Jump to content

2014 15" MBP Iris + R9_280X@16Gbps-TB2 (AKiTiO Thunder2) + Win8.1/OSX10.10 [braynshoc


Recommended Posts

These are some shots of extensions loaded and some basic benches.

Should get 35fps in Valley.

Here is my screenshot with the 2014 15” rMBP Iris Pro + AKiTiO + HD 7970:

post-28870-14495000187093_thumb.png

As said, OpenGL performance is not optimized for OS X, R9 280X should give similar results, maybe a bit better because of higher clock speed.

I got a 7870 under 100€ and considering that, not many frames away from HD 7970:

post-28870-14495000187528_thumb.png

Obviously OS X bottlenecks the pixel rendering speed with high-end AMDs. Windows is much better gaming platform. OpenCL computing is that where AMD rules, but Maxwell cards have catched up somewhat. I am waiting for the new R9 380X. Competition develops technology.

Link to comment
Share on other sites

@Tech Inferno Fan,

I did first test you requested, by removing the added on bridge wires to the slot, and removing the tape to have the ATX power supply feed power back thru the TB/PCIe connection.

There seemed to be no change in stability or instability, card acted the same.

(I'm considering putting in the 750Ti to test it out in Windows, since I now have it installed)

Based on what you are suggesting would it also be a good test to use two separate power supplies, disconnect the 4-wire power feed from the PCIe board to TB2 card, tape the connector again and use the current ATX card to power the PCIe board(and AMD card), and use the original power brick to power just the TB card?

Link to comment
Share on other sites

Obviously OS X bottlenecks the pixel rendering speed with high-end AMDs. Windows is much better gaming platform. OpenCL computing is that where AMD rules, but Maxwell cards have catched up somewhat. I am waiting for the new R9 380X. Competition develops technology.

It's not just OS X.

If I put GTX980 in 4,1 MP I get around 50 fps.

If I [put GTX980 in 2013 nMP via eGPU I still get 50fps.

If I put 7970/R9 280x in 4,1 cMP I get 36 fps.

If I put 7970 R9 280x in 2013 nMP via eGPU I get 22fps.

So while Nvidia cards don't lose much the AMD card takes a 30% speed hit. That isn't anything to do with OS X vs Windows. That is something wrong.

Might be worth trying same trick Netkas found for 3,1 Mac Pro.

Link to comment
Share on other sites

It's not just OS X.

If I put GTX980 in 4,1 MP I get around 50 fps.

If I [put GTX980 in 2013 nMP via eGPU I still get 50fps.

If I put 7970/R9 280x in 4,1 cMP I get 36 fps.

If I put 7970 R9 280x in 2013 nMP via eGPU I get 22fps.

So while Nvidia cards don't lose much the AMD card takes a 30% speed hit. That isn't anything to do with OS X vs Windows. That is something wrong.

Might be worth trying same trick Netkas found for 3,1 Mac Pro.

Yes, thus far my GTX 780 6GB gives the best OpenGL performance, even better than reference GTX 980 in OS X. I don’t have the iGPU acceleration yet, but I do have the full Nvidia external screen acceleration on 2014 15” rMBP and 2014 Mac mini, both running Yosemite 10.10.3. No doubt that Nvidia beats AMD in this field.

So I just wanted to emphasize the practical benefit of OpenGL in OS X. Additional frames don’t give much fun for me. Nowadays I am more interested in offscreen computing. For a OS X Steam gamer your point is very important.

Programmers are lazy to optimise games for OpenGL. Windows / DirectX API games sell like hot cakes, unfortunately. And console games are optimized even further because of limited hardware resources. It’s a bit amusing that programming IDEs have grown to giants over the last ten years, Visual Studio as an example. Better CPUs and 16GB RAM, but UI responsiveness has remained the same. Beach ball rolling on OS X is a rare vision, one of the reasons that I prefer more Mac than PC :)

It’s all about drivers’ optimization generally, and likely something else as well from the part of AMD OpenGL in OS X as you compared. I didn’t look what Netkas found, thanks for the tip!

Link to comment
Share on other sites

Well,

I spent the day doing a lot of tests:

@goalque verified my kext edits were identical to his.

First-

For the fun of it, since I had win8 installed i tossed the 750Ti in, and ran benchmarks.

It ran 3dMark11 3 times in a row no problems, only using the 120W power brick I bought.

When I rebooted to OSX, I modified the kexts necessary, and installed the nVidia drivers, Recognized the card and cuda-Z saw it no problem.

of course no external video, would not see a display connected to any of it's 3 ports (2 DVI and mini-HDMI)

Second-

Tossed in the Radeon 6450, Win8 found it, and had no trouble with running the benchmarks (although slow FPS).

rebooted to OS X modified the AMD6000Controller and AMDRadeonX3000 kexts, came right up, no issues, ran unique Heaven and Luxmark tests.

Other than slow FPS, it worked like a champ on both HDMI and DVI ports.

Third-

Cleared PRAM - reset nvram boot-args=kext-dev-mode=1

Put the 280X back in the ATiKiO, hooked the ATX PSU back up, and win8 recognized it just fine, ran one 3dMark11 test, then failed the second...

Rebooted to OSX, with just AMD7000Controller, AMDRadeonX4000, and AMDSupport the only modified kexts.

System Profiler finds the card, but only calls it DISPLAY (not 7xxx like it should)

Added IOPCITunnelCompatible to IONDRVSupport, rebooted -- fails to boot.

put back original IONDRVSupport, and modified AppleHDAController, rebooted -- nothing new happens, just sees the card like DISPLAY

At no time does AMDFrameBuffer kext load. My guess is there is something causing it not to load that kext. Or it's a power mgmt thing, but i would think it would at least boot up.

This is the last few entries in syslog before i have to power it off and reboot:

5/30/15 9:24:39.000 PM kernel[0]: ** GPU Hardware VM is enabled (multispace: enabled, page table updates with DMA: enabled)

5/30/15 9:24:39.583 PM WindowServer[143]: Session 257 retained (2 references)

5/30/15 9:24:39.583 PM WindowServer[143]: Session 257 released (1 references)

5/30/15 9:24:39.599 PM WindowServer[143]: init_page_flip: page flip mode is on

5/30/15 9:24:39.000 PM kernel[0]: display: Not usable

Anyway, thought I'd share what I had done today.

--BrayNShocK

Link to comment
Share on other sites

Well,

I spent the day doing a lot of tests:

@goalque verified my kext edits were identical to his.

First-

For the fun of it, since I had win8 installed i tossed the 750Ti in, and ran benchmarks.

It ran 3dMark11 3 times in a row no problems, only using the 120W power brick I bought.

When I rebooted to OSX, I modified the kexts necessary, and installed the nVidia drivers, Recognized the card and cuda-Z saw it no problem.

of course no external video, would not see a display connected to any of it's 3 ports (2 DVI and mini-HDMI)

Second-

Tossed in the Radeon 6450, Win8 found it, and had no trouble with running the benchmarks (although slow FPS).

rebooted to OS X modified the AMD6000Controller and AMDRadeonX3000 kexts, came right up, no issues, ran unique Heaven and Luxmark tests.

Other than slow FPS, it worked like a champ on both HDMI and DVI ports.

Third-

Cleared PRAM - reset nvram boot-args=kext-dev-mode=1

Put the 280X back in the ATiKiO, hooked the ATX PSU back up, and win8 recognized it just fine, ran one 3dMark11 test, then failed the second...

Rebooted to OSX, with just AMD7000Controller, AMDRadeonX4000, and AMDSupport the only modified kexts.

System Profiler finds the card, but only calls it DISPLAY (not 7xxx like it should)

Added IOPCITunnelCompatible to IONDRVSupport, rebooted -- fails to boot.

put back original IONDRVSupport, and modified AppleHDAController, rebooted -- nothing new happens, just sees the card like DISPLAY

At no time does AMDFrameBuffer kext load. My guess is there is something causing it not to load that kext. Or it's a power mgmt thing, but i would think it would at least boot up.

This is the last few entries in syslog before i have to power it off and reboot:

5/30/15 9:24:39.000 PM kernel[0]: ** GPU Hardware VM is enabled (multispace: enabled, page table updates with DMA: enabled)

5/30/15 9:24:39.583 PM WindowServer[143]: Session 257 retained (2 references)

5/30/15 9:24:39.583 PM WindowServer[143]: Session 257 released (1 references)

5/30/15 9:24:39.599 PM WindowServer[143]: init_page_flip: page flip mode is on

5/30/15 9:24:39.000 PM kernel[0]: display: Not usable

Anyway, thought I'd share what I had done today.

--BrayNShocK

Did you mod the Tahitiframebuffer kext?

I have seen it appear in the list.

Link to comment
Share on other sites

Did you mod the Tahitiframebuffer kext?

I have seen it appear in the list.

Yes, I did edit the Tahiti section of the AMDRadeonX4000 kext.

the only thing showing in syslog is that IOPCITunnelCompatible is missing from IONDRVSupport.

But when I add it, It fails to boot, producing the logs I pasted above.

Link to comment
Share on other sites

Tomorrow I may have access to a 2014 rMBP 15" with Iris only. (no dGPU)

Should get us an answer.

UPDATE: 3DMark11 on 2014 Mini with 7970 (same as R9 280X)

http://www.3dmark.com/3dm11/9877555

Will try some more runs.

UPDATE: Run #2

http://www.3dmark.com/3dm11/9877592

UPDATE: Run #3

http://www.3dmark.com/3dm11/9877608

Was this the test with reported trouble or is it a different one?

Link to comment
Share on other sites

Tomorrow I may have access to a 2014 rMBP 15" with Iris only. (no dGPU)

Should get us an answer.

UPDATE: 3DMark11 on 2014 Mini with 7970 (same as R9 280X)

AMD Radeon R9 280X video card benchmark result - Intel Core i5 4350U,Apple Inc. Mac-35C5E08120C7EEAF

Will try some more runs.

UPDATE: Run #2

AMD Radeon R9 280X video card benchmark result - Intel Core i5 4350U,Apple Inc. Mac-35C5E08120C7EEAF

UPDATE: Run #3

AMD Radeon R9 280X video card benchmark result - Intel Core i5 4350U,Apple Inc. Mac-35C5E08120C7EEAF

Was this the test with reported trouble or is it a different one?

My MSI HD 7970 and older AMDs can easily loop 3DMark11 physics ten times in a row, the problem seems to be R9 270X / R9 280X specific with the AKiTiO. And the crash happens only in situations when the GPU is not stressed. Playing BF4 one hour won't crash the system. Very odd.

Booting into OS X is a separate issue. I have only experience of Asus R9s, cannot comment how compatible other R9s are with the OS X. Maybe he could try with the different VBIOS. Still suspect that there is something wrong with the kext even though I checked that they were identical..

  • Thumbs Up 1
Link to comment
Share on other sites

I realise this thread is all about AMD cards but I just want to say that this sounds similar the same baffling issue I have with my NVIDIA setup. (Mac mini 2011, GTX 970 in an ATiKiO). In other words, if I don't modify IONDRVSUPPORT kext then it boots but shows only as "NVIDIA chip model" in system information. Whereas if I correctly modify IONDRVSUPPORT then Yosemite kernel pancs at the point it tries to start the login screen / desktop. (And it works great in Windows.) None of us has come up with the answer to this but the only key difference with my system to other Mac minis is that it has a Radeon dGPU.

  • Thumbs Up 1
Link to comment
Share on other sites

My MSI HD 7970 and older AMDs can easily loop 3DMark11 physics ten times in a row, the problem seems to be R9 270X / R9 280X specific with the AKiTiO. And the crash happens only in situations when the GPU is not stressed. Playing BF4 one hour won't crash the system. Very odd.

Booting into OS X is a separate issue. I have only experience of Asus R9s, cannot comment how compatible other R9s are with the OS X. Maybe he could try with the different VBIOS. Still suspect that there is something wrong with the kext even though I checked that they were identical..

I have spent more time testing AMD eGPUs today. Shortly after the 3rd run of 3D Mark 2011, I tried a reboot, it was fine. But then whilst reading some things online I saw blocky artifacts all over the desktop. It was messy but machine kept running. I noted in GPU-Z that the GPU was running at 49% load, just rendering desktop.

I tried running another benchmark, and the artifacts cleared up instantly. Once the bench was done, artifacts back. This was on a 7970. Keep in mind that there is LITERALLY no difference between a 7970 and an R9 280X. They are both Vendor ID 1002 and Device ID 6798. In fact, the 3D Mark results call it a R9 280X.

For the most part a R9 280X is a higher clocked 7970. In fact, many R9 280X are built on 7970 reference PCB, and you could put a reference 7970 fan on them as all screws line up.

I then ran Uningine Valley with OpenGl instead of Direct-X using Extreme HD. And I got 26 fps, so still below what card would do in cMP in OS X.

After this I put in a R9 290X, the Hawaii card. For awhile I ran 3D Mark and it was fine, got to end but couldn't connect with server. Then, failure. Over and over again it would suddenly lose image and blast it's fan at 100%. I tried various methods of powering the card, splitting the AKitio on different PSU and also having both card and Akitio on same PSU. Makes no difference, it will NOT complete 3D Mark. Sometimes it craps out when it starts benchmark, authorities half way through. And one time I saw artifacts appear on an idling desktop, followed quickly by black screen and roaring fan.

I don't think you can miss the conclusion, at least from my tests. In both OS X and Windows there is some issue with power management. In OS X it seems stuck in a lower power state, unable to reach full 100% GPU speed. In Windows the 7970/R9 280X can run at 100% pretty well, it seems to have much more trouble going back into low speed. It was as if parts of it went back down to lower speed but some part of it was still at full speed. The desktop artifacts looked very much like an overheating card or a card with too much speed via an overclock. I haven't tried the R9 290X in OS X and may not bother.

It is possible that an AMD GPU could be used for gaming in Windows, but I don't think it will always return to normal state afterward.

While the 7970/R9 280X was more stable in OS X, it was also running at 2/3 speed.

From this I would have to guess that the Akitio may not provide all the x16 connections that AMD cards uses to run their power management. Somehow, this issue has no bearing on Nvidia cards. After multiple crashes on R9 290X I put a Titan-X in right away. Using same power setup I was able to run latest 3D Mark to completion, no issues at all.

In addition to having the barrel plug on a PCIE plug, this particular Akitio is the one that also has a PCIE power plug routed to Akitio PCB to supply power for the card directly at power pins. I can not imagine any way that the power was not available, there is some other issue to do with managing that power.

King of a bleak picture, Nvidia cards have lots of display issues in OS X as eGPU. AMD cards are much easier to have eGPU display output in OS X, but have issues in both OS's with power management. If it really is a missing PCIE line or some function of handing temp data to the logic board that isn't connected, then will be very hard to fix.

I will have a look around and see if there are reports of stable AMD cards from Tahiti or Hawaii in other eGPUs. Not going to put much more time in this, not an AMD fan and this issue seems completely avoided by Nvidia cards.

Link to comment
Share on other sites

IAfter this I put in a R9 290X, the Hawaii card. For awhile I ran 3D Mark and it was fine, got to end but couldn't connect with server. Then, failure. Over and over again it would suddenly lose image and blast it's fan at 100%. I tried various methods of powering the card, splitting the AKitio on different PSU and also having both card and Akitio on same PSU. Makes no difference, it will NOT complete 3D Mark. Sometimes it craps out when it starts benchmark, authorities half way through. And one time I saw artifacts appear on an idling desktop, followed quickly by black screen and roaring fan.

I don't think you can miss the conclusion, at least from my tests. In both OS X and Windows there is some issue with power management. In OS X it seems stuck in a lower power state, unable to reach full 100% GPU speed. In Windows the 7970/R9 280X can run at 100% pretty well, it seems to have much more trouble going back into low speed. It was as if parts of it went back down to lower speed but some part of it was still at full speed. The desktop artifacts looked very much like an overheating card or a card with too much speed via an overclock. I haven't tried the R9 290X in OS X and may not bother.

It is possible that an AMD GPU could be used for gaming in Windows, but I don't think it will always return to normal state afterward.

While the 7970/R9 280X was more stable in OS X, it was also running at 2/3 speed.

From this I would have to guess that the Akitio may not provide all the x16 connections that AMD cards uses to run their power management. Somehow, this issue has no bearing on Nvidia cards. After multiple crashes on R9 290X I put a Titan-X in right away. Using same power setup I was able to run latest 3D Mark to completion, no issues at all.

In addition to having the barrel plug on a PCIE plug, this particular Akitio is the one that also has a PCIE power plug routed to Akitio PCB to supply power for the card directly at power pins. I can not imagine any way that the power was not available, there is some other issue to do with managing that power.

King of a bleak picture, Nvidia cards have lots of display issues in OS X as eGPU. AMD cards are much easier to have eGPU display output in OS X, but have issues in both OS's with power management. If it really is a missing PCIE line or some function of handing temp data to the logic board that isn't connected, then will be very hard to fix.

I will have a look around and see if there are reports of stable AMD cards from Tahiti or Hawaii in other eGPUs. Not going to put much more time in this, not an AMD fan and this issue seems completely avoided by Nvidia cards.

Is it AKiTiO's x4 electrical link that's causing AMD cards to fail during 3dmark11?

You raise an interesting point. goalque's netstor enclosure without the problem is a 3-slot device utilizing a PCIe bridge to share out the PCIe link. That bridge connects to the video card electrically at x8 or x16 in a similar way to the Sonnet III-D. The AKiTiO connects at x4.

Power management can be tweaked a bit under Windows. In your Power plan -> PCI Express -> Link Power State Management, can you try to set that as Disabled?

Also check AMD Catalyst (or whatever the latest AMD control panel is), looking for a power management option there. You want something that's equivalent to the NVidia "Power Management Mode" = Prefer maximum power.

Then do a 3dmark11 test again observing if stability is achieved during the physics test.

Link to comment
Share on other sites

Isn't the difference between x4, x8 and x16 "just" the amount of lanes (data)? Power is provided within the first x4 slots.

Yes. Point is the pcie bridge in those 3 slot thunderbolt enclosures are probably managing power differently than a 1 slot device.

@goalque , @MVC Pls try to disable pcie link and and card power management per my previous post to see if that gives any better 3dmark11 physics test stability.

  • Thumbs Up 1
Link to comment
Share on other sites

Yes. Point is the pcie bridge in those 3 slot thunderbolt enclosures are probably managing power differently than a 1 slot device.

@goalque , @MVC Pls try to disable pcie link and and card power management per my previous post to see if that gives any better 3dmark11 physics test stability.

Link State Power Management:

On Battery: Off

Plugged in: Off

Also disabled ULPS (Ultra Low Power State) by registry value search and changed the default EnableUlps = 1 to 0.

Didn’t help. 3DMark11 physics gave the blank white screen again, no other choice than power off. And before that it crashed when I was just opening Steam app.

  • Thumbs Up 1
Link to comment
Share on other sites

Link State Power Management:

On Battery: Off

Plugged in: Off

Also disabled ULPS (Ultra Low Power State) by registry value search and changed the default EnableUlps = 1 to 0.

Didn’t help. 3DMark11 physics gave the blank white screen again, no other choice than power off. And before that it crashed when I was just opening Steam app.

The real question now is does this AMD R7/R9 problem affect all 1-slot Thunderbolt enclosures eg: Sonnet SEL, Rocketstar, etc? If it does then we know its some anomoly with the PCIe link. If it doesn't, then it's some anomoly with the AKiTiO Thunder2.

I'm not sure if Macbooks enable ASPM or not. Can you check using the guide at http://forum.techinferno.com/hp-business-class-notebooks/2537-12-5-hp-elitebook-2570p-owners-lounge.html#post38399 (second section)?

Link to comment
Share on other sites

The same "low power" issue seems to be on OS X as well, but not that apparent as on Windows. The taping trick + 75W bypass mod does have an effect, so that you can pass 1-3 physics tests. My record is 5 or something :) As backpowered, it's 100% physics crash with the R9. I ran the tests with the power adapter plugged in.

I saw these lines in the output HTML:

Power Policy:PCI Express ASPM is disabled (On Battery)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

Power Policy:PCI Express ASPM is disabled (Plugged In)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

Link to comment
Share on other sites

My original 2006 Mac Pro has had 100's or even 1000's of GPUs tested in it.

As a result Slot #`1 has gotten very worn. The unique thing about this model is that there is a utility that allows you to allocate PCIE lanes among the four slots.

Very cool thing, actually. Anyhow, after some sparking from a AMD 4870 x2 I realized that I had burned a PCIE connector or trace. With AMD cards the only way to have them run is to put slot 1 in x8 or x4 mode. If slot #1 is in x16 mode, AMD cards don't work. Oddly, Nvidia cards are completely oblivious and un affected by this.

Point is, to me this was a sign that AMD and Nvidia were using whatever line got burned up in a different way. I suspect it is something similar going on here. Especially as there are related issues in either OS, so not a driver thing.

So if AMD cards are out and most Nvidia cards don't offer display output in OS X, really makes slim pickens for Mac users.

Link to comment
Share on other sites

netkas posted up a screen shot of an Nvidia card as eGPU for TB2 mac mini, stating he's gotten external output working...more to come.

hopefully that helps us rMBP Iris Pro folks out...

I'm a bit reluctant to step up to a nestor or sonnet TB enclosure, just purely on the fact that this will still not be a supported solution (i.e. buggy sleep issues, etc) at the cost of over a $1000. And since I have the pieces of my previous hackintosh sitting here (i7-4770k, GB z97, 16gb, 500gb ssd) still sitting here I would probably be better off throwing the 280X in it, and call it a day...

For now, I've got this 6450 in the AKiTiO, driving two monitors, and driving my 3rd from the pass-thru thunderbolt port :)

Not fast, nor completely offloading the iGPU, but at least a partial solution for this week, until external display for nvidia is public... (go netkas!)

Thanks for all the help on this!

If you have more ideas you'd like me to test, toss them out! (especially on making the AKiTiO more stable, I'll be happy to futz with it.

Link to comment
Share on other sites

The same "low power" issue seems to be on OS X as well, but not that apparent as on Windows. The taping trick + 75W bypass mod does have an effect, so that you can pass 1-3 physics tests. My record is 5 or something :) As backpowered, it's 100% physics crash with the R9. I ran the tests with the power adapter plugged in.

I saw these lines in the output HTML:

Power Policy:PCI Express ASPM is disabled (On Battery)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

Power Policy:PCI Express ASPM is disabled (Plugged In)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

Is there power management settings in Catalyst. Some sort of dynamic option that can be set? Also, can you disable Powerplay? Idea is to leave the AMD card running in full power mode.

Some discussion about Powerplay causing issues on desktop systems at How to Fix your HD 7970 from crashing / locking up on monitor sleep - Page 2 - [H]ard|Forum .

Link to comment
Share on other sites

Power Policy:PCI Express ASPM is disabled (On Battery)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

Power Policy:PCI Express ASPM is disabled (Plugged In)

The current power policy for PCI Express Active State Power Management (ASPM) is configured to Off.

I used this tutorial: PCIe Link State Power Management - Turn On or Off in Windows - Windows 7 Help Forums

And disabled ASPM up above, and It made no difference, so I don't this that is the culprit.

Catalyst doesn't have a function like that in the menu, at least that I could find.

The thread mentions using MSI's Afterburner, can I use MSI's app with an XFX card?

worth a try i suppose :)

Link to comment
Share on other sites

netkas posted up a screen shot of an Nvidia card as eGPU for TB2 mac mini, stating he's gotten external output working...more to come.

hopefully that helps us rMBP Iris Pro folks out...

I'm a bit reluctant to step up to a nestor or sonnet TB enclosure, just purely on the fact that this will still not be a supported solution (i.e. buggy sleep issues, etc) at the cost of over a $1000. And since I have the pieces of my previous hackintosh sitting here (i7-4770k, GB z97, 16gb, 500gb ssd) still sitting here I would probably be better off throwing the 280X in it, and call it a day...

For now, I've got this 6450 in the AKiTiO, driving two monitors, and driving my 3rd from the pass-thru thunderbolt port :)

Not fast, nor completely offloading the iGPU, but at least a partial solution for this week, until external display for nvidia is public... (go netkas!)

Thanks for all the help on this!

If you have more ideas you'd like me to test, toss them out! (especially on making the AKiTiO more stable, I'll be happy to futz with it.

Netstor NA211TB is actually $849 and they are one of the rare companies that officially support GPUs.

https://www.youtube.com/watch?v=58b0xVywpNM

AFAIK, the only other company that mentions GPUs is One Stop Systems. Their CUBE is $650 (one slot), but with only 180W external power supply. The dual 180W version is not UL certified. Backplanes, TB1/TB2 cards can be bought separately on their web site (oddly, earlier it wasn’t possible because of Intel rules). The TB2 version is exactly the same card that Sonnet SEL uses and it costs $395. Not cheap.

Intel has the same 30W requirement for them as well:

"30W max, 6W for the adapter and 24W for the Thunderbolt downstream power (12W per Thunderbolt port)"

Thunderbolt 2 Option Card | MaxExpansion.com

Sonnet III-D belongs to the same category, but is more expensive, they don’t support GPUs officially and even though it has a 300W PSU, it provides only one 6-pin up to 75W.

I can guarantee that Netstor product gives stability for AMDs, but it doesn’t solve your particular boot problem with the 280X. Must be something card specific. This might interest you:

Problem with XFX Radeon R9 280x DD and UEFI BIOS - ATi - InsanelyMac Forum

If you had a Kepler architecture card such as GTX 780, the most simple solution would be to switch back to Mavericks with latest security updates installed - that gives you 100% Nvidia eGPU screen output. Also multiple external eGPU monitors are supported out of the box. You don't need Nvidia web drivers.

I am on the trail of more related kexts as I compared Mavericks builds by diff command. As you have Hackintosh background and the same MBP, let’s join the effort for Yosemite. I already proved by screenshots that it’s possible on 2014 rMBP and 2014 Mac mini.

  • Thumbs Up 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.