Lucid Virtu

[grilledcheese]

Seems broken/worthless? Doesn't cut my power use aged and my gtx 570 still does all the work even watching video's I thought my CPU's gpu supposed to take over??

 

[IanM]

it's not broken, it's just a big let down in the way it works. It will only idle the graphics card that isn't being used, so there is absolutely no advantage vs ordinary desktop usage. We need Lucid, AMD or Nvidia offer a version that actually switches off the graphics card while it's not in use.

I don't know if you have to set any other options to the Intel graphics to decode video for playback? but I thought Virtu and Quick Sync was about transcoding & editing video rather than playback.

 

[grilledcheese]

Nope read anadtech's article. http://www.anandtech.com/show/4199/...egrateddiscrete-gpu-on-sandy-bridge-platforms

 

[tangoseal]

Lucid Virtue was really for enabling Quicksync while using Discreet graphics. I use it and it works fantastically.

 

[defaultluser]

Yeah, the bull parroted around about it turning-off the second GPU was just that: likely right out of some marketing literature.

I believe the [H] article has most of the blame for that one:

The Lucid Virtu software allows for switchable graphics. This enables you to use only the iGPU for Quick Sync for transcoding and regular desktop usage which of course can mean a lot of savings when it comes to power usage.

This feature has two modes of operation. "I" mode and "d" mode. "I" mode provides a small hit in regard to performance because of overhead but allows the card to be idle most of the time while enjoying the features offered by the iGPU. This is really the method one uses for power saving.

Two claims in one page about mysterious power savings that were never actually verified with a power consumption test.

When I confronted Dan about it, he gave me a wave of the hand and swept it under the rug, [H] has yet to do another follow-up on the subject, and the rumors persist because of incomplete testing.

 

[jeremyshaw]

Because nVidia and AMD GPU have insanely low idle downclocks and undervolt themselves at idle, too...
making any sort of power saving marginal, since it has to activate the IGP, in addition to the fact that the mobo may not actually support fully powering off the dedicated GPU in a PCIe slot, while the system is on.

 

[Tsumi]

As I recall, nVidia's hybrid power fully powered down the GPU's. A test that I saw said that the gpu fan didn't even spin, and power consumption was only 3 watts or so higher than without the card.

While it may not be that useful for powersaving with one card, it would be useful for multiple cards.

 

[defaultluser]

Because nVidia and AMD GPU have insanely low idle downclocks and undervolt themselves at idle, too...

Okay, let's assume your meandering thought is actually true.

Explain than why Nvidia Optimus actually significantly reduces "non-gaming" power consumption versus having just the discrete graphics card. It should hit the battery twice as hard if you are correct.

The benefits of turning off a graphics card is ALWAYS significant. The difference here is Lucid either (A) plans to add this feature in a later update, or else (b) they figured nobody would question their marketing materials.

 

[jeremyshaw]

As I recall, nVidia's hybrid power fully powered down the GPU's. A test that I saw said that the gpu fan didn't even spin, and power consumption was only 3 watts or so higher than without the card.

While it may not be that useful for powersaving with one card, it would be useful for multiple cards.

yeah, since the IGP needed to be activated

Not only that, after the 8000 series, many vendors already implemented their own "idle clocks," in the BIOS and on the 200 series, nVidia officially implemented idle downclocking/undervolting, which is why the 9000 series is the last to support HybridGFx/HybridPower, whatever it was called.

I think AMD HD2000/HD3000 also idle clocks, but I never had any of those cards

 

[jeremyshaw]

Okay, let's assume your meandering thought is actually true.

Explain than why Nvidia Optimus actually significantly reduces power consumption versus having just the discrete graphics card.

The benefits of turning off a graphics card is ALWAYS significant. The difference here is Lucid either (A) plans to add this feature in a later update, or else (b) they figured nobody would question their marketing materials.

Lets assume laptops are the same as desktops

Okay... lets assume, for a moment, that Optimus is the same as Virtus' i-mode.

IF it were, you would be able to unplug your GPU card from the PCIe slot when it's off

Which was how nVidia Intro'd Optimus tech to the public, by physically removing the MXM card hosting a GPU.

As it stands, it's not possible. Just even try removing the PCIe plugs on the GPU when you are in i-mode.

 

[defaultluser]

Lets assume laptops are the same as desktops

Okay... lets assume, for a moment, that Optimus is the same as Virtus' i-mode.

IF it were, you would be able to unplug your GPU card from the PCIe slot when it's off

Which was how nVidia Intro'd Optimus tech to the public, by physically removing the MXM card hosting a GPU.

As it stands, it's not possible. Just even try removing the PCIe plugs on the GPU when you are in i-mode.

Well no, I'm not about to compare Optimus and Virtu because there are likely platform differences. It may not be possible to do it on the desktop, and let's leave it at that.

What I wanted to highlight was that, contrary to his claims, modern Nvidia and ATI GPUs do NOT idle at such low power levels that they are insignificant and immeasurable. If this were the case, Optimus would make things worse

Even worse: Lucid marketing literature is making power claims similar to that of Optimus without any real-world proof, which is the whole reason for this thread existing.

 

[jeremyshaw]

Well no, I'm not about to compare Optimus and Virtu because there are likely platform differences.

What I wanted to highlight was that, contrary to his claims, modern Nvidia and ATI GPUs do NOT idle at such low power levels that they are insignificant and immeasurable. If this were the case, Optimus would make things worse

The point was, on desktop platforms, the difference is marginal Otherwise, this would be in the Mobile Computing section. On laptops, power draw differences in the hundreds of milliWatts can make a difference, while the same thing on the desktop would be less than 10Watts (and that's more power than it takes for my m11xR2 takes to surf the web - powering a CPU, IGP, and 11.6" screen , and only ~30% of my m11xR2 at near full power, for reference). So yeah, it is marginal and nearly irrevelent, for desktops. Laptops are different, since you brought them into the conversation


EDIT: assuming a GTX580 idles at 35W, that's about the same as my m11x at full battery gaming mode (for note: Dell has a lot of strange power caps for the m11xR2, lol, if I were to override them, probably closer to 50W on the battery).

EDIT: nvm, GTX580 consumes about 26W at idle, the same (26,000 mW via BatteryBar Pro) as my m11xR2 on IGP gaming (starcraft 2)
http://www.xbitlabs.com/articles/graphics/display/geforce-gtx-580_5.html#sect0

 

[TehQuick]

The point was, on desktop platforms, the difference is marginal

I think the opposite point even such a marginal difference makes the Virtu tech useless. I mean what's the point of running in i mode if you are suffering a performance loss and none of the power-saving benefits that virtu is touting? You can just a easily just run everything of the dedicated videcard since the power usage is pretty much the same or, using your words, "the difference marginal", and enjoy full performance benefits whenever. What's more, i'm really concerned with latency penalty of the i-mode which, I suspect, is rather significant.

personally, I'm very disappointed with z68 and virtu tech. I thought there's really a point to switchable graphics but with current implementation, it has only disadvantages and no advantages.

 

[defaultluser]

The point was, on desktop platforms, the difference is marginal Otherwise, this would be in the Mobile Computing section. On laptops, power draw differences in the hundreds of milliWatts can make a difference, while the same thing on the desktop would be less than 10Watts (and that's more power than it takes for my m11xR2 takes to surf the web - powering a CPU, IGP, and 11.6" screen , and only ~30% of my m11xR2 at near full power, for reference). So yeah, it is marginal and nearly irrevelent, for desktops. Laptops are different, since you brought them into the conversation


EDIT: assuming a GTX580 idles at 35W, that's about the same as my m11x at full battery gaming mode (for note: Dell has a lot of strange power caps for the m11xR2, lol, if I were to override them, probably closer to 50W on the battery).

EDIT: nvm, GTX580 consumes about 26W at idle, the same (26,000 mW via BatteryBar Pro) as my m11xR2 on IGP gaming (starcraft 2)
http://www.xbitlabs.com/articles/graphics/display/geforce-gtx-580_5.html#sect0

The point I've been trying to make is that the idle power of a discrete GPU is significant to some. Take the HTPC users, for instance: they want the ultimate in low power consumption because the PC is on 24/7, and because every watt you consume you must use fans to evict.

For some people, integrated graphics is enough...but others would desire more power in their HTPC WITHOUT the increased idle power. Adding a GTX 580 to an HTPC would basically double idle power, which may make a big difference in noise if you have a very small case. Virtu promises the power of a GTX 580 without any idle increase, but does not deliver.

 

[Ruoh]

Perhaps this is where my initial confusion started as well. :/

 

[jeremyshaw]

The point I've been trying to make is that the idle power of a discrete GPU is significant to some. Take the HTPC users, for instance: they want the ultimate in low power consumption because the PC is on 24/7, and because every watt you consume you must use fans to evict.

For some people, integrated graphics is enough...but others would desire more power in their HTPC WITHOUT the increased idle power. Adding a GTX 580 to an HTPC would basically double idle power, which may make a big difference in noise if you have a very small case. Virtu promises the power of a GTX 580 without any idle increase, but does not deliver.

A GTX580 for HTPC's idle heat can be better dealt in so many other ways, however. I could see some use for Virtu in the HTPC sense, since the Intel IGP can power video decoding without requring the nVidia GPU's higher "video decoding," clock state... (which brings up another beef of mine: nVidia, to support 2560x1440/1600 monitors, requres the GPU to be clocked to it's "video decoding/405MHz" clocks ).

Perhaps this is where my initial confusion started as well. :/

I guess defaultuser's first post could of been read out of context in the sense that I just realized Virtu was aiming at.

EDIT: I TAKE THAT BACK!! nVidia 180.xx beta driver allow full downclocking to 50MHz on 2560x1440!! Never did in the 175.xx drivers, now I gotta :wub: some 180 hawtsausss....