Sandy Bridge IGP power consumption with dual screen

[xinaes]

Following on from this thread now that Z68 is emerging.

http://www.anandtech.com/show/4330/asus-p8z68v-review shows that running Virtu with the screen attached to the IGP has some fairly modest power savings and significant performance impact.

The ultimate combination of P67 and H67 was inevitable—if Intel were going to market a series of processors with integrated graphics, it would seem odd not to include connectors on every chipset for those graphics. That's what happened with P67, when Intel decided that enough people wanted discrete GPU performance that the integrated GPU wouldn't get a look-in. Though with the virtues of Quick Sync, there had to be a way of providing one chipset that tried to do everything—overclocking the CPU, overclocking the integrated GPU, multiple discrete GPU setups, the works.

There was a sigh of relief when Lucid developed a software solution to allow them all to work together—the integrated GPU for low throughput graphics, a discrete GPU for gaming, and any situation to be able to use the integrated GPU for highly specialized applications. Virtu is a nice bit of kit, but in terms of power saving, if you're not using Quick Sync, there isn't much to be saved as graphics card manufacturers are pretty good on low power states.

I wonder how different this would be in dual screen setups, given that we know that at present low power states with multiple displays is not something graphics card manufacturers are good at. It seems possible - likely even - that the IGP would use considerably less power idling two screens than dedicated GPUs do and that as such the benefit of running Virtu in this mode could be considerable in terms of power saving.

Does anyone have any experience of how much power H67 uses when driving two displays, or thoughts on how this might be with the new chipset?

[matt_garman]

Does anyone have any experience of how much power H67 uses when driving two displays, or thoughts on how this might be with the new chipset?

I just built a system with the following specs:
CPU - i5-2600K
MB - Intel DH67GD
RAM - 2 x 4 GB 1333
HD - Intel 320 series 300 GB SSD

I drive dual digital monitors, one with DVI and one with HDMI (using an HDMI-to-DVI cable).

I've tried three different power supplies, and idle power consumption ranges from 28 watts to about 33 watts:
(1) The Ablecom 300 Watt 80+ Bronze PSU that came with my Supermicro CSE-731D-300B case gave the best results at around 28 Watts
(2) Seasonic SS-300ET is right around 30 watts
(3) Seasonic S12-II 430W gave about 33 watts

The numbers seem consistent with what other people with similar setups are reporting.

I think that if using two monitors does make a power difference, it's small enough to be hidden in with sample variance, component differences, etc.

[CA_Steve]

I drive dual digital monitors, one with DVI and one with HDMI (using an HDMI-to-DVI cable).

If you still have a watt meter handy, could you try measuring idle power with both monitors plugged in and enabled, and then with just one? A definative answer would be awesome.

If you have any gpu monitoring s/w, it would also be great to see how Intel varies the GPUclock and MemClock speeds for these two configurations.

Thanks!

[xinaes]

Does anyone have any experience of how much power H67 uses when driving two displays, or thoughts on how this might be with the new chipset?

I drive dual digital monitors, one with DVI and one with HDMI (using an HDMI-to-DVI cable).

I've tried three different power supplies, and idle power consumption ranges from 28 watts to about 33 watts:

This is quite promising. Thanks for posting.

It's also interesting to read that you are doing this with two digital outputs; I've read some FUD elsewhere that such a setup would require one screen to use VGA.

[matt_garman]

It's also interesting to read that you are doing this with two digital outputs; I've read some FUD elsewhere that such a setup would require one screen to use VGA.

Last time I checked, that was true for AMD integrated video.

[Vicotnik]

If you still have a watt meter handy, could you try measuring idle power with both monitors plugged in and enabled, and then with just one? A definative answer would be awesome.

If you have any gpu monitoring s/w, it would also be great to see how Intel varies the GPUclock and MemClock speeds for these two configurations.

Thanks!

+1

matt_garman, if you could do this is would be great. If you're on Win7, you can disable/enable screens easily with Win+P.

Last time I checked, that was true for AMD integrated video.

All integrated video beforce Sandy Bridge I believe. I'm very glad this is changing.

[matt_garman]

I just did a quick check. With dual screens, my system was holding steady at 33 Watts AC, mostly idle. Using a Kill-A-Watt.

I pulled the HDMI cable from the back of the PC. I watched the Kill-A-Watt for about a minute. It stayed on 31 Watts, occasionally dipping down to 30 briefly.

I don't have any GPU benchmarks installed, so I wasn't able to see exactly what was going on.

But, at least in my setup, dual screens seems to make a difference of about 2 Watts AC. Assuming my power supply (Seasonic SS-300ET) is 70% efficient at this low load, that's a 1.4 Watt DC difference. Negligible in my opinion.

-Matt

[CA_Steve]

That is awesome.