Power Consumption / Temperature Testing
For this section, every energy saving feature was enabled in the respective BIOSes and the Windows 7 power plan was changed from High Performance to Balanced.
For our idle test, we let the system idle for 15 minutes and measured the peak wattage through our UPM EM100 power meter.
For our CPU load test, we ran Prime95 v26.6 64-bit In-place large FFTs on all available threads for 15 minutes, measuring the peak wattage via the UPM EM100 power meter.
For our overall system load test, we ran Prime95 v26.6 64-bit In-place large FFTs on all available threads for 15 minutes, while simultaneously loading the GPU with OCCT v3.1.0 GPU:OCCT stress test at 1680x1050@60Hz in full screen mode.
At face value, the Core i7-3960X has high power consumption, it’s the second most gluttonous processor on our chart. However, that obviously doesn’t tell the true tale, since unlike Bulldozer , Sandy Bridge-E’s monstrous performance actually justifies its power consumption. The i7-3960X is about 50% faster than the FX-8150 in multi-threaded workloads, yet only consumes 12 watts more. It does not consume much less than a Core i7-980X, but you are getting about 10-25% higher performance across the board, with vastly improved idle results. Compared to mainstream quad-core Sandy Bridge, power consumption is up by 60% but multi-threaded performance gains are above 50%. Overall then, the Core i7-3960X’s performance-per-watt is actually quite excellent.
Having said all that, we can’t help but wonder the extent to which those disabled cores and L3 cache are contributing to the consumption figure. If they are fully power-gated it should be very little, but there is no way for us to tell at this moment.
For the temperature testing, we used the Intel Thermal Solution RTS2011LC liquid cooler. The ambient temperature was 23°C/73.4°F. The application used to monitor temperatures was AIDA64 v1.85.1719 beta Keep in mind that the thermal sensors in most modern processors are not really accurate at measuring idle temperatures, hence the very small delta between the room temp and the idle results.
Idle: The system was left to idle for 15 minutes.
Load: Prime 95 In-place large FFTs was run for 15 minutes.
Since we don’t yet have a compatible LGA2011 bracket for our Thermalright Ultra-120 eXtreme or Prolimatech Mega Shadow heatsinks it’s hard for us to compare the i7-3960X’s heat output to other processors that we have tested. However, based on these results we can tell you that our sample was not a particularly hot-running chip, certainly nothing like the infamous early Core i7D0 LGA1366's.
The Thermal Solution RTS2011LC proved to be an exceptionally quiet-running cooler, with the fan nearly always running at a silent 878RPM. In auto mode, we saw it peak at 958RPM a few times when the system was fully loaded for 15 minutes, but obviously even that was inaudible. We actually had to go in the bios to set 100% fan speed to get the above results. At that level, the fan spins at 2200 RPM, pushes about 73CFM, and outputs roughly 35 dBA. It’s definitely not a terrible product, but it will be interesting to see how much it actually retails for and performs when compared to similar units from Antec, Cooler Master, Corsair, etc. We think a high-end heatsink with a 1600RPM 120mm fan might just outperform it, albeit with greater noise levels.
Power Consumption / Temperature Testing
Power Consumption
For this section, every energy saving feature was enabled in the respective BIOSes and the Windows 7 power plan was changed from High Performance to Balanced.
For our idle test, we let the system idle for 15 minutes and measured the peak wattage through our UPM EM100 power meter.
For our CPU load test, we ran Prime95 v26.6 64-bit In-place large FFTs on all available threads for 15 minutes, measuring the peak wattage via the UPM EM100 power meter.
For our overall system load test, we ran Prime95 v26.6 64-bit In-place large FFTs on all available threads for 15 minutes, while simultaneously loading the GPU with OCCT v3.1.0 GPU:OCCT stress test at 1680x1050@60Hz in full screen mode.
At face value, the Core i7-3960X has high power consumption, it’s the second most gluttonous processor on our chart. However, that obviously doesn’t tell the true tale, since unlike Bulldozer , Sandy Bridge-E’s monstrous performance actually justifies its power consumption. The i7-3960X is about 50% faster than the FX-8150 in multi-threaded workloads, yet only consumes 12 watts more. It does not consume much less than a Core i7-980X, but you are getting about 10-25% higher performance across the board, with vastly improved idle results. Compared to mainstream quad-core Sandy Bridge, power consumption is up by 60% but multi-threaded performance gains are above 50%. Overall then, the Core i7-3960X’s performance-per-watt is actually quite excellent.
Having said all that, we can’t help but wonder the extent to which those disabled cores and L3 cache are contributing to the consumption figure. If they are fully power-gated it should be very little, but there is no way for us to tell at this moment.
Temperature Testing
For the temperature testing, we used the Intel Thermal Solution RTS2011LC liquid cooler. The ambient temperature was 23°C/73.4°F. The application used to monitor temperatures was AIDA64 v1.85.1719 beta Keep in mind that the thermal sensors in most modern processors are not really accurate at measuring idle temperatures, hence the very small delta between the room temp and the idle results.
Idle: The system was left to idle for 15 minutes.
Load: Prime 95 In-place large FFTs was run for 15 minutes.
Since we don’t yet have a compatible LGA2011 bracket for our Thermalright Ultra-120 eXtreme or Prolimatech Mega Shadow heatsinks it’s hard for us to compare the i7-3960X’s heat output to other processors that we have tested. However, based on these results we can tell you that our sample was not a particularly hot-running chip, certainly nothing like the infamous early Core i7D0 LGA1366's.
The Thermal Solution RTS2011LC proved to be an exceptionally quiet-running cooler, with the fan nearly always running at a silent 878RPM. In auto mode, we saw it peak at 958RPM a few times when the system was fully loaded for 15 minutes, but obviously even that was inaudible. We actually had to go in the bios to set 100% fan speed to get the above results. At that level, the fan spins at 2200 RPM, pushes about 73CFM, and outputs roughly 35 dBA. It’s definitely not a terrible product, but it will be interesting to see how much it actually retails for and performs when compared to similar units from Antec, Cooler Master, Corsair, etc. We think a high-end heatsink with a 1600RPM 120mm fan might just outperform it, albeit with greater noise levels.