Comparing the Maximum Power Draw for GRID cards

Hi All,

I noticed an interesting thing about the power stats outputted from the "nvidia-smi -q" command for each physical GPU. If you double the Max Power Limit (as there are two GPUs per board) that equals above the published maximums for the board. For instance the K2 would be 125W x 2 = 250 W and the M60 would be 162W x 2 = 324W. You can see those are above the published maximums below for the K2 and M60 cards.

Published Maximums:
K2: 225 W
M60: 300 W

Perhaps this per-GPU maximum power draw allows the single GPU to enter it’s “boost” mode temporarily. And, that would only occur in the case that the other physical GPU on the board is far enough below its maximum power draw not to exceed the published limit for the entire board. Is this the case?



In the same report you’ll see that there’s an enforced power limit for each GPU of 150W.

Power Management : Supported
Power Draw : 24.45 W
Power Limit : 150.00 W
Default Power Limit : 150.00 W
Enforced Power Limit : 150.00 W
Min Power Limit : 112.50 W
Max Power Limit : 162.00 W

So although the GPU itself is capable of drawing 162W each is restricted to 150W, keeping the board under the 300W published Maximum. These limits are set in the firmware, so do not change with software updates.

Power management in the driver is able to control the load between the Min Power Limit and the Enforced Power Limit. The Max is more to report capability vs what would be drawn.

Perfect. Got it. Thanks Jason!


Hi All,

I have another power related question. In the GRID 2.0 documentation, I see that the passively cooled M60 can be configured down from 300W to 240W. I am wondering if the M6 has a similar option. I ask because I see two power levels noted in the core clock speed specs for the M6. I have noted those below but the two power ratings are 75W and 100W.

722 MHz (TGP: 75 W)
950 MHz (TGP: 100 W)

886 MHz (TGP: 75 W)
1051 MHz (TGP: 100 W)

Thanks as always!


Hi All,

Does anyone know why the M6 has two levels of core clock speeds for both “Base” and “Boost?” I found that “TGP” means “Total Graphics Powerdraw.” I get that you’d draw more power at higher core frequencies. I just want to know now why there are four total levels of clock speed for the M6.



Tesla M6 has been designed for blade servers which have different power requirements depending on OEM.

The different power levels give OEMs the ability to optimize their solutions.

Cool. And how is that level set? I didn’t see anything in the vGPU users guide, etc. on how to change it.

I poked around in nvidia-smi and found the “-pl” switch that does change the power limit but the manual page says it’s for Kepler only. See below.

-pl, --power-limit=POWER_LIMIT
Specifies maximum power limit in watts. Accepts integer and floating point numbers. Only on supported devices from Kepler family. Requires
administrator privileges. Value needs to be between Min and Max Power Limit as reported by nvidia-smi.

Thanks Milan!