EpicBlob

Glad it's working! Another way to repair the kexts is to download kext wizard; after I added the lines to the three kext files I ran kext wizard and it automatically repaired all my kexts. Tested with a 560, 660, and 670.

Also, I have a friend whose 770 shows up as a GPU, but not the specific 770. Anyone know why OS X isn't loading the specific model he has?

Thanks floppah.

I did a quick comparison of your FireStrike scores against the highest scores for the GTX660 (not-SLI) and found:

FPS:

Yours: 20.3fps

Theirs: 47.1fps

This means yours is about 42% of the speed of their (quite ridiculous) gaming machine which is probably completely overclocked and runs a top model i7 desktop CPU. I don't think this is actually a bad result as these guys probably spend their whole lives trying to maximise benchmark scores. Also, I don't think he can just unplug his GPU and take the rest of the machine with him.

I wonder what is constraining this? the Thunderbolt 1.0 interface or your MacBook's older i5 CPU.

I hope we can see some Thunderbolt 2.0 and Haswell i7 benchmarks soon to answer some of these questions.

Most likely the biggest constraint is the i5. I had 13inch macbook pro with a 2415m then upgraded to a mac mini with 3615qm and saw a huge difference.

Hi guys,

sorry my ignorance but do you have to use boot camp and windows in order to do use this GPU or can you use it also on OSX?

- - - Updated - - -

Hi guys,

sorry my ignorance but do you have to use boot camp and windows in order to do use this GPU or can you use it also on OSX?

Yes, OS X is usable with an e-GPU. There are several posts in this forum if you want to check them out for a better understanding.

So you followed the instructions to change the kext files? You might want to download kext utility and let it run. What this application does is it repairs all of your kext files. My card wasn't detected until letting kext utility run.

As far as I know, all nVidia cards will work (I've had os x running fine with a 560, 660, and 670) so that could be something to try. But there are hackintosh users with AMD cards so hopefully yours is supported.

Good luck!

@Nando,

I am really interested in using my MBP for serious gaming, hence an eGPU seems to be a good solution and your DIY eGPU Setup one critical part of it. But actually, I have a SFF case in which i am very soon using to build a small desktop PC, decent CPU and GPU included. What I still need is a mobile monitor. Since my GPU in my SFF case will have a DisplayPort-Out and my Macbook has a Thunderbolt2/Displayport connector, I was wondering if I could send the image from my SFF desktop, on which I would play, to my Macbook's internal Iris Pro and ultimately my Retina display. The guys at Apple told me that it wouldn't work under OS X, but I would use Windows 7 anyway. Could your DIY eGPU Setup enable me to do what I want (just using the MBP as an external monitor for my desktop PC through sending the later's image to my retina display via DisplayPort)? Or does anyone have a clue how to do this?

Thanks a lot.

Btw, I have a late 2013 15" MBPr, Iris Pro, 2,6 GHz.

So you essentially want to turn your MacBook pro into an external monitor? Someone correct me if I'm wrong, but I don't believe any of the connections on the MacBook are Input; they are all Output. Or if I flipped those around, you can only plug in monitors to your MacBook but can't use it as an actual monitor. Atleast that's how some of the older Macs are. Might be a way to hack it though.

Great Job! I've had my Mac Mini running in Windows 8 and OS X for a few months now and it has worked perfectly. Did you have 10.9 then upgrade to 10.9.1? I'm still on 10.9 but I've here 10.9.1 improves performance. I saw quite a dip in performance upgrading to 10.9 from previously having 10.8.3.

I'd also like to add that multiple monitors work in OS X with an egpu. I have a 27inch HDMI and a VGA monitor plugged in and both are working properly.

EDIT: disregard this. Didn't see your post and you don't seem to be using the Sonnet Echo TB/expresscard adapter.

Does:

-shut down with everything plugged in

-boot up mac mini, when you hear the mac chime unplug expresscard from the sonnet echo adapter

-When the windows loading screen appears, plug expresscard back in.

Work?

Yeah, the 670 seems to be the best card to get with 1.2 Optimus. Just bought a MSI 670 PE card and while I haven't done any concrete tests yet, medium settings 1080p resulted in a very consistent 60fps with FEW drops below (hit 55 for about a second then went back up). And got 7566 on 3Dmark11 P. Not too shabby.

I'm finishing up a comparison between a 660 and 670 at 1.2Opt so people can see the difference and make a call from there.

Yes, You can distinguish them by CPU usage as You can see, QM is almost sleeping. BF4 doesn't use additional cores. Maybe more visible differences would be with higher setting and more players but even if, only minimum and average fps would be little bit higher, in my opinion this difference is not worth of buying QM instead of dual core.

I would try upgrading to Windows 8 (if possible). The DX11.1 and 11.2 CPU optimizations have reportedly given significant fps increases in battlefield 4. It uses the CPU more effectively so you'd probably see more usage with the i7 while playing.

This FPS plot above is incorrrect! it's impossible to achieve such a difference because of CPU. I did comparison for 32 vs 64 players. Here You have correct plot for 32 players for both.

[ATTACH]9942[/ATTACH]

So are these 32 player Battlefield 4 comparisons, green being the dual core i5 and red being the quad i7? That's actually pretty close. When I upgraded from an i5 2415m to an i7 3615qm I saw a bigger increase.

I don't own a MACBook but I'm pretty sure you have an iGPU. What's your CPU?

Sadly, Apple disables the iGPU in Windows for the Macbook models with dGPUs in them.

here is BF4, Multi, FHD + medium. Operation: blockade, 64x64. This often shots of fps it's when I was shot.

I reserved rights for this post and plot. Don't use it anywhere else without my permission.

[ATTACH]9882[/ATTACH]

Thanks for the graph bjorm! So compared to my 660 and 3615qm, your system does seem to get a boost in performance. A 760 could probably net be a constant 60fps if I turn a few settings to low but kept most on medium. If I bought a 670, medium should get me a constant 60fps.

I find that lowering the resolution in full screen just completely ruins the way it looks (native resolution trumps all). Keeping it on low is sufficient for me in most cases. And I like full screen so I can see people far away more easily.

Nice!

I can imagine that it's a bit low FPS for a pleasant online play experience?

Yeah, I usually play with all settings on low except for mesh on high to see people from a distance. With that I get a few dips into the 50s but overall a much more consistent 60fps.

64-man multiplayer performance in Battlefield 4. Default Medium Setting and 1080p resolution. Specs are in my description. This was on paracel storm which is one of the most demanding maps performance wise.

Overall average was 53 fps. This was done on just my external screen.

I'll do some test tomorrow on BF4 with 1080p on medium preset, external.

Sweet, thanks a ton man! I'll record the fps for my setup as well so we can get a nice comparison going

and in my post found what? benchmarks?

Whoops completely missed that post lol my bad. But I want to see if I can get a consistent 60+fps 1080p if I upgrade to a 670. I can get a almost perfect consistent 60 on low 1080 with my 660, but I wouldn't mind to bump up a few of the settings. You have a 760 which is close but not quite as powerful as a 670. What kind of framerate would you get if you bumped up the res to 1080 and lowered the settings to medium? Also is this internal or external screen?

page ago posted similar: http://forum.techinferno.com/diy-e-gpu-projects/2109-diy-egpu-experiences-%5Bversion-2-0%5D-222.html#post75157

Looking for actual in-game performance though. Benchmarks can only go so far

@MikjoA 's score is with a really nice OC, my score is with a slight factory OC.

3Dmark11 is much more depended on GPU perfomance, compared to 3Dmark06 which depends much more on CPU performance (that's why I have better 06 score).

So expect a little bit lower 3DMark11 score w/o OC,

Hope it helps!

Thanks! Yeah I usually want to base performance on 3dMark11 because of this (and it's newer lol).

Also, what kind of fps do you get in games? I know the 5Gb/s will limit the boost I get compared to upgrading from a 660 to 670 in a desktop, but I'm sure there will still be a boost. If you've tried out Battlefield 4 that would be great to know

jacobsson - 3820QM:

NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i7-3820QM Processor,Hewlett-Packard 17DF

MikjoA - 3612QM:

NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i7-3612QM Processor,Sony Corporation VAIO

Damn that's a nice improvement. Thanks!

Does anyone have a 3Dmark11 P score that uses a quad core CPU (preferably ivy bridge or haswell), 670 gpu, and 1.2opt? Thinking of getting a 670 and I'm wondering what kind of boost I might see.

Try this:

2. Shut down the Computer ( reboot is not sufficient )

3. Make sure your entire setup is plugged in and on except your computer which should be turned off. (egpu is turned on and the expresscard is plugged into the sonnet echo adapter, adapter is plugged into your macbook thunderbolt bolt. Also if you have an external monitor, it is plugged into your egpu).

4. Turn on the computer. Your screen will turn grey and you will hear the Mac Chime. While the screen in white, unplug the expresscard from your sonnet echo adapter.

5. Your screen will now turn black. When the Windows icon appears and is loading, plug back in the expresscard to your sonnet echo adapter.

You are pretty much unplugging the expresscard when you boot up, then replugging it in when the windows 7 bootscreen appears.

Also, what does your Device manager say under display adapter?

Hey guys I have a question a bit off-topic so please excuse me. Ok, so I've the i3-3210m with GTX660 and I'm having lags in Crysis3 and BF4. Isn't the CPU kind of a bottleneck? Upgrading it to a 4 cores would be good or useless?

It's probably your CPU. I had a 2415m i5 processor in my macbook pro with a 660 then upgraded to a mac mini 3615qm i7 and that got rid of the frame rate issues I had. Especially with Bf4 and Crysis 3 being heavy CPU (and GPU) games.

Oh, I thought you needed a dGPU with Optimus to allow for your eGPU to have it. Or is this just something driver related?

Shows how much I know about this ;p

You need an Intel HD graphics card to enable optimus. It's actually better if you just have an integrated card instead of a discrete card for an egpu system.

another proof that GTX560Ti is not limited by i5 SB or IB (there is no additional scoer for GraphicsScore and CombinedScore as well). Any stronger GPU needs i7QM -> when GTX570 tested on i5 SB vs i7QM SB, there was 1000 point combined score more.

i5-2540M + GTX560Ti

i7-2630QM + GTX560Ti

i7-3632QM + GTX560Ti

Now play a multiplayer game such as Battlefield 3 that utilizes a ton of CPU performance for multi. my 3615qm paired with a 660 is lightyears ahead of my 2415m with a 660.