Lambda

Is there a money-back if setup 1.x doesn't help?

They could make USB 3.0 port and not just USB 2.0. Somehow USB 3.0 is only available on PE3A which is for ExpressCard only.

You can try Windows 10 for free and report your results. You can also try Ubuntu 14.04 or 12.04. They have good support. Many games can be installed successfully there with automated scripts (for wine) and they can be even quicker there. You can give up if simply guidances don't work for you.

Optimus should be the best solution. I consider the discussed method rather as workaround. Now I see that Ultramon cannot tranfer mouse and keyboard input between two screens and this should be done by OS and we should check how some system processes load CPU instead. I'll correct my 1st post. Now I see the discussed method such that we need Ultramon because we need to see the dummy screen (and to set primary display if this cannot be done by OS or iGPU's software) to start a game on the eGPU (I though this can be done easily not seeing the dummy screen, I'll check this). And the limitation with window mode comes from that we have to drag that game to iLCD with mouse. I though this is possible to do in other ways. I'll try the ones mentioned here and try to find others if needed. I have Ultarmon but I don't have PE4C yet and eGPU so cannot make comprehensive test. That may be in one month or later.

It seems I haven't understood completely right what you said about Optimus before and and Intel iGPU + NVidia GTS450 will work with Optimus. And for a slow PCEe connection Optimus might help, that is why I plan to buy one of those nVidia cards. But what about DirectX 10, 11, 12? E.g., GTX 550, 570 and many others relatively old video cards are said to support Directx 11 and partly 12. And what is the most important nVidia Optimus seems to be not a reliable solution. It can disappear because of not appropriate driver support or newer hardware without that support. NVidia can stop it. Also I saw complains that Optimus under WinXP created serious problems... LucidLogix Virtu solutions seems to be more attractive. But I saw somewhere that it's not for all nVidia and AMD video cards. Maybe it is for the listed ones AND newer ones. But one could have older motherboard or just a different CPU. There are simplified CPUs Intel Atom, Pentium, Celeron and others under production, and there are also AMD CPUs. So if you like to play with your pal who was not aware of these hardware restrictions (pretty strict ones), you will have a problem. Ultramon/Chung-gun method seems to be a simpler software realization and one can have the largest expectation that it will work in the future versions of Windows and there certainly will be corresponding software to realize this method also under newer versions of Windows. I believe you this method works. But as I mentioned above, there is no warranty you will be able to start a game in window mode. And this limitation may be not necessary. I don't see why it is necessary yet. Maybe we can ask to make changes in Ultramon or an alternative software and that will be quite small and easy modification. And there is also question about CPU load by Ultramon and by a windowing program if any is used. If it loads one core almost completely, then this may be a problem. If this CPU load is smaller, then it's ok. A modern game may need 2 cores with around 2 GHz frequency. Look at CPU requirements of, say, Crysis 3 or AOW 3.

Hi everyone! I am interested in easy and universal solution for using eGPU with internal LCD. Then the whole eGPU set can be made quite portable. I propose to extend what was discussed and described here and here. Framegrabbers albeit always ready may be considered as relatively expensive solution and they occupy slot/port and add one more cable to your system. Let's share ideas and experience, namely, 1) how to start a game on iLCD by means of some software, 2) CPU load with such solutions, 3) problematic games. 1) The most common solution is through Ultramon. Its developers say it is video card independent as should be according to its functions. Also Actual Window Manager is mentioned as an alternative. Still I don't know what is not enough, if this is so, for Win 7, 8 display cloning to be able to run games on iLCD. This may be asked to be done in the next Windows. Then all report that a game should run in a window. The best solution will be the game's supported window mode. This option can be present somewhere in the game's dialogs (e.g., Oblivion). Also Alt-Enter may help (e.g., AOW SM). In some cases adding parameters "-W", "-w", "-Window", "-window" can help (e.g., Fallout tactics). D3Dwindower succesfully windowed Starcraft, but after its application to AOW SM the image in the window was unacceptable at all and after click the game restored its fullscreen mode. D3Dwindower didn't affect Fallout Tactics at all. With DxWnd and 3D Analyze I've got even worse results albeit DxWnd is newer than D3Dwindower. I hope I'll be able to run all the games in window mode. But is window mode really needed? This part of the text is somewhat technical and may be more appropriate for the stackoverflow site, and a common reader can avoid it. In Windows XP I checked with spy++ the 3 mentioned games and found that all they have windows albeit they run in fullscreen mode. Maybe there is no other possible realization through DirectX or OpenGL. So I don't understand why that window is not enough (I havn't tried that yet anyhow). And if a game is windowed successfully it can try to recover to its original state, as it was in AOW SM. Then a new created window with grabbed frames and tuned input might be a workaround. 2) I propose to find out what CPU load can we have using the method described above with different software. With my preliminary test of mirroring an application at 1280x800 & 25fps through Ultramon (the window's size was not maximized and was around 1024x768) I got around 7% CPU load, not 40% I found somewhere. To make unambigous testing we need to find MHz or GHz (not %) in the following way. You can use Window's Task Manager (open list of processes) or free Procexp (for WinXP or later). We need to know what 100% means there. Also we need to know current CPU's or core's frequency. Most likely your CPU reduces its frequency when its load is small, but I don't know that threshould % so to be sure we need to know CPU's frequency and to have overestimated result in MHz. In WinXP this can be checked in CPU-Z (I have ver. 1.71). In Win 8 or maybe earlier versions Task Manager can also show current CPU's frequency. In the case of present Turbo Boost technology (many CPUs already have it; probably it can be switched off in BIOS) separate cores can increase their frequency above the nominal one. As I have CPU without Turbo Boost, I don't know how to get the current frequency of the core running the processes of interest and what 100% CPU load will mean in Task Manager or ProceXP, say, will there be more than 100% if all the cores are loaded fully? You may need to figure out this. But don't forget to save your unsaved info before 100% CPU load. To load fully a core you can create the following 2-bytes program: EB FE (in hex coding). Alternatively you can enter them in a text editor by their codes through a standard Windows symbol table application. Name it with "exe" extension or "com" as an alternative. In WinXP I had to press "ignore" button and nothing in Win8. Then ntvdm parent process will fully or almost fully load your core. Now examples how to understand the percents. In the case of 2 cores CPU without Turbo Boost one ntvdm process should load CPU approximately on 50% meaning fully loaded one core working on its maximal CPU's frequency. So in this case you should multiply a process's percent by 2 (the number of cores) and then on the CPU's frequency. In the case of 2 cores and 4 threads (or 4 virtual cores, hyperthreading technology) without Turbo Boost one ntvdm process loads CPU approximately on 25%. AFAIK each real core shares its resources between its own threads. So 25% should mean one fully loaded virtual core with 2 times less CPU's frequency. Then a process's percent should be multiplied by #threads/(#threads/#cores)=#cores, as in the previous case. Then 7% in my 2-core CPU with 2.16 GHz would mean just 302 MHz or just 14% CPU load on a one core CPU with 2.16 GHz. Please, present your results on the used software if they are notably different. Also specify your used software, OS and your GPUs. Resolution 1280x800, game's window maximaized or made maximal. CPU load should depend on this. 3) If you a game which could not be run on internal LCD by the mentioned methods, please, specify it to know the situation.

Thank you very much! I don't use my laptop as a portable one any more. Also it is standing on one book for better cooling, which will facilitate plugging to the mPCIe. I saw mPCIe extender in the form of narrow and wide cable somewhere here. I don't know whether it has good enough screening but it should be more convenient. About the EXP GDC V6. From your link I find about possible "faulty mPCIe/EC cables". Don't want to risk now. The PE4C V1.2 has ATX 24-pin slot to conveniently turn on PSU, supports Linux and seems to be better in avoiding registration problems when booting. It is only $18 more expensive. Just want to be sure first that I don't need to solder the cable to the 2 units. Also I have to note that I have Windows XP SP3. I saw confirmations that an eGPU can be recognized in WinXP. One may need to log out and then log in for this. But Optimus may not work albeit nVidia still release drivers for WinXP. I'll try Ultramon if fail with Optimus, or will try installed Linux or plug in external monitor. Win 7 or 8 will hardly ever be installed. First I should better upgrade HDD. And I have GMA950, not a HD iGPU. I haven't got whether Optimus is possible in this case but with smaller performance. I have plans to buy a new laptop but that will be later. Now I would like to have temporary solution. Also the new laptop can have troubles which the current one will not have. So there are several reasons to try eGPU on the current laptop too. And as you mentioned, newer nVidia drivers can have no Optimus support. And again about bandwidth. I didn't take into account bitmapped textures (compressed afaik) if they are loaded frequently. Don't recall about other informationally consumptive ingredients. Then transfer to the eGPU becomes comparable with transfer from the GPU (if used). But if I reduce resolution I have good chance to play a game with high or low details but the eGPU will work. Is there a known case when a game could be played with 25+ fps on eGPU because of low bandwidth of mPCIe v1? So "unsupported" videocards may be unsupported only relatively (but still be able to reveal better performance) if I understand everything correctly. I am also curious about eGPUs perspective as they are so convenient. Update: Here is the source, where GPU switching in WInXP is said to be possible by mean log off/on. At the beginning I though that will be enough to disable iGPU in Device Manager. WinXP may be not smart enough. By the way, since some moment, if I am not wrong, there appeared the copy of IGPU. Also I have 3 monitors in Device Manager...

Hello, everyone! I want to make my choice what adapter to choose. I am prone to save money at the moment and to see whether the proposed adapters are feasible for my laptop. But if everything is ok, that would be good to have effective solution for eGPU. My laptop is HP 530 with Intel Core Due 2.1 GHz CPU. It has mPCIe (probably gen2 and not just gen1) and slow PC card slot (133 MB/s). The PC card slot could in principle be used as a slow variant of expresscard after the corresponding adapter is bought. I plan to buy a GPU like GTX 550-570 with possible further overclocking. So the power supply greater than 220W (as in PE4C v2.1 package) might be needed. Is there any model where power supply could be added, not just somehow selected one, when more than one source is plugged in? From PE4C v2.1 description I find "Power Supply: ATX PSU, DC adapter, 8PIN connector (for D220P AC adapter),Three select one". The PE4C v1.2 is badly described. On its picture I see 4 pinholes near a switcher and 4-pin slot. Are not they for power supply? But it seems to be not a big problem if you have PSU with 6 and 8 pins, though its overall power should be high enough. And does not the cable from the adapter to the mPCIe plug require to solder it? There 3 cables in tha package and I don't see how do they stick into. Also I want to use laptop's LCD, not a cumbersome external monitor. It seems that all models can grant this possibility. From http://forum.techinferno.com/enclosures-adapters/3100-pe4h-vs-pe4l.html I find "PE4L 2.1 and PE4H 3.2 have 1.2 Opt support". Is there an additional support for nVidia Optimus in the mentioned adapters and if so which one has the best support? And about bandwidth. Gen1 mPCIe can pass 250 MB/s. Let's suppose we have 1280x800 32bit (I don't have larger) resolution with 25fps. Then we need 100 MB/s bandwidth if we want to transfer these uncompressed frames back to the laptop's LCD. Even if we have 1000 objects to render (game developers can give more realistic number) and to send them to the GPU, this seems to be a small percent of 100 MB/s. Then 250 MB/s seems to be enough for all the modern games and probably for the ones to appear in 5-10 years provided we use the mentioned resolution. Taking the above mentioned into account I would buy the PE4C v1.2. It is the cheapest one and can allow up to 500 MB/s bandwidth for a larger resolution. And the cables from its package look certainly different than the PCIEMM-060A cable which is said to be for gen1 PCIe. For, say, PE4H both PCIEMM-060A and PCIEMM-060B cable are proposed but I could not find how to choose PCIEMM-060B.