DIY eGPU experiences

oh yea kizwan how do you check to see what ports your devices are on....i am guessing i am screwed with the ports being on the wrong sets....just knowing my recent luck with this thing. BTW if i lost you...I am talking about the port 1+2 and so forth with the mPCIe and expresscard

 

Try this one:-

You can use AIDA64 >> Motherboard >> Chipset >> South Bridge.

 

is there a free program to use?

 

Youch.

PE4H is the premium product for users with pci-e 1.1 devices (Series-5 and older) where is bests the PE4L 2.1 by offering up to x4 link width and a detachable cable. For the X220T the PE4L 2.1's pci-e 2.0 capabilities are definiteily worth it for the improved performance (see x1.1Opt vs x1.2Opt here). So you would have a ebay buyer for your PE4H if you wanted the performance upgrade to a PE4L 2.1.

Quite a few Dell systems have TOLUD=3.5GB. So no amount of using Setup 1.x will help allocate the eGPU into space that doesn't exist. The solution is to do a DSDT override.

See Prepurchase FAQ-> 2. How can I check if my notebook is compatible with an eGPU? for details on how to check TOLUD and a link to the DSDT override page.

 

Tried and the old 8800GTS is now shown as "3D-Videocontroller". No drivers were found by windows. If I force the latest nVidia driver (296.10 x64) it fails before installation. Error message says something about not useable for this windows version.

 

Thank you, nando4, for the expedient response!

I've made it to fixing errors in my .dsl file, and while there's 3 warnings, 12 remarks and 11 Optimizations, there's only one error. I used Kizwan's guide (Thanks Kizwan!) to sort out the parse error, but the line of code it took me to was not an If statement, but just Arg0 (the 0 seems to be what's specified as the parse error)

What does that mean, and how would I resolve the issue?

Here is the .dsl file in question
And here is the error reported:

Code:

dsdt_CALPELLA.dsl 4751:                            Arg0
Error 4096                                     parse error ^

UPDATE: Nevermind. It looks like I managed to solve the issue. GTX560 up and running.

 

So ive tried to hault the system on the boot screen where you select the operating system / options... And then switching on the Egpu.. but it didnt work.. wasnt detected in devmgmt.. I have also tired using the delay switches on the Egpu thing as well as the minipcie thing where the wireless card usually goes.. but none work.. I can see by the LED's the actual delay side of things is working, but the card doesnt show in device manager, and I have tried scanning for hardware too.. What am I doing wrong is there a step by step guide?

If I switch it all on before boot it just make the laptop screen and the screen blank. Can anyone help me?

 

I hope my post belongs here... I bought a vidock 2 from ebay for around a 100€ since it seemed like a good deal considering the reduced waiting time (for me) and the inexpensive price. I removed the front end of the enclosure, plugged in a pci-express 6-pin for my 8800gt (bios flashed to quadro fx3700) and plugged the expresscard into the slot of the T520. The card is driving two external displays at 1440x900 and 1680x1050. So far everything works well, GPU-Z reports pci-e 2.0x16@x1 2.0... Using 296.10 drivers for Windows 7 64-bit, I disabled the NVS4200 gpu as it reported code 43 when used alongside 8800gt. The HD3000 of the i5-2520M cpu drives the 1600x900 internal display of the notebook. I have 8 gigs of ddr3-1333 memory installed.


I've only ran 3d mark 06 once and the results are as such...

3DMark Score 7,222.0 3DMarks
SM2.0 Score 2,995.0
HDR/SM3.0 Score 2,758.0
CPU Score 2,959.0

 

Hey guys,

I just finished my eGPU, but I got some problems with it.

Laptop is and old:
HP Compaq 6720s ICH8 bridge,
T9300 2.5 GHz
4 GB DDR2
Intel X3100 iGPU
WinFast GTX 560 eGPU

In first stage I used a no-name 450 W power supply 12V/18A (1 FDD, 4 molex connectors, and that's it, all were used by the eGPU, because the FDD powers the PE4H, and the 4 molex converted to 2x6 pin PCIe) but when the GPU is at high capacity after 10 minutes it tends to shut-down (BSOD) and reboot. In desktop applications works like a charm.

So I decided and bought a Cooler Master GX 500W Lite, with 12V/36A. When I set it up and plug the express-card in, the 3 leds (3.3, 5, 12V) on the PE4H lights on but in 5 seconds when the monitor should turn on, the 12V led on the PE4H (v2.4a) shuts down, the cooler on the eGPU and PSU stops and I got a freeze or BSOD and reboot. (So most probably something is causing PSU to shut-down...)

I have to mention that I tried with different drivers, same result, and tried with and without the 12V bypass jumper set on the PE4H.

Do you have any ideas what is this cause, some protection on the PSU (because it has OVP, SCP, OPP protections), the no-name has no protection and it's working while it's not heavy loaded, why does the PSU stopping? Has anyone else used a Cooler Master PSU with the eGPU?

I also managed somehow to make it boot in windows, but right after the 'Welcome' loading screen the PSU and eGPU stops (no power from PSU) and laptop reboots. Any help please?

 

So is there no enclosure available for purchase yet? I won't have time to build one for a long time

 

Firstly, your bios must not see the eGPU on bootup. Now without Setup 1.x to help you along, I'd proceed as followed:

- Boot the system with wifi card, confirm it's working (this overcomes any whitelisting)
- disable the wifi card in Windows device manager
- carefully remove the wifi card while system is still on.
- power on the eGPU and *carefully* hotplug the PM3N in the wifi slot
- do a Device Manager scan. The eGPU should appear as a device. If not, then there is a x1/x16 switch on the PE4L 2.1. Toggle the switch.

Not clear at all why that Coolermaster is doing that. However, the cheap PSU does sort ot work. Could try to reduce the load the GTX560 places on that PSU by underclocking and undervolting it with MSI Afterburner. Likely to run without BSODing after 10mins then.

PE4H 3.1 will have an enclosure option. It's tentatively released in the next two months. A PE4L 2.1 otherwise can be housed in a small shoebox as an interim solution while contemplating a proper enclosure build.

 

I just ordered a PE4L 2.1 from PE4L ( PCIe Adapter ver2.1 ) I was wondering what the difference between PE4L 2.1b and PE4L 2.1c is since I keep seeing the 'b' and 'c' versions pop up on this thread. Will I be receiving the latest revision?

My second question is how a video card (lets say GTX 560Ti) physically fits on the small PE4L PCI-e board. I take it that half of the pins on the video card will be connected to nothing?

P.S. I own a ( price errored) Lenovo T520 with an i7-2860qm and Optimus NVS4200M/HD3000 and plan to run games off of external monitor(s). Does my computer + PE4L 2.1 + GTX 560Ti + appropriate power supply sound like the right setup?

 

You're allowed to use AIDA64 trial version.

You might want to try older version. Every time cold/warm boot, always connected/power on eGPU at "Starting Windows" + animated windows logo screen.

Did windows 7 successfully re-allocate eGPU to 36-bit bridge? If yes, please post screenshot of Device Manager ( >> View >> Resources by connection) showing eGPU in 36-bit bridge.

You should/will get the latest version.

Yes, it is PCIe x1 slot with the other end opened, so you can attach full-size GPU to the slot. Approximately 3/4 of the pins will not connected to anything.

Sounds right.

 

Mac OS X + eGPU + DSDT Override

<strike> DSDT Override also does work in Mac OS X (tested on Mac OS X Lion). Mac OS X also re-allocate eGPU in 36-bit space. The "Large Memory" I use is 0xC20000000 - 0xE0FFFFFFF but</strike> Mac OS X re-allocate eGPU to 0xF10000000.

EDIT: I just tested again without DSDT Override & I get the same result. Look like Mac OS X Lion able to re-allocate eGPU to 36-bit bridge when there is insufficient 32-bit space without DSDT Override.

 

Do you use a hackintosh or a "real" MBP ?

 

It is Hackintosh. For real Macbook, will need to figure out how to load DSDT override.

 

OK, to be honest, i don't understand anything about DSDT override

 

I just tested again without DSDT Override & I get the same result. Look like Mac OS X Lion able to re-allocate eGPU to 36-bit bridge when there is insufficient 32-bit space without DSDT Override.

 

osx is also more egpu friendly than win7

 

Hi all,
I've been following the eGPU thing for some time now and am wondering, what the next couple of month will bring in that area.

The reason for me asking is that I got 2 options here:
1. use an old laptop (got 3 to choose from) via express card or mPCIe and the PE4L or PE4H KIT
2. wait until there is a reasonable solution with thunderbolt and then buy a awesome ivy bridge thunderbolt ultra book.

Now I got the problem, that I dont really know if it would be worth waiting for a Thunderbolt Solution. As far as I know TB would be 4 PCIe Lanes is that right ? So generally speaking, you should be able to quadruple the bandwith compared to current solutions like the PE4L or PE4H ?
Is that assumption right ?
If yes, it would be possible to use some high end gfx cards as bandwidth should be no limitation anymore, right ?

As far as I know, there are already some out of the box Thunderbolt Solutions available ( e.g. from magma) but they are way too expensive and on
http://www.hwtools.net I cant see any news of such a solution coming soon ?

So it would be very much appreciated if someone could enlighten me here ! Maybe there is a sub-forum where you can find information about future products or development.

Thanks,
Dave

 

Looks like OS-X ignores the root bridge definition and allows 64-bit BARs to be allocated in 64-bit space. So a Hackintosh is not going to see the same sort of problems a Windows system does. Good work yet again that's been linked to the DSDT override page.

The best performance from current expresscard solutions is "x1.2Opt". It is around x2 2.0 bandwidth, so Thunderbolt will get double that. Additional feature of TB is Displayport input into the LCD so the eGPU can drive the internal LCD without incurring a bandwidth penalty. T

No subforum but I wish we did have one (requested twice already and got knocked back by the mods). See the Thunderbolt, USB 3.0, PCIe 2.0 eGPU update for latest status on Thunderbolt products.

The MSI GUS-II is going to be the most accessible Thunderbolt enclosure if the $150-$200 pricepoint rumor is true. Magma/Sonnet are both over $600. Villagetronic and BPlus need to wait in queue for Intel to release Thunderbolt chips to small developers before they can provide Thunderbolt enclosures with no ETA in sight.

 

In my limited experience with the pci-e x1 2.0 bandwidth operation of a 8800gt (a notch above entry level nowadays, only 512MB of gddr3) I managed to get a general grasp of things related to bandwidth. For this purpose, I used gpu-z on one display to monitor the gpu as I played games and ran benchmarks on the other. What surprised me the most was running the ps2 game black on pcsx2 as I already had experience with that game using the 1GB NVS4200 of my T520. With the NVS4200 on the x16 internal pci-e bus, the game ran slow overall (not choppy though, just slow motion). With the 8800gt (which has less memory but much higher memory bandwidth) on the expresscard interface, the game ran very smooth and fast for a couple seconds, before it froze for five seconds or more and this cycle was consistently repeated throughout the level. Now, I believe this is because PS2 has very little memory so fps games are designed to constantly load data from the dvd to the graphics memory, instead of loading them at once or in larger chunks, causing such behavior.

This became more clear to me when I decided to test crysis on an empty multiplayer map. This time, the game ran smooth until I covered some distance in the map, which caused a second or two of stuttering before it was smooth again. Mind you, the game used around 500megs of graphics memory, and with lower settings I believe this might also be avoided. In short, I think the bandwidth is only an issue as a result of game design. If you have enough graphics memory and on a multiplayer map the game loads all large textures before the map begins, you should have a smooth experience, close to what you would achieve with x16 pci-e bandwidth. Since most of my favorite games are map based such as multiplayer fps, rts and sports, I hope to have a better gaming experience especially as I switch to a next generation nvidia card, enabling optimus compression and a large graphics memory. I too had some doubts about not waiting for thunderbolt, but seeing that you already have the laptops, it won't cost you much to get some decent gaming experience.

 

First of all , thanks nando and ufster for your quick reply.

x1.2Opt means 1 PCIe Lane on the PCIe 2.0 Standard with Optimus enabled, right? And Optimus I can only use if I got a GPU integrated in the Processing Unit like the hd3000 or is it also possible with an older "on board" gpu chip?

Is there already a PSU included ? Seems pretty cheap to me compared to the other solutions out there.

Do you know when a DIY Thunderbolt Kit is available ? I am just not sure if I should buy a Express Card Solution now, when you can have a TB Solution for the same price ( or probably a bit more expensive ) in June.





Another Question I still have is this:
Can I upgrade parts of the Set ? e.g. when I buy the PE4L now, is it possible to just swap the express card with a thunder bolt adapter ?

 

optimus can be enabled ONLY if you have an Intel 4500MHD, HD or HD3000 iGPU.

 

Not sure if this is the info you're looking for, but here's the card in device manager (large memory is a new entry):



I assume I got it into 36, but I wouldn't recognize it by looking. All I know is that the card is outputting video to my monitor (I'm using it now) and that things like The Witcher 2 actually run with decent settings on my XPS now.

A great many thanks to everyone and their help, be it direct or vicarious.

 

Optimus' pci-e compression engages when there is a X3150, 4500MHD, HD, HD3000 and likely Ivy Bridge HD4000 iGPU as the active video device and a x1 eGPU link.

THe MSI GUS-II supposedly comes with a 150W PSU. Yes, if the GUS-II has a 150W PSU and is a Thunderbolt enclosure and costs $200, then it will storm the market. Sonnet are charging $800 for their Echo express with similar specs.

You cannot upgrade the PE4L. Though when you consider it's US$92-shipped you'd likely lose a max of $40 if you sold it on ebay and replaced it with a Thunderbolt enclosure.

Yes, the GTX560 is now allocated to 36bit space.

Could you please post an experience post with details of what was required to get it all working with 4GB+? The purpose being to list the system details, steps required and some benchmark results so another user with the same/similar system could get a very good headstart on what's required. Pls include the modified DSDT file. Note too that your XPS 15 (L501x) is likely x2 1.0 capable where you combine expresscard+mPCIe or mPCIe+mPCIe ports to double bandwidth and improve gaming smoothness. Those slots need to correspond to be [port1+2], [port3+4], [port5+6] or [port7+8].

 

While I won't have the time to do a write up until later next week (it's the end of my semester, so final exams are upon me) I do plan to chronicle my setup.

Unfortunately my unit doesn't have/wasn't offered with an expresscard port. I do have an additional mPCIe port if I remove the wireless card, but it's a shorter space. Is there a shorter PM3N, perhaps without the legs?

 

I have a dumb question. Is it possible to attach a GTX590 or GTX690 eGPU with those new DSST Override method to allocate more memory ?

I googled and found a small bit in Setup 1.x history which confirmed that there's a workaround for GTX590 to work ??
http://egpu.maeth.net/history.txt

 

There is a perforation along the "legs" designed to allow you to snap the PM3N into a half-height card. Suggest identify your port numbers first using SIVS or AIDA64 to see if you would want to use the wifi slot to do a x2 implementation.

The DSDT override method makes it possible to host a lot more GPUs as there is an abundance of pci-e space to work with. After extensive PM-based troubleshooiting with NewNet we were not able to overcome error10/43 with the GTX590. Which is not to say it may not be able to be made to work. It might be as simple as captureing a 'good' pcie configuration space dump and pre-loading that via Setup 1.x prior to chainloading. If I had a dual-GPU card like a GTX590 I could play with it to see what is necessary to get it to work.

 

I bought an EC2C a month ago and disassembled it.
EC2C 1.41 is not mentioned here.
What is the difference between 1.41 and 1.40?

 

Thanks for this info, but I have to ask just to be 100% sure. Are these iGPU's (X3150, 4500MHD, HD, HD3000 and HD4000) strictly the only ones, where one can use Optimus on a Express Card Solution ?
What about the rest like the 4700MHD, 4500M or X3100.
I am asking because I have an old Latitude D530, Core 2 Duo T7250, 2 GB Ram, X3100 GPU with a Express Card Slot, which I would love to make use of for this experiment.

If that's not working I would have a newer Dell Latitude E6410, i5 560M, 4Gb Ram, Intel HD GPU with a Express Card Slot aswell.
Here I already checked that my TOLUD is 3,5 GB so I probably need to use that Override hack, but besides of that, I think with that setup it should work what do you guys think ?

 

Power requirements question. I have this power supply already:
Thermaltakeusa»Power Supply»TR2 Series»TR2 Standard 500W PSU : TR2 Standard 500W PSU W0093RU

It says it can supply 14A on +12v1 and 15A on +12v2.

I would like to run:
Newegg.com - EVGA 012-P3-1571-KR GeForce GTX 570 HD w/Display-Port (Fermi) 1280MB 320-bit GDDR5 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card

which apparently needs upwards of 219W... if these rails are indeed additive, I should be in good shape... the plan was to use my one PCIe power adapter and combine two other Molex plugs into another one.

Should be fine for now at least? Or do I need to go investing in another psupply?

 

Hey guys,

I was just wondering if I am able to have an eGPU with an Acer Aspire AS5750Z-4835. If I am able to have one how exactly do I go about it (just point me in a general direction)?

Best regards,
MST

 

Optimus ONLY possible with Intel X3150, Intel 4500MHD, Intel HD Graphics & Intel HD 3000. Optimus also most likely possible with Ivy Bridge Intel HD 4000. Intel X3100 doesn't support Optimus.

Do you mean you already have Dell Latitude E6410? If yes, you can get either PE4H-EC2C (v2.4) or PE4L-EC2C (v1.5). If no, then better get SB or wait Ivy Bridge notebook instead & you can get PE4L-EC2C v2.1. If TOLUD is 3.5GB, you can use DSDT Override to help Windows 7 re-allocate eGPU to 36-bit space.

14A on +12v1: +12V1 * 14A = 168W

PSU 168W vs. GPU 219W, this show that PSU is not suitable to drive GTX570. You need a PSU with at least 20A/21A on +12V1 or +12V (single rail).

Your notebook specification show it doesn't have ExpressCard slot. This means you'll need to use mPCIe port for eGPU & you probably need to sacrifice wireless when you're using eGPU. Your notebook have Intel 6 Series chipset, so you'll need to wait for PE4L-PM3N v2.1 ¹.

¹ The product official name is "PE4L-PM060A V2.1b". PE4L portion is same as PE4L-EC2C v2.1b、and PM060A stands for Pcie Minicard (Half size) and 60cm cable. (Source: Thunderbolt, USB 3.0 and PCIe 2.0 eGPU update)

 

Hi Kizwan, thanks for your help.
Yes, I do have a Latitude E6410. Only Problem is, that this is the Laptop from the company I am working at. But that shouldnt be that big of a deal, cause I got admin rights and its not that I can destroy the machine with an eGPU setup right ?

The difference between PE4H-EC2C (v2.4) and PE4L-EC2C (v1.5) is that with a
PE4H I can add more PCIe Lanes if available (like a mPCIe slot) and with a PE4L I can only use 1 PCIe Lane, is that right ?

Both of them are able to use PCIe 2.0, right ?

So if I only want to use my Express Card Slot, then the best choice would be the PE4L-EC2C (v1.5) setup ?

Thanks again to all you smart people out there

 

Ugh... I knew I'd be SOL there. Any recommendations?

Also, are the 570's overkill for eGPU? I was mainly only interested because they were among the few nVidia cards to have DisplayPort out on them, but if the rest of the card can't be utilized, I'm not sure it would be worth it.

 

I think dual-GPU graphic card can work on dynamic TOLUD notebook, e.g. Lenovo x220 & my notebooks. I hope so.

Yes, no mod needed on the notebook to use eGPU. Even if you use DSDT override, you can uninstall it anytime.

PE4H:-
Support up to x4 link & PCIe 5Gb/s concurrent bandwidth.

PE4L v1.5:-
Support PCIe x1 link only & PCIe 5Gb/s concurrent bandwidth.

PE4L v2.1:-
Support PCIe x1 link only & PCIe rev 2.0 10Gb/s concurrent bandwidth. Sandy Bridge & Ivy Bridge notebooks support PCIe rev 2.0 10Gb/s concurrent bandwidth.

EC2C:-
Referring to ExpressCard.

PM3N:-
Referring to mPCIe card.

Dell Latitude E6410 is an Intel 5 Series chipset notebook (1st gen Intel Core). PCIe on this notebook only support 5Gb/s concurrent bandwidth. So, you can get either PE4L v1.5 or v2.1. Getting PE4L v2.1 is good idea in case you're going to get Sandy Bridge or Ivy Bridge notebook in the future.

For the PSU, look for PSU with at least 20A/21A on +12V & preferably single rail PSU. For example, the Corsair Builder series PSU not expensive & have more than enough amperage on the +12V than your eGPU need. I don't think 570 is overkill if your notebook support Optimus setup. Higher graphic card will definitely provide more FPS, as you can see at first post. There are also several GPU comparison at first post which can help you decide.

 

Not necessarily. If you are prepared to solder two two yellow and two black leads from the 24pin ATX connector on your PSU to a pci-e power lead then you can utilize the second rail. Eg: 75W floppy molex + 75W pci-e (150W from 12V1) and 75W pci-e (75W from 12V2). You'd void the warranty in cutting and soldering those wires.

So you might just want to consider getting a higher power single-rail PSU instead. Any of these will do the job, just depends if you want to use the PSU for other purposes later on and get one with a higher max power output than you need for your GTX570 now.

12V/24A: Newegg.com - Broadway Com Corp OKIA-BLACK-550 550W ATX Power Supply
12V/25A: Newegg.com - LOGISYS Computer PS575XBK 575W ATX12V SLI Ready Power Supply
12V/30A: Newegg.com - Diablotek PHD Series PHD450 450W ATX12V V2.2 Power Supply
12V/34A: Newegg.com - COOLMAX ZX Series ZX-500 500W ATX12V v2.2 / EPS12V v2.91 SLI Ready CrossFire Ready 80 PLUS Certified Active PFC Power Supply
12V/35A: COOLER MASTER GX 450W RS450-ACAAD3-US 450W ATX12V v2.31 80 PLUS BRONZE Certified Active PFC Power Supply
12V/40A: Newegg.com - Diablotek DA Series PSDA600 600W ATX12V v2.2 Power Supply
12V/41A: Newegg.com - Diablotek UL Series PSUL575 575W ATX 12V v2.31 SLI Ready CrossFire Ready Power Supply

 

Hey Nando, do you know if setup 1.x is compatible with Sis chipset or is only intended to work with Intel chipsets?
Because I tried some time ago when it was still freeware, and it always hanged.

 

Hi,
I downloaded the latest nvidia driver "296.10-desktop-win7-winvista-64bit-english-whql"
and I can't find any nvam.inf files
I can see only nvae.inf & nvak.inf

 

The configuration file name is different for desktop drivers than laptop drivers. For desktops it's called something like nvdisp.inf. The main thing though is that the desktop driver will already contain your graphics card ID and will install without any modification.

 

Hi all,
Just to share my experience, it was very smooth

MSI GTX 560 TI 480 core
PE4L 2.1
Dell E6420, 8GB RAM, Intel HD 3000, Win 7 64bit

• Plugged the Express card and turned the power supply on while booting (windows logo)
• Windows detected a VGA card, tried to install the driver, but gave me black screen
• Restarted my laptop while express card in, power ON
• Laptop detected the External monitor, and started using it
• I installed graphics card driver on MSI CD
• Restarted again, and here it is working

I have 2 questions ,
is this the maximum performance I can get from my card ? if not what else I can do ?
How can I make use of Optimus thing?
(I downloaded driver "295.73-notebook-win7-winvista-64bit-international-whql", replaced "nvam.inf", when I run the setup it doesn't detect my nvidia card and doesn't install anything)

 

hit a wall.. So ....I did end up getting code 12 error ,despite my T520 having the latest bios 1.36 (which was supposed to fix tolud issues)

specs of lenovo T520

i7 2860qm cpu
16gb (8gbx2) 1600mhz memory
NVS4200m + intel 3000
windows server 2008r2

I am using the latest PE4L 2.1b, with my new gtx680.

I have set the bios to use just integrated graphics mode (essentially disabling the dgpu nvs4200)

I have tried setting pcie to force gen1 instead of auto.. I have tried disabling all unnessary devices in bios (wlan, memcard,esata,etc) .. I have tried just one stick of mem, so only 8gb).

regardless.. windows will either hang at load, or if I do get through, under device properties I have code 12 on my gtx680.

Why is this still occuring? I dont really want to downgrade to less memory, I could pull out the physical wlan card out if that helps free up resources... could 2008r2 have something to do with this? (I figured this should have been plug and play, err)

do I need that setup software? or should I attempt a DSDT override?

Any ideas would be helpful... thanks guys

I am assuming the DSDT override will fix this definitely, if I have to go that route

 

The Dell E6420 bios will use the eGPU as the primary video device if the bios sees it on bootup. So you either need to hotplug it by halting Win7 load with F8 or try setting the PCI Reset Delay to 7 seconds. You need the iGPU to be the primary video device. Then you'll get x1.2Opt where the eGPU can drive the internal LCD and you'll see pci-e compression improve performance by 20-333%.

The modded drivers' nvam.inf files do not have an entry for the GTX560Ti-448 in it. Need to add it to get the install to go through.

We don't know if Lenovo implemented their dynamic TOLUD on dGPU equipped systems. It's important that the PCI Reset Delay be set to 0secs on the PE4L 2.1 (not 6.9secs) so the bios does see the eGPU on bootup and can adjust TOLUD accordingly. If still no go then follow the Pre-purchase FAQ->2. How can I check if my notebook is compatible with an eGPU? steps to see what TOLUD is when you disable the dGPU and boot with the eGPU active.

If TOLUD isn't being adjusted to 3.0 or 3.25GB, then yes, need another solution to allocate the eGPU and get rid of the error12. That can be either the DSDT override OR enabling the dGPU in the bios (to lower TOLUD) then configure Setup 1.x to disable it and hijaak it's resources for use by the eGPU.

 

thanks for the info (and ya, my pci reset is 0.0s), I will look more into more once off work tomorrow..

its odd, since looks like Tom_ger, who has a similar system, but with only 8gb ram, had no issues with his gtx560.

 

Ok so i went full retard and in my haste i purchased a pe4l 2.1a(a 2.1b came) not realizing i had a pc card slot...not an expresscard slot...... .. .. ..
what i was wondering is can i use pm3n off of the port on the pe4l? or am i just burnt 80 bucks.?

 

Setup 1.x is tested on a Intel platform with some Intel-specific features are disabled when running on an NVidia or SIS chipset. The distribution will still boot and can be used to manually configure the system if needed to overcome error 12. Compact has been updated since the free versions so is less likely to cause system freezes. Still, with the DSDT override procedure you may not even need Setup 1.x.

So youe Dell E5400 has a PC-card rather than expresscard slot?

PE4L 2.1 has a soldered EC2C on the end.. necessary to run at pci-e 2.0 speed. The older PE4L 1.5 uses a mHDMI cable and the end can be swapped for an EC2C (expresscard) or PM3N (mPCIe).

So I'm sure you could find someone here that would swap their PE4L 1.5 for your PE4L 2.1a. You'd be able to get a PM3N and use it and they've get a pci-e 2.0 specced device.

Alternatively, there is a Dell E5400 systemboard on ebay for $65. If it has an expresscard slot then I'd suggest grab that and sell yours. The changeover costs would be lower and you'd get a more convenient expresscard DIY eGPU implementation:

E5400 motherboard, systemboard | eBay

 

Thanks for the quick response man and thanks for the research,i think ill go ahead on the board...except ill use this as my excuse to man up and pll mod my existing board for practice. thanks again

~edit~ actually ill refrain from mod to help recoup funds by selling board

 

so i was over at dells website looking at the pdf for the e5400 and in the expansion card section it says

"One Type I or Type II PCMCIA card; support for 34 mm
Expresscard® via caddy"

i take it this means it was only produced with pc card? meaning 1 gb/s transfer ?

~edit~ or is it some sort of thing resembling a hdd caddy?

 

Please let me know if you solve your issues. I may run into the same as I have a T520 + 2860qm. I'm still in the phase of waiting for my parts (PE4L 2.1 and GTX 560TI)