DIY eGPU experiences

ITM adapter works in ports #2 and #4 normal and detected in AIDA64 .
If I connect wi-fi adapter in any slot - notebook will not turn on (only black screen, but the LEDs are lit)

 

Your test confirms that port1 (expresscard) + port2 (mPCIe) do work, so you can do a x2 (NVidia/ATiAMD) or x2E (ATiAMD) link. X2E will however knock out your wifi in port4 from being concurrently functional. Please refer to benchmarks on first page to see if you prefer x2 NVidia or x2E ATiAMD. The latter will give higher DX9 3dmark06 results. I would suggest also looking at overclocking the system's pci-e bus to squeeze more performance out of it.

 

Hi, i'm new to this forum and very interested by eGPU. So i'm looking for a 11 or 13" laptop to fit with my desktop gtx280 but it looks like its not usable with optimus drivers (too old ?).
What do you think about the dell vostro v131 ? I can't find any laptop with an expresscard port which is strange as i can see in your 1st page 3dmark benchmarks that almost all of you use EC (expresscard right?) on recent laptops. So what do I miss there?

Thanks ! (sorry for my bad english )

 

Click the Thin-and-light Sandy Bridge iGPU, dGPU and eGPU notebooks link in my sig.

 

I have a question regarding slow movements in RPG games in an eGPU setup.

I have an e6520 (i7) hooked up with a GTX 560 ti in an eGPU setup.

I can play pretty much any games including Battlefield 3, the latest COD, Crysis 2, etc without much problem. However, whenever I play a RPG like the Last Remnant or recently Skyrim, the character's slow movement is unbearable. The first prominent thing I notice is the mismatch between the speech and lips. Mouth is keep moving very slowly after the speech is done. In the Last Remnant the character walks very slowly. I lowered all the resolutions and quality, and it is still the same. I checked my 3D setting in Nvidia control panel and I see Auto-select as it should be. Physx as well. I checked the FPS with Fraps, and the FPS is well above the norm especially in low resolution and low quality. However, the games are still slow. I compared their movements with the Youtube videos, and clearly there's something wrong.

Can anyone give me some tips on how to fix this issue?

 

If it's a pci-e bandwidth issue then:

- wait for the pci-e 2.0 compliant PE4L/PE4H to be released.

- liberate some bandwidth: run lower color depth (16-bit), disable NVidia audio devices in Device Manager.

- force DX9, DX10 or DX11 via the commandline or other settings then compare if performance has improved.

- try newer drivers in case they are the problem

- run with 'highest performance' power profile and ensure pci-e link management is disabled.

- disable wifi. Seems some folks are noting what appears to be interference between wifi x1 pci-e port and the expresscard slot.

<strike>Also, if you have a i7-quad + GTX560Ti, can you post benchmark results (3dmark06, vantage, RE5, dmcv4, etc) to be included on the first page? Khenglish's result is the highest i7-quad we have but that's with a GTX460.</strike>

 

That show both mPCIe ports are working. You can use PCIe port #2 & ExpressCard slot for eGPU. Sorry about your WIFI card, it shouldn't happen like that.

 

Thanks Nando. I tried everything you said and one hit it. I got the latest verde driver with the modified nvam.inf and now it's all good. Thank you again.

My benchmark results are in your list right below Khenglish although I haven't done RE5 and dmcv4. I'll try to do that whenever I get the chance. By the way, mine isn't Quad. It's simply i7-2620m.

Anyways, thanks!

 

I would like to use the GTX560 Ti, so I can not get x2E-link ...
Actually, I'll be able to test 9600GT, GTX460-OC and HD6870, before buying a GTX560-Ti Where can I find how to make x2E-link for the HD6870 and then return as x2-link for NVidia cards?
Now I wait when PE4H will be delivered to me (Russian post works very slowly).
How I can overclock the system's pci-e bus? As far as I know, PLL is locked for 5920G (I was not able to increase fsb frequency with SetFSB)...

Yes, it's sad. But I don't regret, because I don't use wi-fi on this notebook.

 

So I am having some problems with my setup, I am using the following:

HP 4420s, Intel Integrated HD Graphics, GTX275, 4GB RAM, Windows 7 64bit

PE4H with ExpressCard

At first I had code 12 from my GTX275 so I began troubleshooting.

I have been using Setup 1.x and have been attempting to compact. I am getting no solution for 32 bit and 32A bit while setting it to compact all.

36 bit successfully finds a solution when compacting iGPU and eGUP without forcing either to 32bit. When I get into windows 7 I find that the eGPU no longer has code 12, but now my iGPU has a code 12. In addition the external screen is still not working, nvidia control panel will not open (says no screen exists).

I also attempted 36bit compaction forcing iGPU to 32bit. When I do this windows will not boot past the screen where the windows logo appears. Hangs before login screen. I tried keeping the eGPU unplugged for the boot, but it still hangs.

Can anyone provide me any assistance? Thanks!

 

Physical connection for x2 & x2E are the same:-
- Connect PE4H port x1 (using miniHDMI cable) to ExpressCard (PCIe port#1)
- Connect PE4H port x2 (using miniHDMI cable) to PM3N mPCIe card at PCIe port#2

Then, just need to use Setup 1.x to set x2 or x2E at PCIe port#1.

 

Hi guys!

I have extended my Lenovo X220 (i5-2540 8GB) with the PE4H+EC2C with a Nvidia Quadro 600 graphics card for CAD design like Autodesk Inventor 2011.



I've done a few quick tests in order to compare the performance between the external gpu and the internal intel hd3000 graphics.

3Dmark06 quadro600: 6700 pts
3Dmark06 intel hd3000: 4800 pts

Specviewperf shows a 10 times performance increase which is to be expected using a CAD optimized graphics card. Now I do have a problem using Inventor 2011:



Inventor switches to the internal Intel HD3000 graphics card! So I won't benefit from the external egpu at all... SPecviewperf, Cinebench and the like do utilize the quadro card but Autodek Inventor doesnt seem to recognize the external Quadro gpu! Is there a way to solve this problem?

I've installed the "265.70_desktop_win7_winvista_64bit_english.exe" driver from the Nvidia website.

 

Is this running on internal screen? Try verde (mobility) driver instead. The link to the driver is available at first post (modded drivers).

 

It's running on my external HP zr24w monitor. I have tried the Verde drivers but I receive this error message the driver is not suited for my graphics card (quadro line).

I have also tried to disable the intel graphics in the Windows Device Manager but after a reboot the external GPU isn;t recognized anymore so this doesnt help either.

 

Will people who run the optimus (1.x?) way still benefit from getting the new pcie2.0 and get even more performance than people running amd cards?
Just if I will benefit from getting the new pcie 2.0 when it comes instead of my standard if I get a GTX560 or that a HD6870 will perform equally well when we switch to 2.0

 

Strange. If using external monitor, it should use eGPU because external monitor is connected to eGPU. Sorry, I have no idea why it use Intel HD 3000 instead.

 

Hello everyone

I have an eGPU setup up and running, however, I want to enable audio over HDMI (which is currently disabled, probably by the driver). I know there is a performance hit but I need it as my HDMI out goes to my receiver which needs the audio input.

How do I get re-enable HDMI audio and get it back?

I have a GTX 570 on an HP Elitebook 2540p Win 7 32bit system. Thanks!

 

Nvidia Control Panel -> Resolution -> HDMI - HDVT (Audio Enabled)
If the options not there you better go check in device manager that NVIDIA HIGH DEFINITION AUDIO exists under sound devices. If not, time to reinstall drivers.

 

Thanks but not much luck. I don't have the nVidia HD audio in the Device Manager. I just downloaded and installed the latest drivers 285.62, but no luck again even after re-scanning hardware. I have attached a picture with a few screenshots.

I know I did something to disable HDMI audio a long time ago when I set up the eGPU for the first time, but can't remember what it was . Using a different nvam.inf or something similar (which shouldn't make a difference now that I installed the latest drivers).

Any ideas?

 

Reinstall 285.62 and choose the clean install option and see if it loads HDMI Audio drivers this time during setup.

 

OK, I had to use the 270.61 drivers because I have a 32 bit OS and cannot find a 32 bit nvam.inf except for this one here

Now I managed to reinstall the 270.61 driver, but I still do not see the audio device in my device manager. Screenshot attached. Also I can't get the eGPU to work. I used to do an Optimus setup and hotplug previously, now hotplugging doesn't work at all. I hotplug and nothing happens, when previously the display would switch to my external monitor.

Just as a note, I did a clean re-install of the nVidia drivers, and the installation asked me to restart in the middle of the process (after it removed the old drivers). After restarting, Windows as usual attempted to do its own default driver installation at startup - and it was installing an HD audio device! However, the nVidia installer asked me to cancel that, else it would not continue with its installation, so I had to cancel it. And I still don't have the HD audio device in my device manager.

Looks like I'm back one step and have to get the eGPU setup working again. Any suggestions on what I should be doing? Is it worth trying the Setup 1.x?

Thanks in advance guys.

 

I have been troubleshooting my problem further and I think I have found why I cannot do the 32 or 32A compaction.

It seems there is a 6MB Motherboard Resource in my D0000000 pci bus memory block. Am I correct that this is not being moved during compaction? If this is the problem how do I fix it?

Attached is a photo of my memory from device manager.

Thanks

 

Hey avlan, first of all thanks a lot for providing these hack for the E4300 users!

I am currently trying to work this done, could you maybe give a short instruction how to exactly do this? I am not sure which steps and how I have to do them (just tried MS ASL Compiler with your .aml and it reports some errors...)


EDIT: Got it! Just did a mistake using "asl /loadtable dsdt.aml".

Now I am running Dell Latitude E4300 with 4 GB Ram (thx to avlan!) with GTX 560.

Drivers: 285.62-desktop-win7-winvista-32bit-international-whql - directly from Nvidia no modded ones (should I?). I did not need to use Setup 1.x.

Benchmarks:

3D Mark 2006: 10.500
3D Mark 2011: 2.500

 

Update 1: I got my eGPU setup to work again, by hotplugging it. It just randomly worked. Still no HDMI audio though, so I'm back to square one. After doing more research on the forums, I think my no-audio-through-HDMI problem is related to this which is in nando4's FAQ link Looks like I'll have to give Setup 1.x a shot now even though I don't have error12 or any other reason to use it. Hopefully it won't mess up my existing setup

Update 2: I had difficulties running Setup 1.0 and while brainstorming, I realized a possible option would be to route my audio directly to my receiver via optical and HDMI to my TV (previously I was trying to do audio and video through HDMI to receiver). This setup worked but only with PCM 2.0 audio output. Bought a new USB Dolby Digital 5.1 card for $40 which does real time 5.1 encoding (through DD Live) and I now have 5.1 sound from my games and x.1 Opt video performance on my laptop

carage, thanks for your support!

 

I know it's not the completely right place to post it, but i preferred to speak here than opening a new thread about this suggestion.
I'm planning to change my notebook, and maybe get a Macbook Pro 13 (late 2011), but before to spend the cash, I would like to have some impressions about my thoughts.

I actually own an XPS L501X (in signature), but the GPU card is starting to limit my game experience even at 720p (GT420m ocked @ GT540m), so I started thinking about an upgrade that could grant me mobility and power (I don't like desktop replacement and 15" is the maximum size i would consider).
My main purposes are mid-high gaming while @ home and maximized mobility while i'm on travel. I've never played on battery and i'm not planning to.

ViDock is actually prototyping the Thunderbolt eGPU solution and the only "affordable" notebooks that today could take advantage of this technology are the Macbook Air and the Macbook Pros.
From what I've read there are 3 TB controllers actually, with the cheapest one (Cactus Ridge, with only 1 channel) that could be implemented in the firsts ultrabooks made by Asus and Acer in Q1 2012; those ultrabooks are going to match the price of the Air probably.

The TB controller built in the Air has 2 channels (Eagle Ridge), while the one in the Macbook Pro has 4 channel (Light Ridge so 4x10Gbps in each direction aka 80Gbps I think). In both the cases the solution is better than the express card available today or mPCI-e). Honestly i would like to get a "commercial" solution, just because i tried to read the posts here but i would like to get an enclosure and support due my lack of time and skill

Furthermore VillageTronic said that with the new revision of ViDock we will be able to use the internal display on the notebook on a single cable and without any hassle (Thanks to the TB cable itself I guess).

Now, considering the CPU in the Air (a bit less powerful than my i3), I'm moving to the 13 inches. Considering that the ViDock preorders are opening during early December (from what i saw on the FB official page), and considering that the price of the Macbook won't drop till the new release (that I could think will come in Mid to Late 2012 with Ivy Bridge release in March '12), do you think could be a good solution to get a Macbook Pro 13 (i5 SB base model), bootcamp W7 for gaming and get the dock once it's available with a GTX6xx Equivalent to a GTX560Ti? The whole config would cost about 1500€ (1150€ MB + 150€ GPU + 200€ DOCK) and in Italy, with the same budget, i could buy only a 17" with a GTX560m (half the power of the Desktop Ti) and without the possibility to upgrade the GPU in the future.

Thanks in advance!

 

Hi Lichtenstein!
I have a 1310 with "Geforce 8400M GS" and as you may know, it has alot of problems with VGA!
I saw your post about DIY ViDock and surprised with your results!
But I have some questions. hope you kindly answer them:
1-why did you use "PE4L" instead of "PE4H" !? is there any problem using it ignoring higher price!?
2- can I have higher performance with "PE4H" or not?
3-considering the config difference between my laptop and yours and the fact that I'm not a very advanced user, do you think this DIY works for me!?
Thank you already!

 

Hi kazem.gh,

There is NO difference in performance whether you use PE4L or PE4H. Both deliver the same job. PE4H is more suitable for non-Optimus setup, e.g. x2 or x4 link setup. It also have sturdier base. PE4L is suitable for Optimus (x1.Opt) setup because this setup only need one PCIe x1 link. eGPU setup performance comparison: x1.Opt >= x2 > x1.

If you have Dell Vostro 1310, you can do x2 link setup using PCIe port#3 (Robson/Intel Turbo Memory) & port#4 (WIFI) (both are mPCIe ports on this notebook according to schematic). You also can do x1 link using ExpressCard (PCIe port#2) but performance will not good. You'll experience stuttering & lagging when too many effects.

 

Thank you Kizwan!
Unfortunately I don't know about "PCIe port#3 and #4" !!.
-Are they something within ExpressCard ports as well ? Can I use them without being advanced user?
-It seems that I can't use x1.opt option. is this correct ?
-Considering that I need these only running 3D Max on my laptop, which product is more suitable and easy to setup?
-I have Adapter (5v & 12v out) belonging to my DIY External CD ROM. can I use it to run these product instead of a complete PC power supply?

Excuse me because of my extremely low tech questions! :")
Thank you already!

 

I played with my laptop further, and removed 2GB of memory. At this point I got the eGPU to work. Unfortunately this leaves me with only 2GB of memory, and I would much rather use my full 4GB!

With 4GB of memory my TOLUD is 3.0GB, so theoretically I should be able compact correct?

Here is my post with the screenshot of my memory!
http://forum.notebookreview.com/gam...851-diy-egpu-experiences-678.html#post8087178

Thanks guys!

 

Yes. It does look like a motherboard resources prevents you from having a full 256MB of free pci-e space to work with. We see in the sample GTX280 lspci output below that it requires a 256MB+32MB+16MB bar, so this could not be accomodated on your system. A GTX275 would be the same.

Code:

02:00.0 VGA compatible controller [0300]: nVidia Corporation GT200 [GeForce GTX 280] [10de:05e1] (rev a1) (prog-if 00 [VGA controller])
        Subsystem: Device [19f1:0998]
        Interrupt: pin A routed to IRQ 16
        Region 0: Memory at da000000 (32-bit, non-prefetchable) [size=16M]
        Region 1: Memory at c0000000 (64-bit, prefetchable) [size=256M]
        Region 3: Memory at d8000000 (64-bit, non-prefetchable) [size=32M]
        Region 5: I/O ports at df00 [size=128]
        [virtual] Expansion ROM at dbf80000 [disabled] [size=512K]
        Capabilities: <access denied>
        Kernel driver in use: nvidia

However, i f you upgraded to a Fermi card like a GTX460 then it uses a 128MB+64MB+32MB configuration that could fit starting at F0000000 on your system.

Code:

01:00.0 VGA compatible controller [0300]: nVidia Corporation GF104 [GeForce GTX 460] [10de:0e22] (rev a1) (prog-if 00 [VGA controller])
	Subsystem: Giga-byte Technology Device [1458:34fc]
	Interrupt: pin A routed to IRQ 18
	Region 0: Memory at f8000000 (32-bit, non-prefetchable) [size=32M]
	Region 1: Memory at d0000000 (64-bit, prefetchable) [size=128M]
	Region 3: Memory at dc000000 (64-bit, prefetchable) [size=64M]
	Region 5: I/O ports at ef00 [size=128]
	[virtual] Expansion ROM at d8000000 [disabled] [size=512K]

Since your 4430s has Intel HD3000 graphics, a GTX460 would also get pci-e compression to greatly accelerate performance over anything your GTX275 could offer. There's some 768MB GTX460's at BB for $82AR: Galaxy GeForce GTX 460 (768MB) $81.99 after MIR Select Best Buy B&M (YMMV) - Slickdeals.net.

 

Considering I was going to buy a better graphics card if I could get my GTX275 to work as an eGPU, this is good news! I also want to use optiumus drivers so I can use the internal dispay, so I need the upgrade.

Would a GTX560 ti use the same 128MB+64MB+32MB configuration?

Thanks for the speedy reply!!

 

Yes. A GTX560Ti also uses 128MB+64MB+32MB pci-e BARs so would also work on your 4430s. Below is how it would appear in your Device Manager, as simulated on a Series-6 2560P.

 

PCIe = PCI Express
mPCIe = mini-PCI Express

I think you already know PCIe ports on desktop computer motherboard. Notebooks also have PCIe ports. The PCIe ports on notebook is smaller & it's called mPCIe ports. This is what mPCIe port look like:-


ExpressCard slot also connected to PCIe lane. To know the PCIe port number, you can use software like AIDA64. This is an example of AIDA64 report on the Chipset \ "South Bridge" section:-

• Ethernet device is integrated & not removable which means you can't use this PCIe port.
• WIFI, WWAN & Intel Turbo Memory are mPCIe devices & they are removable which means you can use the mPCIe port.
  • Note: x2 link setup required any of these combinations:-
    • PCIe port#1 & port#2
      OR
    • PCIe port#3 & port#4
      OR
    • PCIe port#5 & port#6

For empty PCIe port(s) we don't know where it's located & whether they are physically available or not. This is snippet from Dell Vostro 1310 schematic:-


This is AIDA64 report of the Chipset \ "South Bridge" for Dell Vostro 1310 ( source):-

*Broadcom BCM4310 USB Controller is actually a WIFI card.

Base on the schematic & AIDA64 report, we know:-

1. PCIe port#1 is not electrically connected, so you can't use this port.
2. PCIe port#2 is ExpressCard slot.
3. PCIe port#3 is Robson. Robson is referring to Intel Turbo Memory. Remember, I did mentioned earlier, Intel Turbo Memory is a mPCIe card. PCIe port#3 doesn't show up in AIDA64 report because Intel Turbo Memory is not connected to this port. You can confirm this port is available by opening the memory/HDD compartment, located underneath your notebook.
   
   TIPS: If there are more than one empty mPCIe ports, you can determine their port number by connecting WIFI card to them, then boot into windows & check in AIDA64.
4. PCIe port#4 is WIFI. WIFI card is removable.

Base on this analysis, your notebook can do x2 link setup with [PCIe port#3 & port#4 ]. Since both is mPCIe port, you'll need PE4H, two PM3N cards & two miniHDMI cables. You also can do x1 link setup using the ExpressCard slot with PE4H-EC2C kit but with x1 link only, performance will not good.

• Your notebook can't do x1.Opt setup because it missing the required/compatible iGPU (Intel GMA 4500MHD).
• I have tried 3ds Max Design but not extensively. Most consumer graphic cards are not tested. So, if you don't want to get FirePro or Quadro, Nvidia Fermi or ATI Radeon HD 4XXX/5XXX should work I think.
  
  Code:

  //////////////////////////////////////////////////////////////////////////////
  //////////   Review GPU Feature Support   ////////////////////////////////////
  //////////////////////////////////////////////////////////////////////////////
  Graphics Vendor Information: Direct3D 9.0 (ATI Radeon HD 5800 Series)
  GPU Shader Model Support: SM3.0
  Maximum pixel shader instruction count: 32768
  Viewport Shading Quality: Best
  Real-time Sun and Sky: Supported
  Real-time Architectural Material: Supported
  
  Visit http://www.autodesk.com/max-hardware for a list of certified graphics cards
  //////////////////////////////////////////////////////////////////////////////
  

I hope this answered your questions.

 

I was unable to get it to work with my x220 because of the junk bios. I think Ill be selling my diy dock, pm if you are interested

 

I'm assuming you're using latest BIOS v1.23 & it stuck at Windows logo while booting, right? Did you try my workaround?

 

~ deleted ~

 

Hi Kizwan,

Thanks for your help! I was able to install these optimus drivers. PhysX is now also enabled compared to my previous driver. I ran 3Dmark06 quickly and my score is now 6933 pts. Although I was more interrested in my Autodesk Inventor performance.

In the nvidia control panel I have put the preffered graphics adapter to Quadro 600 for the inventor.exe. So this should tell Inventor to use the external graphics card. But is doesnt seem to use it still! :X
Zooming an assembly is still as laggy as with the internal hd3000 graphics.

My Quadro 600 is not listed as a certified / recommended graphics solution following this page:
Autodesk - Certified Hardware - Find Graphics Hardware
(Although the new 2000 series (and up) based Fermi cards are recommended, unlike my fermi based 600 card)

Could this be the reason for Inventor to revert back to the internal graphics?

 

I think inventor will complain the 600 incompatibility (if it really not compatible), not the Intel.

I read in several forum, many "real" Optimus notebook unable to run Inventor using Nvidia graphic card. It simply use Intel HD Graphics. I think the problem is not Optimus. Probably Inventor only able to use/detect the primary GPU, which in this case Intel HD Graphics.

This might sound silly but anyway...try run GPU-Z in the background (select Nvidia at the drop-down menu in GPU-Z window) & then run inventor. See if it able to detect & use the Quadro.

EDIT: I found one person using Quadro 600 with Inventor.

 

You could be on to something here Kizwan! Thanks for the help already.
Putting GPU-Z in the background didn't solve the problem however.

I've also found some threads where people are having troubles using Inventor (or Autocad) using an Optimus laptop. Possibly have to do some more searches on a Inventor forum...

 

Nando with the new 2.0 version will there be any idear to get amd cards then? Or will optimus still be way better? Just now that the new ati cards are coming properly takes 2-3 month until we see GTX 660 ti

 

Hi, Bios 1.24 solved the TOLUD problem, just remember in power setting in bIOS set
"BIOS > Config > Power > Express Card Speed is set to "Generation 1". Leaving it on 'Automatic' causes windows to freeze when its trying to negotiate 2.0 link speed."
Like Xav said on his post

http://forum.notebookreview.com/7572285-post4023.html

It's working for me in 1.24 with 8gb ram

 

Thats exactly what I did and the setup is running now.



I was wondering, what's the performance limit of running these cards in x1 setup like in the picture? Can I do an estimate on the performance gain between (stronger) graphic cards? Or is there some way I can select a specific card based on troughput for example? Somehow I have the feeling putting a Quadro 6000 card in the PE4H adapter won't give me the full potential ;-)

 

Quick question for you guys.
Just finished setting up my dock. Im running the PE4H With a 6970 on a lenovo x220.

I can only run cod mw3 at high 1920x1080 w/ shadows, depth of field, smoke effects and bullet impact turned off otherwise its extremely laggy.

I know that the dock/expresscard setup cuts the performance down a lot but this seems like a bit too much?

Am I missing some part of the setup or something?
Thanks

 

You're missing NVidia Optimus' pci-e compression to give approximately doubling of bandwidth. Means that even ~ US$120 GTX460-1MB will outperform your three-times-more-expensive HD6970 on your x1 link.

Optimus' pci-e compression means that a NVidia solution will still have effectively doubling of bandwidth. Not to mention it transparently manages the internal LCD mode. See above response to see the pitfalls of going an AMD solution.

 

Sorry typo there its a 6870 so I didn't spend that much.

How does running optimus work since my laptop doesnt have an optimus enabled gpu?

So I'm assuming from what you said the x220 can only run x220 pcie and nvidia cards are significantly better because the drivers can compress data more or something along those lines?

So basically swap this thing for a gtx 460 esque gpu? What kind of performance can I expect with a gtx 460. Thanks for ur help

Sent from my GT-I9100M using Tapatalk

 

Your eGPU is x1.Opt setup. x1.Opt is the best configuration so far. You can get better performance when PCIe 2.0/Gen2 compliant kit available. Quadro 600 is not strong GPU. If you want to get better performance, you need better GPU.

Performance comparison:-
- Quadro 600 = GeForce 9800M GT
- Quadro 5000 = Radeon HD 6790
- Quadro 6000 = GTX560 (non-Ti)

Your eGPU is running only at raw PCIe x1 link speed when using ATI GPU. You'll experience stuttering & lagging when too many effects. With Nvidia Fermi GPU, you can do x1.Opt setup. Optimus provide PCIe data compression which boost the performance over PCIe x1 link. If we want to compare the performance, the x1.Opt performance is roughly similar or slightly better than x2 link setup.

If your x220 have Intel HD 3000 iGPU, you can do Optimus setup with Nvidia Fermi desktop graphic card.

 

Awesome thanks ill try yo get my hands on a Fermi then thanks

Sent from my GT-I9100M using Tapatalk

 

Shouldnt the 2.0 version be ready now btw?

 

The quadro 600 isn't a very stong GPU, I was aware of that. What I was trying to say is, what does limit the performance from this external GPU setup? The current PE4H x1 adapter or the graphics card?
As the PE4H setup is only x1 PCI bandwidth, can I use a (for example) quadro 4000, 6000, GTX560 and the like to full potential?

Thanks this is useful information. But then again, this x1 Optimus setup should provide roughly 5 GT/s right? Can I find the maximum transfer speed from specific graphic cards to compare?

 

bad news, no mxm adapter. good news, new versions of current products on the way.