.T.A.L.K.E.R.: Call Of Pripyat
Page 10:Benchmark Results: 3DMark
Page 11:Power Consumption And Battery Life
Page 12:Nvidia Returns
Power Consumption And Battery Life
AVADirect’s build appears to have lower idle power consumption than any of its competitor’s configurations. Though we have serious doubts about the GeForce GTX 485M’s ability to run 17 W lower at idle than the GeForce GTX 470M, improved GPU power-savings techniques could account for at least part of this improvement. Other system optimizations, such as an improved BIOS from Clevo, an improved CPU stepping from Intel, or even a “luck of the draw” CPU with lower gate leakage could account for the remaining idle power differences.
The GeForce GTX 485M uses far more power than either the GTX 470M or HD 6970M at full load, but appears slightly more miserly at low load. Battery Eater Pro is a great application for applying a low load, and could help us to reveal more about the single 485M’s power behavior.
The single 485M’s big lead in idle power use is seen as a noticeable improvement in battery life. The 485M SLI’s smaller power advantage over the 470M SLI is seen as a far smaller improvement in battery life. That is to say, the battery life chart proves the usefulness of the power consumption chart, at least at idle.
Jumping back up to the top, we also see that FurMark pushes the 485M SLI system to 336W at the wall while using only one CPU thread. Previous tests have shown that a 420W draw trips the power supply’s 15A protection circuit, and that continuous loads of 400 W activate heat protection after several minutes. The possibility that some users will attempt to push both CPU and GPU power limits simultaneously explains why AVADirect supplied this unit with two power bricks. The more-miserly (at full load) Radeon HD 6970M CrossFire configuration gets by (barely) with only one.
Because the Chicony PA09-022A 300 W power brick is roughly twice as big and far less efficient than SilverStone’s 450 W SFX unit, we’re going to make the unprecedented recommendation that Clevo should either design or contract a specialized, 350 W power unit for its next workstation-class notebook using the highest-efficiency components it can find. While high heat in a sealed power supply normally decreases efficiency, a custom design could use an extruded-aluminum, finned housing to cool its components. People who pay thousands of dollars for the “best” hardware would likely tolerate an extra $20-50 added to the notebook’s price in exchange for the added portability of a single power unit.
- Better? Faster? Cheaper?
- Taking It To The Limit, One More Time
- Test Settings
- Benchmark Results: PCMark And Sandra
- Benchmark Results: Non-Gaming Applications
- Benchmark Results: Crysis
- Benchmark Results: F1 2010
- Benchmark Results: Just Cause 2
- Benchmark Results: S.T.A.L.K.E.R.: Call Of Pripyat
- Benchmark Results: 3DMark
- Power Consumption And Battery Life
- Nvidia Returns