Jump to content

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


Recommended Posts

Been lurking for a while. I wanted to share some of my experience with the AKiTiO enclosure.

I am specifically trying to make AMD cards work with Yosemite, as I have a Iris Pro only mid 2014 rMBP, and cannot get external displays working with my GTX 750Ti...

I have only been able to get a couple of cards to work (Microcenter is starting to get tired of me exchanging cards every other day.). Finally, decided I am going to fight to get the R9 280X to work with Yosemite, even though it's larger and more expensive than I originally wanted. While I;m having trouble with Yosemite, I decided I would try Win8.1/BootCamp and see if at least the hardware worked.

First I bought a larger 10A(120W) power brick, did the tape trick and soldered power to the PCIe slot to support 75W to the slot. I used a Corsair CS650M ATX power supply for the 6 and 8 pin power to the XFX R9 280X DD 3GB card. Not wanting to mess with the enclosure yet (in hopes someone will figure out the nvidia output on the iris pro rMBP), I have the PCB naked on my desk.

First boots into windows, were ok at the desktop, but unique heaven at anything above 1280x720, or using full-screen caused failure (blinking screens, crashes, or snowy)... I thought maybe it was the two different power supplies, I went ahead and soldered 2 x 18g yellow and black wires to the CPU 8pin connector on the Corsair ATX PS. Making the ATiKiO completely running off of the Corsair PSU.

Immediately saw improvement in Windows 8.1, boots perfectly, and runs Heaven at full screen @1080p, with HIGH settings, and 2X AA, it ran smoothly and perfectly. AVG FPS was somewhere around 50FPS (can't remember off the top).

So the dual power supplies were causing some instability.

Now if I can figure out how to get Yosemite to boot fully with the IOPCITunnelCompatible injected into kexts, I would be happy...

Thanks for all the insights here, especially goalgue, Tech Inferno Fan, mvc -- Thanks!

  • Thumbs Up 1
Link to comment
Share on other sites

Been lurking for a while. I wanted to share some of my experience with the AKiTiO enclosure.

I am specifically trying to make AMD cards work with Yosemite, as I have a Iris Pro only mid 2014 rMBP, and cannot get external displays working with my GTX 750Ti...

@braynshock , awesome you have a R9 280X going successfully with a AKiTiO Thunder2.

I wish you could unsolder the 75W bypass, remove the tape off the slot and test stability. Based on latest update here there should be no reason why that should not be equally as stable.

One other point, @goalque had problems running a R9 280X reported here, even with the 75W slot bypass. He couldn't get a full run of 3dmark11. That's put a big question mark over the Thunder2's stability with AMD cards.

Would you mind testing if you can get one full 3dmark11 run, then repeat if OK to confirm stabilty across multiple runs?

You can see at http://forum.techinferno.com/diy-e-gpu-projects/6578-implementations-hub-tb-ec-mpcie.html#Thunderbolt we have a heavily NVidia favored implementation list. Would you mind creating a guide so can add one to the red side? Would help those wanting best OpenCL performance.

Link to comment
Share on other sites

@Tech Inferno Fan,

I will try removing the tape and unsoldering the direct wires to the PCIe slot later tonight. But I did go ahead and download the 3Dmark11, and true to form it did fail twice (machine froze) about 3-5 seconds into the physics test. sigh...

I am now questioning if i should just put the 750ti back in, and spend my effort on editing frame buffer to get the external monitors working... I have really had very little success with these AMD cards and Yosemite (or mavericks for that matter).

I've seen some of the other forums mentioning that the Nestor box may actually be more stable, would that be better option for running AMD cards? or do you think it's really more of a question of TB2 and AMD? I didn't mind dropping $200 to try the AKiTiO, but $1k for the nestor is a little much for experimenting, you know? :)

Last second thoughts:

Windows has no problem driving 3 x 1080p displays, but it seems the physics test is just too much (on only one display of course). I did try under clocking the card a bit... I may try it a few more times to see if bringing it down 50-100mhz will help with stability.

--BrayNShocK

  • Thumbs Up 1
Link to comment
Share on other sites

@Tech Inferno Fan,

I will try removing the tape and unsoldering the direct wires to the PCIe slot later tonight. But I did go ahead and download the 3Dmark11, and true to form it did fail twice (machine froze) about 3-5 seconds into the physics test. sigh...

I am now questioning if i should just put the 750ti back in, and spend my effort on editing frame buffer to get the external monitors working... I have really had very little success with these AMD cards and Yosemite (or mavericks for that matter).

I've seen some of the other forums mentioning that the Nestor box may actually be more stable, would that be better option for running AMD cards? or do you think it's really more of a question of TB2 and AMD? I didn't mind dropping $200 to try the AKiTiO, but $1k for the nestor is a little much for experimenting, you know? :)

Last second thoughts:

Windows has no problem driving 3 x 1080p displays, but it seems the physics test is just too much (on only one display of course). I did try under clocking the card a bit... I may try it a few more times to see if bringing it down 50-100mhz will help with stability.

--BrayNShocK

Yes, 3DMark11 physics is a good stability indicator for R9 270X / R9 280X AMDs. Would you tell the exact model?

Note that my Asus AMDs NEVER crashed when stressed by Valley, other parts of 3DMark11, Fire Strike (except the physics), or playing intensively BF4.

I have examined AKiTiO’s AMD issues thoroughly. Soldering mod helps to run 3-4 physics in a row at best, but eventually the whole eGPU will crash in normal tasks such as browsing the web. The common factor is low power consumption. I tried with the default factory clocks.

Netstor is pricey but rock stable, loops 3DMark11 physics test infinitely. Older architecture AMDs seems to run more stable with the AKiTiO. I can help with the kext mods.

Link to comment
Share on other sites

I performed some more tests, under clocking the card. something came to light while i was testing... The CPU fans were never spinning up, and the rMBP was getting very warm, so I installed fan control for bootcamp, and cranked up the fans... It got a little more stable...

I managed to get thru the full test at 850mhz, one time, then crash, same at 800mhz, and again at 775mhz... It only seems to like the test once, before crashing. will not run a second time. this is a little more positive, based on the fact that it wouldn't get but 2 or 3 seconds into the physics test... now it completes the test once...

and on the second time thru, it's 10-15 seconds into the physics test before it crashes.

however... I'm still stuck with Yosemite... just won't boot... I could create a provisional guide on this, but I certainly wouldn't call it successful yet.

--BrayNShocK

- - - Updated - - -

@goalque,

Thanks for the feedback...

It's a XFX R9 280X 3GB Double Dissipation...

When you say "low power consumption", do you mean when it's being used at low power, or do you mean, not getting enough power under high usage?

and as a follow-up, if i understand, the stability issues could be with the ATiKiO board, if i switch to Nestor, It would be rock solid?

I'm going to start another thread as to not high-jack this one, but I need help getting OS X to boot with this 280x... super irritating. I've gotten 2 other AMD cards to boot, but no acceleration, and 2 nvidia cards to boot (full CUDA) but no external video (2014 iris pro MBP)... but this stupid card refuses to let OS X boot.

--BrayNShocK

Link to comment
Share on other sites

I performed some more tests, under clocking the card. something came to light while i was testing... The CPU fans were never spinning up, and the rMBP was getting very warm, so I installed fan control for bootcamp, and cranked up the fans... It got a little more stable...

I managed to get thru the full test at 850mhz, one time, then crash, same at 800mhz, and again at 775mhz... It only seems to like the test once, before crashing. will not run a second time. this is a little more positive, based on the fact that it wouldn't get but 2 or 3 seconds into the physics test... now it completes the test once...

and on the second time thru, it's 10-15 seconds into the physics test before it crashes.

however... I'm still stuck with Yosemite... just won't boot... I could create a provisional guide on this, but I certainly wouldn't call it successful yet.

--BrayNShocK

- - - Updated - - -

@goalque,

Thanks for the feedback...

It's a XFX R9 280X 3GB Double Dissipation...

When you say "low power consumption", do you mean when it's being used at low power, or do you mean, not getting enough power under high usage?

and as a follow-up, if i understand, the stability issues could be with the ATiKiO board, if i switch to Nestor, It would be rock solid?

I'm going to start another thread as to not high-jack this one, but I need help getting OS X to boot with this 280x... super irritating. I've gotten 2 other AMD cards to boot, but no acceleration, and 2 nvidia cards to boot (full CUDA) but no external video (2014 iris pro MBP)... but this stupid card refuses to let OS X boot.

--BrayNShocK

Thanks for the input. My two Asus R9s behaved exactly the same. The first run of physics will pass, but the second will most likely give white screen and crash. I meant the situations when the GPU was not stressed at all. If you happen to own an energy meter, you will see that the physics part doesn’t eat many watts from eGPU, the CPU does the calculations.

280X should be compatible with the Yosemite, please follow my instructions:

http://forum.techinferno.com/implementation-guides-apple/8585-2014-15-macbook-pro-2x-r9_280x%4016gbps-tb2-netstor-na211tb-win8-1-osx-%5Bgoalque%5D.html

I haven’t updated the post, so the cursor positions may be incorrect. The missing information is that you may need sudo nvram boot-args="kext-dev-mode=1”.

The same “low power” issues are lurking there as well. I have came into conclusion that AMD R9 stability issues are due to AKiTiO only.

If still won’t boot, see the system logs.

Link to comment
Share on other sites

Thanks for the input. My two Asus R9s behaved exactly the same. The first run of physics will pass, but the second will most likely give white screen and crash. I meant the situations when the GPU was not stressed at all. If you happen to own an energy meter, you will see that the physics part doesn’t eat many watts.

..

The same “low power” issues are lurking there as well. I have came into conclusion that AMD R9 stability issues are due to AKiTiO only.

The AKiTiO has just has two filtering caps - C1 (100uF/16V) and C2 (10uF/25V). Can you study the DC input capacitor arrangement of your Netstor NA211TB to see what they are doing? Wonder if those caps are inappropriate for AMD card use?

Link to comment
Share on other sites

Hi all...

Let me start by saying, I'm not a gamer... like my experience with gaming is BlackOps 2 with my son on the PS3... (when he was in high school, a few years ago), or PC gaming, I played Quake 3 arena, and got talked into Tribes for a while... anyway... Not a Gamer...

I set out on this eGPU quest, because when i got rid of my Mac Pro2,1 I decided to build a hackintosh... After a couple years of on Ivy bridge, I upgraded to Haswell, and the futzing around with the nuances of getting sleep working after upgrading to Yosemite, i gave up and bought the rMBP.

So I'm used to 3 1080p monitors, and using every port on this mac to get 3 monitors, external gigabit, wireless key/mouse, external raid array, speakers, time capsule, printer, etc... I decided, ok, I need a way to dock this thing, and allow portability... Also I want to offload the video for audio/video projects to an external GPU instead of heating up the macbook.

Long intro, but wanted you to have the background.

Out of the gate i determined GTX750Ti was the way to go, no external PSU needed, if i upgrade the power brick for the AKiTiO...

Got it working in Yosemite, with very little fuss, only one problem (which i found after i set it up) -- NO external video for Iris Only rMBP...

So I set out on a quest to get similar AMD card... Tried the following cards:

XFX R7 250X 2GB

Saphire R7 260 2GB

Saphire R7 270 2GB

XFX R9 280 2GB

XFX R9 280X 3GB

The only one to completely boot in Yosemite was the 260, but it had no acceleration... the rest will not finish boot cycle once I either added their Device Ids to the appropriate section of X3000 or X4000 kexts and added the PCI Tunneling to the 7000 or 8000 kext files.

So since I know that the R9 280X is a rebranded 7970 and the Hackintosh community has had great success with it, I am planning to stick with it for now...

Setup...

mid-2014 15" Retina MBP Iris Pro 16GB, 256GB SSD

AKiTiO TB2 Enclosure that I have modded in the following ways:

  • Corsair CS650M ATX power supply (474 watt 12V rail)
  • soldered power to PCIe slot from main power (per Tech Inferno Fan's guide)
  • taped off pins 1-6 on thunderbolt to PCIe connector
  • I added C3 and C4 capacitors (same guide)
  • I added soldered connections ( 2 x 18gauge wires 2 yellow/2 black) to the ATX power supply (CPU) for the main power connector to ATiKiO.
  • Removed PCB from chassis, naked on the desk...

XFX R9 280X DD 3GB (rebranded AMD 7970 (clocked at 850Mhz / can be boosted to 1000Mhz))

I tired kext edits in both Mavericks and Yosemite, before i add the IOPCITunnelCompatible line to the AMD7000 kext, the computer boots and the other AMD kexts are loaded... and shows the device in System Profiler, as soon as I add the kext change to AMD7000, it will not finish booting.

So, I decided to install windows to test the hardware setup under bootcamp. installed win8.1 on a partition, and card boots like a champ.

It does fail to complete 3DMark11 Physics test, unless I underclock it. If I under clock the CPU, it will complete the benchmark test, once, upon second attempt, it will fail the physics test.

Now keeping in mind, I'm not a gamer, I don't really care about 3d benchmarks personally... but... i do want the offloaded graphics...

So I would like to get Yosemite working.

Here's what I did to fresh Yosemite install:

added IOPCITunnelCompatible to IONDRVsupport, AMD7000Controller, AMDSupport, AppleHDAControler

boot args = kext-dev-mode=1 -v (even tried dart=0, nv_disable=1)

no peripherals attached (except the USB SSD with fresh Yosemite install).

Symptom:

gets thru loading the OS and the last step before it loads the GUI, then freezes.

It appears to me either an issue with Power Management kext or AMDFramebuffer kext, based on what the Hackintosh community has to say about getting this card to boot. Since we don't use Clover, i don't know the best way to inject frame buffer changes, or how to use some of those fixes.

I've thought about modding the ROM on the card (either using netkas tutorial or sending to MVC) adding AMD apple EFI to the ROM, but not sure that will do any good, or harm.

Any experience you all have that might help would be appreciated.

TIA

--BrayNShocK

Link to comment
Share on other sites

@goalque, I used your guide :)

One thing I noted, I think since you had probably done some work with Nvidia cards, you had already modded the IONDRVSupport kext, it was missing from your guide for the 280x, but system logs had the "missing IOPCITunnelCompatible" line for that kext, so I added them it there as well.

Understood on the low power situation, thanks for the explanation.

It looks like i either need to go on a quest to figure out how to make Nvidia cards produce external output on my rMBP or buy a Nestor to make my AMD stable. hmmm what to do, what to do :)

@Tech Inferno Fan, I did add C3 and C4 caps to my AKiTiO... didn't seem to make any change in stability for the AMD card. (Using the ATX PS as only power source and ditching the 120W power brick did however.)

  • Thumbs Up 1
Link to comment
Share on other sites

Since I'm only using the ATX PSU for all power, I'm just using the built-in power switch on the PSU. (of course I have the green wire bridged to permanently ground)

I'm wondering if I need to wipe my internal Yosemite install and start from scratch again...

Do you think using an apple EFI would help, or unnecessary, it's something else?

Link to comment
Share on other sites

The AKiTiO has just has two filtering caps - C1 (100uF/16V) and C2 (10uF/25V). Can you study the DC input capacitor arrangement of your Netstor NA211TB to see what they are doing? Wonder if those caps are inappropriate for AMD card use?

NA211TB’s PCIe board design is much complicated, the power is shared by 4 slots from a standard 24-pin ATX plug. There are at least twenty tiny capacitors beside the upstream SLOT4 PCIex8, TB card narrows the connection to x4. After them one bigger capacitor (54475/100/35V) and yellow one before the SLOT3 (PCIex8) where the main GPU can be attached. Cannot distinguish the markings of yellow capacitor without magnifier glass but it starts with 476A:

Free shipping SMD tantalum capacitors 476A 47UF 10V B type 3528 1210 AVX capacitor yellow bile polarity-in Capacitors from Electronic Components & Supplies on Aliexpress.com | Alibaba Group

The same yellow can be found near 12V of 24-pin, and there are a lot of chips that I have no idea.

I am running eGPUs in GEN2 + x8/x16 compatibility mode.

Link to comment
Share on other sites

Since I'm only using the ATX PSU for all power, I'm just using the built-in power switch on the PSU. (of course I have the green wire bridged to permanently ground)

I'm wondering if I need to wipe my internal Yosemite install and start from scratch again...

Do you think using an apple EFI would help, or unnecessary, it's something else?

That’s quite a bunch of AMDs! :D

Did you add IOPCITunnelCompatible true value for AMDTahitiGraphicsAccelerator block in AMDRadeonX4000.kext when tried 280X?

I have a plan to make the process fully automatic by using regular expression patterns in kext mods instead of XML structure looping or query.

No need for ROM flashing if you use AMDs. I have tested newer and older AMDs on Yosemite with the AKiTIO and Netstor, they work out of the box, giving 100% external screen output, blow away Nvidia in OpenCL performance. OpenGL is notably slower with AMDs. The only issue is R9 + AKiTiO combination in low power conditions, and it likely has nothing to do with your boot problem.

I made some progress with the Nvidia eGPU screen output, found the Iris Pro device ids from the frame buffer binary data. If I am able to accelerate the internal retina screen as well, I will let you know. For free and open source.

Send PM and I am happy to check if the kexts are ok. I have the same Mac.

Link to comment
Share on other sites

Thanks :)

Yes I added the IOPCITunnelCompatible true value to Tahiti section of X4000 kext.

The only two AMD cards that I've been able to boot are a lowly 6450 and the XFX R7 250X, but neither had acceleration. They ran unique heaven at 3-5 FPS.

The rest would not boot. Even adding device IDs and the IOCPITunnelCompatible=true to the kexts (each card of course required different edits).

My guess is I've done too many kext edits and something is causing the problem, so I'm running a backup, and I'll blow Yosemite away and try a fresh install on the internal SSD. Maybe I'll try clearing the PRAM and SMC reset first...

I'll keep at it, once I've done the fresh install, I'll send you the kexts so you can double check if I still have the problem.

Thanks,

--BrayNShocK

Link to comment
Share on other sites

Thanks :)

Yes I added the IOPCITunnelCompatible true value to Tahiti section of X4000 kext.

The only two AMD cards that I've been able to boot are a lowly 6450 and the XFX R7 250X, but neither had acceleration. They ran unique heaven at 3-5 FPS.

The rest would not boot. Even adding device IDs and the IOCPITunnelCompatible=true to the kexts (each card of course required different edits).

My guess is I've done too many kext edits and something is causing the problem, so I'm running a backup, and I'll blow Yosemite away and try a fresh install on the internal SSD. Maybe I'll try clearing the PRAM and SMC reset first...

I'll keep at it, once I've done the fresh install, I'll send you the kexts so you can double check if I still have the problem.

Thanks,

--BrayNShocK

Hmm.. the external SSD may have an influence. I just bought Samsung SSD T1 250GB, up to 450MB/s reading speeds via USB3, that’s amazing.

And yes, IONDRVSupport + AppleHDAController kext mods are recommended, but I remember that they were not necessary - even though system log showed missing IOPCITunnelCompatible rows for them. kext-dev-mode=1 only, no other boot arguments.

The device id for 280X is 0x67981002, should be ready in the kexts.

Good luck with the fresh install! By the way, some AMD vendors offer UEFI compatible VBIOS, I think that's the reason why my Asus AMDs show smaller boot screens icons in retina screen, but some other show the normal icon size.

Link to comment
Share on other sites

yeah, don't start guessing and adding things.

As goalque stated, the AMD cards are usually pretty easy. More then likely you have a buggered kext somewhere. I always create a folder called "Virgin 14D136" or whatever build it is and first I copy all of the kexts to be modded there. Then I duplicate that folder. That way if I bugger them up I can just install from the Virgin folder instead of a full re-install.

EFI on AMD cards won't help or hurt OS X boot. Adds ability to choose boot disk via "option" (alt) key.

We have found that Tahiti cards on 2014 Mini run at approx 2/3 speed. They get 22FPS in Uningine Valley Extreme HD preset while same card in cMP would get 35fps. Likely similar to the bug they have in 3,1 MP.

Very few AMD cards AREN'T in the kexts, so it is quite possible that you were adding to wrong section or wrong kext.

Link to comment
Share on other sites

I have examined AKiTiO’s AMD issues thoroughly. Soldering mod helps to run 3-4 physics in a row at best, but eventually the whole eGPU will crash in normal tasks such as browsing the web. The common factor is low power consumption. I tried with the default factory clocks.

The note about the CPU fans never ramping up made me think of our past issues with 7950/70 in 3,1 Mac Pro.

Reduced 7xxx / R9 Performance in OS X on 2008 Mac Pro 3,1

16 pages of fighting with them. The gist of it was that the cards stopped getting full speed in a 3,1 Mac Pro in some version of 10.9 IIRC. There was one update that fixed it but then the problem came back. The cards ran at 40-60% of speed that they would get in a 4,1 or 5,1 Mac Pro. The real slap in the face was that the earlier 1,1 and 2,1 Mac Pros weren't affected. So the Tahiti cards ran better and faster in a PCIE 1.0 machine than a 3,1 that had PCIE 2.0.

Eventually Netkas found a fix, and here is where it gets interesting considering your observation about CPU fan speeds remaining locked at a low speed.

The fix was to remove "intelCPU power management" kexts from the Extension folder. Many folks reported it to work, though I don't know if we knew exactly why.

Goalque, do you still have a R9 280x in a eGPU setup? Curious how they did in Uningine Valley.

When we tried in 2014 Mini the 7970 would have same speed issue as in 3,1. It ran Uningine Valley in Extreme Preset at around 22FPS. Should be around 35FPS.

So it has some power management issue. I will try with the nMP and see if it is same. I may try to run the Physics thing, though I haven't put any AMD drivers on my Windows installs and have been trying to avoid it. In brief testing in nMP I had same "error 12" issues as with Nvidia cards so I gave up.

bryan, netkas can write you modded kexts that enable the 750Ti to have display output in Yosemite.

Link to comment
Share on other sites

I re-installed Yosemite on primary drive.

Only changed 4 kexts:

AMDSupport

AMD7000Controller

AMDRadeonX4000

AppleHDAController

Then of course ran 'kextcache -system-caches'

It boots on the LCD, System Profiler shows the generic "Display" under graphics/displays, Device ID 6798 as it should.

Does not show it as an AMD 7xxx

Syslog shows "IONDRVFrameBuffer missing IOPCITunnelCompatible"

If I add IOPCITunnelCompatible to IONDRVSupport kext, it does not boot, I have to disconnect the AKiTiO boot.

Any suggestions?

post-36017-14495000179226_thumb.png

post-36017-1449500017994_thumb.png

post-36017-14495000180663_thumb.png

post-36017-14495000181866_thumb.png

post-36017-14495000182951_thumb.png

post-36017-1449500018409_thumb.png

--BrayNShocK

- - - Updated - - -

Link to comment
Share on other sites

I would guess bad mods?

Did you send your kexts to goalque to verify?

Do you have all pins powered on the R9? SWEX/Paperclip happening?

Kext dev mode definitely enabled? (nvram -p in terminal will tell you) A single PRAM reset disables it so important. But I guess those kexts wouldn't have loaded if not.

goalque is definitely the AMD pro around here, I can just verify that when I tried it was easy, breezy, covergirl.

Shouldn't be this difficult.

Link to comment
Share on other sites

Will zip them up shortly and send to goalque. I'm re-installing win8.1 to retest power mods Tech Inferno Fan requested in AKiTiO thread.

As far as I can tell I've got the power setup working. I mean it does run under windows (albeit unstable). And I'm watching win8 install running off the 280x card now :)

Must be something wrong with my kexts...

Link to comment
Share on other sites

I've seen some of the other forums mentioning that the Nestor box may actually be more stable, would that be better option for running AMD cards? or do you think it's really more of a question of TB2 and AMD? I didn't mind dropping $200 to try the AKiTiO, but $1k for the nestor is a little much for experimenting, you know? :)

The Sonnet III-D or Netstor boxes are ultra expensive. Furthermore, they are a multi-slot device that introduces a PCIe bridge to share the link. That introduces two issues:

- some latency, however small

- there have been reports of boot hangs due to that bridge by squinks and another user

Best to stick with single-slot devices. If can't hack the AKiTiO to get an AMD card working with 100% reliability, inc 3dmark11 physics test, then the next affordable devices to try would be one of these:

http://forum.techinferno.com/enclosures-adapters/9638-us%24340-highpoint-rocketstor-6361a-pcie-chassis-16gbps-tb2.html

http://forum.techinferno.com/enclosures-adapters/9476-%5Bupdated-%5D-%24329-sonnet-ee-sel-%5Bnew%5D-%24349-sonnet-ee-se-i-16gbps-tb2-discussion.html

Link to comment
Share on other sites

@braynshock , before replacing the enclosure with a more expensive one, I'd suggest we do a simple test to see if the filtering is the issue. The AMD card may be adding low frequency noise that isn't getting filtered which causes the TB board to fail.

Seeing you are handy with a soldering iron you may want to have a go?

To do the test, you'll need a true multi-rail PSU so can split off power say PCIe power and slot power on 12V1 and TB board on 12V2. Eg PSU with true dual-rails is a Corsair HX1000: Professional Series . This would be a workaround for suspected insufficient decoupling capacitors on the AKiTiO board.

The mod is similar to the 75W bypass mod. You apply the cellophane tape to isolate the TB board and PCIe slot. However, instead of soldering the two wires across in that manner you want to:

- solder a molex connector to 12V and GND at reverse side of the AKITiO PCIe slot 2V (B1, B2, B3, A2, A3) and GND (A4, B4, B7). You could get by with 2 yellow-12V and 2 black-GND wires.

- either add a barrel-to-CPU connector or solder a mating 4-pin (CPU) connector directly to where the barrel input 12V/GND lines are.

This way the molex (12V1) and CPU (12V2) are on independent rails.

Attach to your ATX PSU and do some 3dmark11 runs to see if that helps. Would really like to nut this AMD problem out.

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.