Unusually low Power Consumption after SSD/Wifi Upgrade?

So, I know this is a bit of an out of the basket thread... But i litterally am stumped as can be with this one.

I have a Clevo w110er notebook, upgraded to the hilt, which usually maxed out my 120w PSU during Rendering and games. I'm talking 113-120w.

After I installed an Intel 7260 Wifi card, to replace my Atheros what ever model, and my 1TB 7200RPM HGST HDD, with a Samsung EVO 850 500GB SSD, I've found that my power consumption has gone down to a staggering 89w as an absolute MAX under any stress test I can find, Games, Rendering etc. Even Apps that were notorious for casusing a power shutoff due to over consumption (Overclocking) are now only at the mid 80w's...

Anyone have any idea what the hell is going on?

Specs:
w110er
11" 1366x768
Intel Core i7 3920XM @ 3.9Ghz
GT 650m @ +70/+150
Prema UV Bios, going to try UOV now i have this headroom.
500GB Samsung EVO 850
120w PSU (Generic)
Intel 7260 Wifi.

 

It probably is the SSD, although the wireless card can help a bit. What happens with a slow HDD during data transmission and HDD access is the CPU and GPU, while still at high power draw, are just sitting there waiting for data. That bottleneck costs a lot more than most realize. Since even the GPU is part of the PCI bus, freezing the interrupts on the bus can cause issues there as well.

 

Optimized drivers (Intel card) will do that...

The SSD won't make much difference at idle (~10 minutes more battery life vs. ~5Hr total with HDD), but when the storage subsystem is going full bore, an SSD is probably using more power than the HDD it replaced (but doing more work too, of course).

 

That... depends. Usually on ignoring how ssds spend a lot more time on complete idle than a hdd. But yes, agree it's probably the driver. Given that you had a common atheros nic, and it's one of the travelstar drives. Then we're talking something like... adding some extra margins here.. 6-10w for the wifi depending on what sort of encryption is going on, how much correction has to run, which band, things like that. And the hgst drive runs at peak 5w. And you switched it out with a similar intel nic, etc.

So then I guess at least 16w magically disappeared. ..No, I'd look for a lack of "alive" wmi-type "authorized" interrupts for the wifi driver as well.

And I'm guessing what happened is that you were running on the 5ghz band, or some not fully hardware supported mode and encryption combination, so part of the decoding happens on the processor, instead of on the module. So more than that the card is using more power, the processor is just allowed to idle normally again.

If that's not it, a reinstall of the intel driver package maybe allowed the motherboard to use more power-management functions? Really, no one knows when windows is involved. Worst part - you can't really test for it either, because you can't tie the driver behavior to which specific power state the processor is allowed to return to. You can make a qualified guess, and you can say that 32-bit wrappers are always more likely to queue up the type of operations over time that prevents the processor from reaching the lower power states. But you can't actually guarantee it (it can be a combination of this driver and a specific other program, for example).

If you used the same kind of science for hammering down a nail, you would of course eventually get the nail hammered into the wood, as you wanted. After you've crushed your fingers, your hand, your arm, and broken the plank you were nailing something into, bent the hammer as well as the nail, etc. But it would be a success, because studies show that you always hit the nail a sufficient amount of times for it to get stuck in the plank given enough time.

 

That doesn't make any sense.

You didn't say what the old wireless card is, but comparing the Intel 7260AC with something as old and power-hungry as the Intel 3945ABG can only save a 4 watts. Unless your old card was literally getting very hot to the touch, the wireless card's power savings would be even less.

Comparing the hard drives, there is also only a difference of 1-2 watts at idle and at load between the 850 Evo and the HGST 7K1000.

That doesn't make sense either. Besides the fact that an SSD or a HDD shouldn't make a difference, if the CPU and GPU are starved for data, they are working less hard and have lower power consumption.

 

Actually, the CPU & GPU works harder when starved as they are still clocked high while waiting. If the data is there quicker the program completes its work and can relax. I can see this more easily on my old P79 when booting. The Q9200 while peaking at the same CPU usage % levels does so for a very short time with the SSD compared to booting from a HDD.

 

You don't understand what the OP is talking about. First of all, for games and rendering, the hard drive is not a bottleneck for the GPU and CPU. Once the data gets loaded into memory, the hard drive makes almost no difference. You can play any game you want, but changing the hard drive between a 5400rpm laptop drive and a brand new 1TB M.2 SSD will not change your frame rates. The only things that are going to bottleneck the CPU or GPU are each other or the system RAM.

Secondly, you seem to be talking about power consumption over time. If we are trying to do a low-level format to a hard drive, the system with the SSD is going to get done faster than the system with a mechanical hard drive. The total system energy consumption over time will be less for the system with the SSD, since the system got done the work faster. However, the peak system power consumption will not be any different. The OP is talking about peak system power consumption. Watts are what we use to measure power. Watts have no relation to time. Energy is combines watts and time. There is a big difference in saying that my computer runs at 50 watts and my computer consumed 50 kilowatt-hours.

Lastly, if a CPU or GPU is bottlenecked, it means that although you want to use them at 100%, they have a lot of idle clock cycles where they are not doing any work. During these idle clock cycles, even if there are no power saving measures present to reduce power consumption during these times (there are though), they are going to use less power since they just sitting there not doing much. It would be inconceivable to think that a CPU or GPU would use MORE power while spending a lot of time idling than it would while working at full load.

 

I can only comment on my own observations, while not a gamer. Agreed it will not increase frame rates but if you have a SSD it will tend to load faster giving the CPU time to do its job and even delays for the CPU to get going are reduced as well. If he were running a RAIDx this could be further reduced as the stripes need not be combined in the background, this does not seem to be the case here.

Time for a CPU is not like you and me measuring it, it is in us. Every time the PCI bus is waiting for data it is tying up the CPU, just the nature of the beast. This is a bit less true where power savings and the CPU and GPU are at their lowered speeds as they go through a lot less and more miserly power cycles waiting for data and or release of the bus. The HDD being active for longer will cut into the power numbers too, not as much as he is saving though.

So you are correct, performance is not affected by much if any. Power savings, that can (and I repeat CAN) be affected.

 

I appreciate the responses. I had no intention on starting a argument.
Just thought I'd show some pictures, of the parts

http://imgur.com/a/tyUtp

 

No argument here, I am just stating what I have observed myself with the systems I've had to deal with. In the end my observations could be off as again I have seen this but it does not mean that is what is going on at your end. Even all that prior wireless card was CPU driven in the background eating away at power.