DIY eGPU experiences

[Derbish]

I think it's an optimus settings and not a eGPU fault, i had to manually set the *.exe in the dGPU whitelist, so i think you have to do the same...

Im using an ATI card till my GTX 560ti gets here, but what does that mean "manually set the *.exe in the dGPU whitelist"

[SimoxTav]

Im using an ATI card till my GTX 560ti gets here, but what does that mean "manually set the *.exe in the dGPU whitelist"

It was a consideration of an option in the nVidia Control Panel (i thought u already had the 560Ti). On the ATI counterpart i know that only in case of BACON, you have to set the rule for the executable, i don't think such option is enabled with an eGPU setup. Anyway the issue sounds quite strange

[Derbish]

It was a consideration of an option in the nVidia Control Panel (i thought u already had the 560Ti). On the ATI counterpart i know that only in case of BACON, you have to set the rule for the executable, i don't think such option is enabled with an eGPU setup. Anyway the issue sounds quite strange

Fixed it. Right clicked the icon from desktop and used troubleshoot compatibility and whatever it did worked.

[viki@]

I was readng about usb 3.0 and thunderbolt, even with these two tech the full capability of gpu cannot be accessed.
Only x4 is max. So it cannot match the performance of the desktop...
Is my understanding correct ??

Sent from my HTC Vision using Tapatalk

[Grozar]

Hey all!

I ran into another problem: I've got a Dell Latitude E6410 notebook and PE4L with HD 4830 working fine, but...the card cannot accelerate flash videos...at least, there is no flash acceleration. The flash video goes green and every browser I tried freezes. HTML5 videos working fine, by the way. Is this a typical problem or it is a simple bug in the AMD driver? I'd like to use my ATI card with working flash acceleration.
I'm looking forward for your answer.

I believe this is just a bug.. I will ask around..

[timohour]

Hello I run some tests on my HP 4530s and 460GTX and I am getting the sense that Nvidia shut us down, stopping the Optimus acceleration on SB motherboards. I first installed desktop drivers 270.11 that came with the card and the result was 10975 Vantage points. Then I did a clean install of Windows and installed Optimus driver 285.36 (the latest modded) and the result was 10934. This made me nervous, so I ran several tests but none was better. All of them were around 11000. I am sure that the Optimus driver was installed because I could run on internal screen. Uninstalling the drivers and installing 285.36 Desktop (I would like to have exactly the same driver for the test) and it gave me around 10904. This is ridiculous. There is no performance increase at all. Maybe NVidia put something in the driver and when it tests the port and it is pci-e 2.0 optimus “acceleration” doesn’t engage. Firstly I thought it was my CPU (2310M) that bottles the system but we have seen greater NVidia Optimus increase with worse or same CPUs (i5-430M, P8700, T8100, T5800 etc).

Furthermore since I am moving occasionally I usually don’t carry a TV screen with me, and that was the reason that I grabbed a 460 NVidia, internal screen mode with Optimus (I personally like more Ati GPUs). Lately Lucidlogix announced a driver that would provide internal screen mode on SB H61, H67, Z68. Luckily for me the driver worked fine on HM65 chipset (HP motherboard 167C) with a watermark showing that my motherboard is not Virtu GPU (that’s how it is called) capable. So I ran back to back tests between Virtu GPU (with 285.36 desktop drivers) and optimus Performance (with 285.36 modded verde drivers). On the Virtu GPU Control Panel I slide the slider towards performance and… unbelievable. Scores talk by themselves 7455 Vantage gpu points for Virtu GPU driver while 6732 Vantage gpu points with optimus, both tests on internal screen 1366x768 (CPU: 2310M, 4GB DDR3 mainstream). As soon as I saw those results I was so disappointed by NVidia (for disabling Optimus acceleration on SB) that I decided to trade my 460s for a 6950 and yet I am looking for a new psu and waiting for pci-e x1 2.0 compatible hardware. Pictures below are added to show the facts. This also explains why serialk11r would score that low with GT240 @x1 2.0 Optimus. If anybody doesn’t believe me, he can try installing plain Desktop drivers and see the “difference” by himself. I am done with NVidia! Not only they don't add support but they cut for new hardware?

Images added

Hewlett Packard 4530s GTX 460 @ x1.1 (270.11 Desktop)

Hewlett Packard 4530s GTX 460 @ x1.1 Opt. (285.36 Verde)

Hewlett Packard 4530s GTX 460 @ x1.1 (285.36 Desktop)

Hewlett Packard 4530s GTX 460 @ x1.1 Opt. (285.36 Verde) internal screen 1366x768

Hewlett Packard 4530s GTX 460 @ x1.1 (285.36 Desktop) internal screen Virtu GPU 1366x768

Adding some more OC images just for instance @810 core 1860 memory

Hewlett Packard 4530s GTX 460 @ x1.1 (285.36 Desktop) @ 810 core 1860 memory Physx on

Hewlett Packard 4530s GTX 460 @ x1.1 (285.36 Desktop) @ 810 core 1860 memory

Hewlett Packard 4530s GTX 460 @ x1.1 (285.36 Desktop) internal screen Virtu GPU 1366x768 @ 810 1860 memory


Results with 2630QM and 6950 soon…

[SimoxTav]

To me sounds like there is a piece of the puzzle is missing. So many successful experiences and now this bad experience? Have you uninstalled the nVidia Audio driver? On one screen the CPU score is quite different from the other ones (probably cause Physx pushed the CPU)

[timohour]

To me sounds like there is a piece of the puzzle is missing. So many successful experiences and now this bad experience? Have you uninstalled the nVidia Audio driver? On one screen the CPU score is quite different from the other ones (probably cause Physx pushed the CPU)

Yeah thats the physics pushing the CPU on the last one. this was just for fun.
Since internal screen was running fluently, this means that Optimus has engaged. My card was running @ x1 (not x2 or x3) so no excuse for Optimus "acceleration not to engage, just the obvious. Nvidia shut us down.However just to be fair I would like someone with a SB laptop to confirm, since it could be just my problem
oooorrrrr
perhaps the Optimus Driver just uncrippled the Intel pci-e ports using an smart way to Enhance them. As we can see similar performance cpu post similar results eg p8700 @ x1 Opt. 10331 (very close to my results since mine was an Evga FPB board with slightly higher clocks). That's my guess on that. This is also why Optimus wouldn't run on x2 implementations.

[User Retired 2]

[timohour]

You can see my results with the 285.62 Verde driver confirm pci-e compression remains.

Offcourse Optimus internal screen mode is what everybody wants, but Virtu GPU works as well. It also runs smoother than Optimus (at least that's what I believe), runs great on AMD too (tested my 6950 soon posting results) and they are going to add support for amd Intergrated drivers (at least that is what they claim).