HD 3000 and discrete GPU dont operate properly together

I stressed both HD 3000 and 525m on my laptop and it seems that they cant be both at 100% usage. That affects my gaming performance. If HD 3000 is under use then the maximun usage percentage on the 525m drops which affects my gaming. Why does this happen? Is this a general problem to all optimus baring machines?

For example with windows aero running on HD 3000 and a game on 525m the game has less fps than without aero.

 

I think maybe you don't understand how Optimus works?

 

All i see is that optimus is not optimus after all.

 

Why would both cards be running at 100% at the same time ever?

 

Nvidia's Optimus solution works by sending the Geforce GPU stuff over the Intel IGP.
If the IGP was stressed then the Geforce GPU would be bottlenecked.

Same goes with Nvidia based 3D computers.
Nvidia 3D notebooks = No Optimus.

Your problem comes off that the game renders off the IGP rather than the Geforce GPU, thus affecting your performance since the Intel IGP does not perform as good with gaming.

Hopefully someone with knowledge about Optimus pops up.

 

optimus is named for its battery optimizing function...... it doesnt mean both igp and discrete gpu will work together.

 

Yes, you do not have Hybrid SLI with the 525M + HD 3000.

Choose one or the other, the other being the 525M which you should solely be using while gaming.

 

Technically they are working together, nVidia does the calculation, Intel displays the result..
But they are not working as in SLI or something like that.

 

Personally I think for those who have discrete graphics, switchable graphics bad idea.

Intel should just release CPU without integrated graphics. And then Nvidia and AMD should figure out a way so that on battery 50% or more of the cores are disabled to increase battery life.

Right now on my 5870M I can run the cores at 100mhz, but that's still 800 cores running. If there was only 200-300 cores running, would make a pretty big difference I think.

Both Nvidia Optimus and AMD Switchable Graphics seem to have lots of issues with Intel's IGP when it comes to gaming.

Personally on my next gaming laptop rig, I hope it has switchable graphics disabled.

 

Personally, I think the switchable graphics thing would be fine if manual switching was always an option and if AMD and Nvidia built switchable graphics support into their monthly driver releases. Then again, that would involve cooperation between competing companies, and it may or may not be possible (as I'm not certain how much of the switchable graphics tech changes from notebook to notebook).

For my own purposes, I'm very happy with my manual switchable graphics, I just wish it were easier to update the drivers for the respective GPU's. I feel like limiting your switchable GPU technology to an automated process is counter-intuitive since different users may use the same program in different ways, or have different expectations from it as far as performance, perhaps the same user might even want to use the IGP with a program in one situation and the discrete GPU with the same program in a different situation. Does Optimus or AMD's dynamic switching allow for that? Sure, for some programs, it's easy to say Yes I need the DGPU or No I don't need the DGPU, but I'd rather be given the opportunity to choose for myself rather than relying on a program to choose for me, especially if it's known for making the wrong decision from time to time.

 

Actually, Nvidia's Optimus are part of their official driver releases for quite some time.

Just only only a few didn't qualify if their OEM did something special outside of Nvidia's specification.

 

That's pretty much what I've been thinking. Just add a little extra info into the power profiles so that on the 2D profile 80% of the cores are disabled and then you could have the equivalent of an intel HD3000 when you need it, without the issues of current hybrid solutions.

The problem lies with the fact that if you have a certain amount of cores you can't simply half the frequency and then half the voltage. Once you get below about 0.9V you get start to tail off on the voltage required for a certain frequency because silicon simply won't function below a certain threshold. That works out as saying that running all the cores at 100Mhz rather than 200Mhz does not mean you half the power draw. If you simply turned off half the cores and then had the rest running at 200Mhz, you'd have the exact same performance as the previous solution, but you would half the power draw.

Also, a quick note. The current integrted graphics aren't compulsory. While they will always reside on the chip (assuming you have a processory with them built in), their use isn't compulsory. There's nothing from a hardware perspective to say that you can't simply use a single dedicated GPU.