nando4

DIY eGPU Setup 1.30 (previously Setup 1.x)

2089 posts in this topic

AcqKcth.png
Setup 1.30

 

j3junyB.png KPvnBEd.png 4EoyAFp.png tCFq0Ji.png ID1Nyjh.png XqpQnzj.png uK8H5u6.png bRuy5In.png
 
 
DIY eGPU Setup 1.30 >> US$25: Paypal to 1NG6ScX.png << (Jan-2016)
In msg, pls include your email address and TechInferno username if have one.
Check your INBOX and SPAM folders for a reply message.

Don't have Paypal access but have a credit card? Please email me on 1NG6ScX.png instead.
 


Setup 1.30 now being distributed with many new features: Series-7/8 chipset support, Win8/10 support, polled eGPU detection, Macbook installation, 13" Macbook PCI compaction, 36/64-bit compaction matching DSDT override + more.

 

GqLz4Gx.pngnfsaawS.pngB8D36iA.png mh4X0vn.png c6I96WA.png26PHcef.png

Above: sample screenshots from DIY eGPU Setup 1.20. Hitting F1 over a menuitem providing context-sensitive help

 

   
Above: second video's first 15s shows how to configure a GTX670 eGPU for use on a Acer 5750 using Setup 1.30 (details). Shown is how to configure the mPCIe port, perform PCI allocation of iGPU+eGPU, then chainload to Win7.  

 


Setup 1.30 is available with quite a few nice new features over the previous 1.0x versions. These include using the faster FreeDOS 1.1 platform with no licensing issues, larger disk image with 26MB free space, a html-text based help system, being able to run from Linux' grub2 bootloader, Series-6+ 2.5-5.0 GT/s southbridge PCIe port switching, link retraining, simpler grub4dos setup; additional Reassign buses, ALL port and root port anti-whitelisting method, significantly faster F5 and status refresh, consistent iGPU/dGPU/eGPU naming, mouse support, cache option to speedup startup.bat runtime, press F3 to add the last command into your startup.bat, browse your NTFS/EXT2/EXT3 partitions, do screen capture using the ALT+DEL key, play a retrogame with ALT+G hotkey, provide an uninterrupted disk image boot into your OS and numerous bug fixes. Looks quite a bit better too

 

Why would I need DIY eGPU Setup 1.30? >>>


Setup 1.30 is a FreeDOS environment used to configure your DIY eGPU before chainloading to your OS. Required if you:
 

  • encounter Windows error 12:cannot allocate resources requiring automated PCI Reallocation or IGP relocation. Note: Win7/8 users may want to try a simpler 36-bit root bridge DSDT override instead of using Setup 1.30.
  • want to use x1E/x2E/x2/x4 higher performance pci-e links (x1E/x2E on Series-4 or older, x2/x4 on Series-5 or older chipsets)
  • want to set pci-e 1.0 (2.5GT/s) or pci-e 2.0 (5GT/s) link speed (Series-6 or newer chipset).
  • require mPCIe anti-whitelisting (HP/Lenovo) to allow a wifi slot to work with a PM3N
  • want to disable a dGPU in a hybrid graphics system to free up resources to host the eGPU
  • want to disable a dGPU in a hybrid graphics system to activate NVidia Optimus and if using a x1 link, engage x1 pci-e compression
  • want to initialize a NVidia video card prior to Windows boot to prevent hang/BSOD on startup or error43 in device Manager. Eliminates the need for a standby, attach, resume to overcome this problem.
23 people like this

Share this post


Link to post
Share on other sites

I got a couple of questions.

Would this Setup 1.x allow me to disable white-listing checks imposed by Lenovo on a laptop like Y510? Not talking only about eGPU, for example I want to use another WiFi card instead of the original one, until a whitelisting modded bios emerges.

Also, can I use this program to disable the discrete Nvidia GPU 750M and enable the integrated Intel one? I understand this could be a hardware issue and I could end with nothing displaying on my screen, just asking.

Share this post


Link to post
Share on other sites
I got a couple of questions.

Would this Setup 1.x allow me to disable white-listing checks imposed by Lenovo on a laptop like Y510? Not talking only about eGPU, for example I want to use another WiFi card instead of the original one, until a whitelisting modded bios emerges.

Also, can I use this program to disable the discrete Nvidia GPU 750M and enable the integrated Intel one? I understand this could be a hardware issue and I could end with nothing displaying on my screen, just asking.

If the notebook doesn't halt bootup with the 'unauthorized wifi card detected' then potentially Setup 1.1x could be used to allow another vendor's wifi card to work. That was what was used in [ALL ProBooks/Laptops in General] Our WiFi WhiteListing Days Are 0v3r! .

Setup 1.x can disable/enable the iGPU port on Series-5 or older systems (1st i-core or older) but it is only of value if the notebook has the iGPU wired through. More valuable is disabling the dGPU so an eGPU can use it's pci-e space instead as well as receive NVidia driver priority to enable Optimus features (internal LCD mode, x1 pci-e compression)>

1 person likes this

Share this post


Link to post
Share on other sites

Hey Nando, got the setup in my signature working however I got a few issues/questions;

1) Should I be using grub4dos MBR, straight up "chainload MBR", bootmgr, or? I am currently using grub4dos MBR and it seems to be working but since it is Windows 7 I thought I should be using bootmgr, no?

2) iGPU + eGPU compaction doesn't seem to be working via your software. It says "no solution found" every time. Compact all PCI lanes doesn't seem to work either and just hangs the computer everytime. If I want better DX9 performance, what option should I choose for compaction? Maybe just the eGPU one? The first one (all except x,y, and z) does work but I am leary of compacting PCI lanes not related to my eGPU....GPU-Z currently displays "x1 2.0".

3) It seems I cannot set my PhysX processer to be my iGPU? It only gives me 2 options, CPU and the GTX650. I heard it is a good idea performance wise to atleast use the iGPU as the PhysX processer (is that true first of all?) so I'd like to be able to do that....

Thanks in advance Nando!

2012 13" Macbook Pro | 8GB Gskill RAM | 124GB Crucial M4 SSD | OS X + Windows 7 (MBR) + Linux

GTX650 Ti eGPU | 400w PSU | PE4L v2.1b | Sonnet Pro Adapter | Setup 1.2x

1 person likes this

Share this post


Link to post
Share on other sites
Hey Nando, got the setup in my signature working however I got a few issues/questions;

1) Should I be using grub4dos MBR, straight up "chainload MBR", bootmgr, or? I am currently using grub4dos MBR and it seems to be working but since it is Windows 7 I thought I should be using bootmgr, no?

2) iGPU + eGPU compaction doesn't seem to be working via your software. It says "no solution found" every time. Compact all PCI lanes doesn't seem to work either and just hangs the computer everytime. If I want better DX9 performance, what option should I choose for compaction? Maybe just the eGPU one? The first one (all except x,y, and z) does work but I am leary of compacting PCI lanes not related to my eGPU....GPU-Z currently displays "x1 2.0".

3) It seems I cannot set my PhysX processer to be my iGPU? It only gives me 2 options, CPU and the GTX650. I heard it is a good idea performance wise to atleast use the iGPU as the PhysX processer (is that true first of all?) so I'd like to be able to do that....

Thanks in advance Nando!

'chainload mbr' is fine.. and in most cases works better than 'chainload bootmgr'. Hence why it's set as the default.

The PCI compaction will be dependent on the system and as such the list of combinations to try is shown if you hit F1. If "All except SATA+..." works then stick with it. A lot of BIOS/firmware doesn't pre-set devices on boot and depends on the OS to do it. When there is a huge chunk needed by the eGPU then it can confuse matters. Compact is doing what the firmware didn't do AND adding the eGPU into the allocation.

iGPU can't be used as a Physx processor. Only a NVidia dGPU/eGPU.

Share this post


Link to post
Share on other sites

Anyone who is looking to convert their GPT disk to MBR please see below. I needed to do so in order to properly run this software for my HP 2570P laptop. In order to make this work, you'll need a Live-CD Linux that supports your network card (I used Backtrack 5, it has Intel 6205 drivers), a Windows 7/8 retail installation media (OEM won't work), and a few hours. I highly recommend that you make a system image using Windows Backup prior to starting, as there is a serious chance of messing something up. Spoiler has download links

Backtrack 5:

Windows 7 Install media, choose your language and SP1:

Details on repair install:

If you have a large enough flash drive, at least 8gb, YUMI was super helpful:

In order to reclaim the 100mb EFI partition after everything, I used GParted.

Check your install media by booting into it, selecting your language/currency/time, then click repair in the bottom left. Mine said that I couldn't repair my windows install with that particular disk, but it worked after I got started.

If you get lucky, then you can follow these instructions tailored to your flavor of Linux without any extra hassle:

How to Convert a GPT disk layout to a MS-DOS/MBR layout without data loss (and Gigabyte Hybrid EFI) | firewing1

For me, after step 14, Windows Repair couldn't locate my installation so I had to set my primary partition as active using diskpart from the Windows RE command prompt:

Diskpart

LIST DISK

SELECT DISK (followed by the number of the disk . most likely 1)

LIST PARTITION

SELECT PARTITION (followed by your partition number. most likely 1)

ACTIVE

EXIT

Windows startup recovery should now work.

After that, I started again from step 13 and it worked. Don't forget to change your BIOS to Legacy or MBR boot. Good luck!

1 person likes this

Share this post


Link to post
Share on other sites

Anyone tried bf4 on a egpu setup? I am worried that my 3360m might not be enough.

Thanks.

Phil

Sent from my iPhone using Tapatalk - now Free

Share this post


Link to post
Share on other sites

I dismounted the eGPU image and removed the eGPU folder from C:, but at startup I still have to choose Windows 8.1 or Setup 1.x . How to remove it completely without the boot interruption?

Share this post


Link to post
Share on other sites
Quote

I dismounted the eGPU image and removed the eGPU folder from C:, but at startup I still have to choose Windows 8.1 or Setup 1.x . How to remove it completely without the boot interruption?





Prior to removing the c:eGPU directory you'd run c:eGPUuninstall-disk-image for it to remove the DIY eGPU Setup 1.x bootitem from BCD/boot.ini. If you've deleted it then at the commandline run 'bcdedit', followed by 'bcdedit /delete {ID}'. eg:



C:>bcdedit

...

Real-mode Boot Sector

---------------------

identifier {abadcceb-40c1-11e3-b075-b2a36dbf094f}

device partition=C:

path eGPUgrub.exe

description DIY eGPU Setup 1.30



C:>bcdedit /delete {abadcceb-40c1-11e3-b075-b2a36dbf094f}

The operation completed successfully.

Share this post


Link to post
Share on other sites
I have a problem. When I try to disable the dgpu, it freezes. The keyboard works fine, as the capslock light still responds, but it just sits there at the setpci command and I have to restart the system. Any ideas?

If the dGPU is the primary bootup device, rather than the iGPU, then that is expected. The dGPU should only be disabled if the console is being viewed on an iGPU. The only other instance the dGPU can be disabled is if you are using XP or Win7/8 with XP drivers and are seeing an error 12 on your eGPU AND have insufficient 32-bit PCI config space to perform PCI compaction. There would need to ignore[dGPU] to get a PCI compaction solution. Then you can add a 'call iport dGPU off' to your startup.bat and boot blindly into Windows.

If your configuration differs from the above scenario and are using the latest 1.30 version then please email me to discuss further.

Share this post


Link to post
Share on other sites

Here ya go:

Almost sleek narration to make it seem like I have some remote idea of what I'm talking about lol.

---

As requested, showing minimal interaction and a fast bootup:

2 people like this

Share this post


Link to post
Share on other sites

Hey nando,

i just want to to ask if you has received my payment. i would like to send you a pn, but with less then 5 posts it´s a bit difficult :). i´m posting here because i´m just wondering haven´t received an answer since my payment on 5th. january. if you have answered me, i´m sorry but i didn´t get anything :(. please excuse my lag of englisch, but school belongs to the past and reading is easier then writing :) . hope to hear from you soon.

best regards

jan

Share this post


Link to post
Share on other sites
You could get one of these for $200.

Firmtek TT-PX ThunderTek/PX Thunderbolt to PCIe... in stock at OWC

Should be able to pull out the useful bit in a few minutes. Downside: only 10Gbps.

As posted by Relentless, there is a $200 10Gbps TB1 enclosure ( FirmTek ThunderTek/PX) and a $480 16Gbps TB2 enclosure (OWC Helios 2 - rebadged Sonnet Echo Express SE II with 3yr rather than 1yr warranty) : http://forum.techinferno.com/diy-e-gpu-projects/2109-diy-egpu-experiences-%5Bversion-2-0%5D-259.html#post80870 . Only the Haswell 2013 13" rMBP and 15" MBP have TB2 controllers. All other Macbooks have TB1 controllers.
Yeah I have a 2012, and I'm pretty sure a GPU doesn't come close to 10Gbps so this would be perfect
Does the III-R Mobile Rack kit include PCIe to TB? Doesn't look like it does, but says this:

Note!PCIe cards sold separately; requires Thunderbolt cable

It was listed under sonnets site as a PCIe-TB thing.
The $199 III-R Mobile Rack kit is an extra that can be added to a $1075 Echo Express III-R. You can see how it all pieces together at Sonnet - Echo Express III-R Thunderbolt 2 PCIe Expansion Chassis . Yes, the Echo Express III-R should be $199 instead!! Sonnet obviously targetting the premium end of the market.

Cheapest so far being the $199 FirmTek ThunderTek/PX. At that price it doesn't make any sense to purchase a US$92 PE4L 2.1b + US$132 Sonnet Echo Expresscard II adapter anymore since it (1) costs more overall and (2) is a 5Gbps rather than 10Gbps solution.

Warning - a expresscard eGPU adapter vendor is pushing their x1 2.0 5Gbps solution instead of native 10/16Gbps ones

I've been alerted to a competing vendor that is claiming 5Gbps + Optimus x1 pci-e compression is faster than native 10/16Gbps Thunderbolt1/2. This is not true. 10Gbps TB1 is at least 14% faster: http://forum.techinferno.com/diy-e-gpu-projects/3062-%5Bguide%5D-2012-13-mbp-gtx660ti-hd7870%40x2-2-th05.html Besides, the 10Gbps native TB1 solution can have the eGPU second lane taped so can run at 5Gbps (x1) if really want to in those rare DX9 instances that may see performance benefit in doing so.

Same vendor is also claiming that Optimus internal LCD mode only works in x1 mode. Again, misinformation disproven by same link : http://forum.techinferno.com/diy-e-gpu-projects/3062-%5Bguide%5D-2012-13-mbp-gtx660ti-hd7870%40x2-2-th05.html .

This illustrates misleading pre-sales info that favors their x1 2.0 expresscard solution using a EC-to-TB adapter for Thunderbolt systems above native Thunderbolt ones for completely false reasons. Buyer beware!!

Ya lost me there.... but I agree all he PCIe-TB solutions are marketed to business and professionals with a way inflated price, and every link I could find googling (MAc external gpu) or (PCIe to tunderbolt) ect. just turned up the PE$L + echo expresscard link from this site.

Share this post


Link to post
Share on other sites

Hi @Nando.

The Firmtek seems to put out about 36 Watt (specs say 12V at 3.0A) so if you were to power something like a Geforce GTX 770 which needs 230W TDP (for example) and has a 6 pin (75W) and 8 pin (150W) power connector you would have 225W from the power supply and any extra (5W) from the Firmtek?

Is there a minimum power (or percentage of power) that must come from the PCIe connector? or that the graphics card expects to draw from the PCIe connector?

If not then could you just:

1. buy a Firmtek and rip out it's logic board.

2. mount a Geforce card in a chassis with a standard power supply capable of delivering 225W+

3. connect the Firmtek board in place of the motherboard. I am guessing that it will hold on to the PCIe slot well enough without proper mounting.

4. UEFI boot Windows on a Macbook Pro with both the power supplies powered on.

Cheers

Share this post


Link to post
Share on other sites
Hi @Nando.

The Firmtek seems to put out about 36 Watt (specs say 12V at 3.0A) so if you were to power something like a Geforce GTX 770 which needs 230W TDP (for example) and has a 6 pin (75W) and 8 pin (150W) power connector you would have 225W from the power supply and any extra (5W) from the Firmtek?

Is there a minimum power (or percentage of power) that must come from the PCIe connector? or that the graphics card expects to draw from the PCIe connector?

If not then could you just:

1. buy a Firmtek and rip out it's logic board.

2. mount a Geforce card in a chassis with a standard power supply capable of delivering 225W+

3. connect the Firmtek board in place of the motherboard. I am guessing that it will hold on to the PCIe slot well enough without proper mounting.

4. UEFI boot Windows on a Macbook Pro with both the power supplies powered on.

Cheers

Huh? don't you just connect it via the TB port?

This is the only thing I have found with a decent power supply but its not TB, I'm just going to find a small chassis and throw a 500w PSU in there.

I would run a 780 but I heard they don't work on mac for some reason.

- - - Updated - - -

Hi @Nando.

The Firmtek seems to put out about 36 Watt (specs say 12V at 3.0A) so if you were to power something like a Geforce GTX 770 which needs 230W TDP (for example) and has a 6 pin (75W) and 8 pin (150W) power connector you would have 225W from the power supply and any extra (5W) from the Firmtek?

Is there a minimum power (or percentage of power) that must come from the PCIe connector? or that the graphics card expects to draw from the PCIe connector?

If not then could you just:

1. buy a Firmtek and rip out it's logic board.

2. mount a Geforce card in a chassis with a standard power supply capable of delivering 225W+

3. connect the Firmtek board in place of the motherboard. I am guessing that it will hold on to the PCIe slot well enough without proper mounting.

4. UEFI boot Windows on a Macbook Pro with both the power supplies powered on.

Cheers

Huh? don't you just connect it via the TB port?

This is the only thing I have found with a decent power supply but its not TB, I'm just going to find a small chassis and throw a 500w PSU in there.

I would run a 780 but I heard they don't work on mac for some reason.

Share this post


Link to post
Share on other sites
Hi @Nando.

The Firmtek seems to put out about 36 Watt (specs say 12V at 3.0A) so if you were to power something like a Geforce GTX 770 which needs 230W TDP (for example) and has a 6 pin (75W) and 8 pin (150W) power connector you would have 225W from the power supply and any extra (5W) from the Firmtek?

Is there a minimum power (or percentage of power) that must come from the PCIe connector? or that the graphics card expects to draw from the PCIe connector?

If not then could you just:

1. buy a Firmtek and rip out it's logic board.

2. mount a Geforce card in a chassis with a standard power supply capable of delivering 225W+

3. connect the Firmtek board in place of the motherboard. I am guessing that it will hold on to the PCIe slot well enough without proper mounting.

4. UEFI boot Windows on a Macbook Pro with both the power supplies powered on.

Cheers

Easiest way where no mods to the Firmtek are made would be to plug one of these into the Thundertek/PX slot : PCI Express PCIe 16x Riser Card Ribbon Extend Cable Molex Power Capacity Filter | eBay

$_3.JPG

That's a x16 to x16 pcie extender with molex power. The molex power providing the 12V/75W necessary to drive the slot. That eliminates using the supplied and insufficient 12V/36W AC adapter. It *might* be necessary to shorten the extender from x16 to x4. I say might because I can't quite see if the Thundertek/PX's slot is an open ended x4 one or not and unfortunately ebay doesn't seem to have a x16 to x4 extender. If the slot isn't open ended, then would need to cut down the end that plugs into the Thundertek/PX to be x4 OR you could cut into the Thundertek/PX slot to allow x16 cards to be inserted.

Once the pcie extender has beem added just attach a ATX PSU molex to that connector, attach pci-e power to your video card and that's it for a UEFI implementation. A bootcamp/MBR implementation would need an additional PCI Reset Delay mod added.

Share this post


Link to post
Share on other sites
Easiest way where no mods to the Firmtek are made would be to plug one of these into the Thundertek/PX slot : PCI Express PCIe 16x Riser Card Ribbon Extend Cable Molex Power Capacity Filter | eBay

That's a x16 to x16 pcie extender with molex power. The molex power providing the 12V/75W necessary to drive the slot. That eliminates using the supplied and insufficient 12V/36W AC adapter. It *might* be necessary to shorten the extender from x16 to x4. I say might because I can't quite see if the Thundertek/PX's slot is an open ended x4 one or not and unfortunately ebay doesn't seem to have a x16 to x4 extender. If the slot isn't open ended, then would need to cut down the end that plugs into the Thundertek/PX to be x4 OR you could cut into the Thundertek/PX slot to allow x16 cards to be inserted.

Once the pcie extender has beem added just attach a ATX PSU molex to that connector, attach pci-e power to your video card and that's it for a UEFI implementation. A bootcamp/MBR implementation would need an additional PCI Reset Delay mod added.

I am soooooo lost... I want to do bootcamp so what is a PCI Reset Delay mod? Is using s Helios way easier or about the same difficulty?

Share this post


Link to post
Share on other sites

Helios is the same board. Just a different brand name outside, hardware is the same.

You need to cut the wires from the extender if you power the videocard from the PSU, because you don't want the current to flow back in the thunderbolt daughterboard.

Share this post


Link to post
Share on other sites

Hey Guys,

I am running a GTX660 in a Sonnet Express Pro and almost everything is running out of the box.

All I had to do was edit two kext files: NVDAStartup.kext and IONDRVSupport.kext to enabler Thunderbolt support.

This gives me native support in OSX 10.9 and any game I have tried runs as smooth as it gets.

I have however encountered two minor issues and was hoping somebody here can help me solve:

1. Clamshell mode is not working properly anymore. The MB just sends itself to sleep all the time. I am currently circumventing the problem by running NoSleep! in the background.

2. Hardware acceleration when playing HD vides is all screwed up. Just greenish stripes all over, nothing really decipherable.

Share this post


Link to post
Share on other sites

Hey,

From what I've heard, it's an issue with Mavericks. I was running a Mac Mini with a 660 with 10.8.3 and everything worked fine, but upgrading to 10.9 caused some hardware acceleration issues.

Here's an article on the TonyMacx86 site that has some people with similar problems in mavericks: Nvidia Hardware Acceleration Issue After Mavericks

Share this post


Link to post
Share on other sites

Are you fully updated to 10.9.1 ? 10.9.0 was terrible for me when using a external card and with no external card connected.. Update and you should be fine... 10.9.2 is just around the corner also...

Share this post


Link to post
Share on other sites
Are you fully updated to 10.9.1 ? 10.9.0 was terrible for me when using a external card and with no external card connected.. Update and you should be fine... 10.9.2 is just around the corner also...

Yes newest version of everything. I'll try installing rEFInd on the week end and try to boot my MBP as a "Mac Pro" which allegedly solves these issues.

Share this post


Link to post
Share on other sites

Be careful with sn issues when using refind, I had to reactivate all my adobe software the last time i used that...

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now

  • Similar Content

    • By nando4
      [Originally posted 3-14-2012 in DIY eGPU experiences (NBR). Relocated to T|I due to being banned by Lenovo fan ZaZ].
      Due to it's importance created a new official updated thread.

      no more 4GB+ system RAM issues: use a DSDT override to extend the root bridge into 36-bit space (Windows 7/8/10 only - MacOS/Linux are unaffected)
       
      Introduction>>>
      This problem and solution is only relevant to Windows operating systems. MacOS ignores the root bridge and can allocate in 64-bit space as kizwan found here. Linux has a "noCRS" kernel parameter to ignore the root bridge boundaries so the OS can also also allocate in full 64-bit pci-e space.

      The this process aims to solve the problem of seeing an error 12 (This device cannot find enough free resources that it can use) issues against a eGPU in Windows Device Manager due to insufficient 32-bit addressing space. By manually adding a memory range outside of this 32-bit space, we can force Windows to operate in 36-bit space instead to host eGPUs.

      Know systems that will require this process are Dell Latitude Ex410 series, Samsung Series-4, Sony F-Series and MSI CRx20, HP-Elitebook/Probook xx60x/xx70x with AMD GPUs. Those have insufficient free 32-bit PCI space to accomodate a eGPU if have 4GB or more of RAM installed. They can't use Setup 1.30's PCI compaction to create sufficient space because they either have TOLUD=3.5GB, have lower TOLUD like 3.25GB but with an unmovable systemboard device occupying candidate free pci space making it unusable or are using an AMD or GTX650/GTX750 card that requires over 256MB contiguous space. Refer to Pre-Purchase FAQ #2(>> 2. How can I check if my notebook is compatible with an eGPU?) for details on how to check your TOLUD.

      The same solution can be applied to any 3-gens old PM965 chipsets (2007) or newer system. They have a 64-bit CPU and chipset so are fully PAE-36-bit/64-bit compatible. Dell Inspiron 1440/1525 (core2duo) and Dell Vostro 1015 (core2duo) have BIOS-configured 36-bit root bridges already so don't require this workaround.

      I happened to run into one of these problematic systems. A Dell E4310 with TOLUD=3.5G which can be maxxed out at 16GB of RAM. It's too nice an ultraportable to be need to downgrade RAM to 3GB of RAM to use an eGPU. I set of in pursuit of how to add a DIY eGPU to it which is documented below.

      Using Setup 1.30's 36-bit PCI compaction on the eGPU to relocate it in such space will see the eGPU still give an error 12 in Device Manager. Why is that? Avlan confirmed that the 32-bit root bridge definition limited allocation only to 32-bit space. The fix being a modified DSDT loaded as a registry override that extends the root bridge (ACPI PNP0A08 or PNP0A03 device) into 36-bit space. Avlan's fix was cryptic so wasn't attempted on other systems.

      After a bit of digging from the following references I found the QWORDMemory DSDT static resource entry can be used to easily extend root bridge entry into 36-bit space. Ref: [Qemu-devel] [PATCH 6/6] 64bit PCI range in _CRS table plus two examples at virtualbox and Coreboot.
       
      Tools required>>>
      The iasl and asl tools used are in 2570P-dsdt-override or (older) DELL-36b-root-bridge.rar. Allternatively, can download them from their original sources:
       
      The latest Windows Binary Tools (WBT): Download
      The Windows Driver Kit (WDK), which contains the Windows ASL Compiler: Download.
      Notepad++ (or any other text editor)
       
      Step-by-step DSDT override>>>
      Need a 36-bit root bridge DSDT override

      The test Dell E4310's root bridge was extended into 36-bit space with the steps below. See also http://forum.techinferno.com/diy-e-g...html#post49833 , user angerthosenear's alternative instructions to those presented below where can use a DSDT Editor to simply the patching process if experience a compilation error:
      iasl -g acpidump -b 3.Decompile dsdt.dat to get an output dsdt.dsl text file that can be editted:
      iasl dsdt.dat 4. Open the resultant dsdt.dsl file and look for the PNP0A08/PNP0A03 'DWordMemory' resource entries. Under the last DWordMemory entry in that area, add a 'QWordMemory' (64-bit) entry  shown as the second paragraph below. I ensure that I stay in the 36-bit range (< 64GB) so the location is compatible with the PAE-capable Win7 32-bit. I also chose a location above 48GB so we won't have issues with maxxed out memory. The range chosen was 48.5GB to 56.25GB.
      DWordMemory (ResourceProducer, PosDecode, MinFixed, MaxFixed, Cacheable, ReadWrite, 0x00000000, // Granularity 0x000A0000, // Range Minimum 0x000BFFFF, // Range Maximum 0x00000000, // Translation Offset 0x00020000, // Length ,, , AddressRangeMemory, TypeStatic) QWordMemory (ResourceProducer, PosDecode, MinFixed, MaxFixed, Cacheable, ReadWrite, 0x0000000000000000, // Granularity 0x0000000C20000000, // Range Minimum, set it to 48.5GB 0x0000000E0FFFFFFF, // Range Maximum, set it to 56.25GB 0x0000000000000000, // Translation Offset 0x00000001F0000000, // Length calculated by Range Max - Range Min. ,, , AddressRangeMemory, TypeStatic)
      5. Create a dsdt.aml file with the 'iasl -oa dsdt.dsl' compiler directive. '-oa' disables all optimizations found necessary to get working solutions. It is unlikely this will succeed first time, as the compiler is very strict. In general warnings are acceptable, but errors may require some fiddling. Those will need to be corrected with common patches in Common DSDT patches for Ivy/Sandy/Haswell laptops, How to fixed DSDT errors? (kizwan). Consider asking for guidance at tonymacx86.com DSDT forums. DSDT editting alone could have it's own forum. Then try compiling your DSDT again.
       
      iasl -oa dsdt.dsl 6. Load the dsdt.aml file into into the registry as a DSDT override. You must be an admin user for this to be successful.
      asl /loadtable dsdt.aml 7. [Windows 8.x/10] Enable test signing mode for the registry override to apply. You must be an admin user for this to be successful.
      bcdedit -set TESTSIGNING ON An alternative to disabling testsigning in Win8/10 is to do a Win8: Perform a DSDT substitution within Setup 1.30 to overcome TOLUD issues. Win7 users who don't want to do the registry override can also do this DSDT substitution.

      8. Check you now have a new Large Memory entry in Device Manager->Resources by Connection->Memory as shown below:
       
       
      9. The same fix was done on a Dell E4300 and Win7 did automatically allocate the eGPU into the 36-bit space. If you still see error 12 then there are two options to pursue:  
       
      Re plug'n'play the devices. Go into Device Manager -> Video cards, delete the eGPU. Also go into Device Manager -> System devices and delete the PCI Express Root Port x entries.

      Upon restarting the system, halt Win7 startup with F8, attach and power on your eGPU and then proceed to load Win7. Win7 should now be clever enough to re-allocate the eGPU into 36-bit space upon realizing there is insufficient 32-bit space to host it.

      Can go one step further and delete your eGPU NVidia/ATiAMD driver, restart the system and reload it. That may help as well.
        Hard allocate the eGPU into 36-bit PCI space before booting Win7 using Setup 1.30. Select PCI compaction->Endpoint=56.25GB (36-bit) and then select Run compact. When prompted for the scope select eGPU.

      The result will be similar to that shown in the spoiler in step 8 above, but with the eGPU rather then the Intel HD iGPU being in 36-bit PCI space. Can then proceed to automate this by editting your %DRV%:\config\startup.bat (or V:\config\startup.bat in windows) so can just select the Automated startup via startup.bat Setup 1.30 bootitem and have it do everything for you: call speedup lbacache call vidwait 300 call vidinit -d %eGPU% call pci call chainload mbr  
       
      Testing results>>>
      * Lenovo X230 (jay_dark_dry): SUCCESS! Allocated the eGPU into 36-bit space.

      * Lenovo T430 (arteart) : SUCCESS! Allocated the eGPU into 36-bit space.

      * HP 2570P (nando4): SUCCESS! Allocated the eGPU into 36-bit space.

      * Dell E4310 (nando4): SUCCESS! I relocated the Intel HD iGPU into 36-bit space using Setup 1.30 as shown in step 8 above freeing 256MB of contiguous 32-bit space to host an eGPU, so it will definitely work. Plus there is plenty of 36-bit space for the eGPU too. Unfortunately my PE4L 2.1 isn't working so I can't show the iGPU+eGPU both being active but it will work

      * Dell Studio 1557 (Kizwan): SUCCESS! His HD5870 successfully relocated to 36-bit pci-e space.
       
       
      FAQ>>>
      1. When do I need to hotplug my eGPU to overcome the error12 with a DSDT override?(kizwan)
       
       

      2. How to clean the DSDT override properly? (kizwan)

      (*We will need to delete DSDT key in registry)
      HKEY_LOCAL_MACHINE\SYSTEM\ControlSetXXX\services\ACPI\Parameters\DSDT * where XXX are 001, 002, 003 & so on. and HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\ACPI\Parameters\DSDT I didn't really tested whether deleting "DSDT" key in ControlSetXXX (XXX are 001, 002, 003 & so on) really necessary but I will test this when I test your latest Setup 1.30. For sure deleting "DSDT" key in CurrentControlSet is necessary. You can access registry via command line. So, you can put it in .bat file.