Jump to content

DIY eGPU Macbook experiences


oripash

Recommended Posts

2014 Mini requires special EFI card to show boot screens and display support in OSX with Nvidia cards.

For Windows I find the EFI isn't needed.

Here is a 3D Mark score with Titan-X

NVIDIA GTX TITAN X video card benchmark result - Intel Core i5 4350U,Apple Inc. Mac-35C5E08120C7EEAF

NVIDIA GTX TITAN X video card benchmark result - Intel Core i5 4350U,Apple Inc. Mac-35C5E08120C7EEAF

Strange 3dmark11 result there. My OCed GTX970 is beating your Titan-X when it should be ~25% faster per NVIDIA GeForce Titan X 12 GB Review | techPowerUp . When not overclocked my GTX970 was getting ~12.6k for 3dmark11.gpu which is still faster than your result. Consider:

[email protected] + Titan-X@16Gbps-TB2 (Mac Mini, Win8.1)

3dmark.FS-gpu=12545

3dmark11.gpu=12091

i7-4810MQ@4Ghz-4c + GTX970OC@16Gbps-TB2 (HP ZBook 17 G2, Win8.1)

3dmark.FS-gpu=11139

3dmark11.gpu=13934

Link to comment
Share on other sites

Yeah, not sure why. It is connected to a 4K 60Hz display so pushing lots of pixels. I'll try with a 24" Dell and see if that changes anything.

Also consider that this is entry level Mini with 1.4 GHZ CPU

Link to comment
Share on other sites

I have a question. If i have iGPU(Nvidia 650M) in my Macbook Pro Retina, how to make eGPU(Nvidia GTX970) primary to my macbook's display?

Pls try http://forum.techinferno.com/implementation-guides/8558-%5Bguide%5D-macbook-enabling-optimus-internal-lcd-mode.html and report in that thread if it works on your 2012 15" Macbook Pro.

Link to comment
Share on other sites

ok

- - - Updated - - -

I read it, but it's not what i want! i want my eGPU to use my internal macbook display. Support from this site says that i can't do that: http://translate.google.ru/translate?hl=en&sl=ru&u=http://bizon-pc.com/ru/faq-egpu#egpu3 I'll be very upset if they are true

Link to comment
Share on other sites

I read it, but it isn't what i'm looking for. I want these:

Using under Windows:

The video card can display an image on an external monitor or on the internal screen.

To operate on the internal screen output, the configuration of your Mac must be equipped with only a video card Intel (Intel HD 4000 and above, Intel Iris, Intel Iris Pro). If you have two video cards, this function will not work.

For output to an external monitor, any combination (Intel, AMD, NVIDIA, Intel + NVIDIA, Intel + AMD).

So it says that it can't do what i want if i have iGPU.

Link to comment
Share on other sites

Hello,

Would it be possible to use two Thunderbolt2 ports to increase the bandwidth of a eGPU? I'm using a 5k iMac and would love a bit of extra performance in Bootcamp for gaming.

Cheers

R

Can't say for sure about a iMac, but for a Macbook Pro it's not possible to get 32Gbps bandwidth across two 16Gbps-TB2 ports due to their bus layout as explained. It's more than likely that a iMac uses the same bus layout.

Link to comment
Share on other sites

Can't say for sure about a iMac, but for a Macbook Pro it's not possible to get 32Gbps bandwidth across two 16Gbps-TB2 ports due to their bus layout as explained. It's more than likely that a iMac uses the same bus layout.

Thanks for the quick reply! How would I check?

Even if you could get 32Gbps over two ports would it be very tricky to implement?

Link to comment
Share on other sites

Thanks for the quick reply! How would I check?

Even if you could get 32Gbps over two ports would it be very tricky to implement?

If you have 2x 16Gbps TB2 channels, then you'd use two enclosures to do it as has already been demonstrated. However, it's likely that your two ports are just one 16Gbps link in which case using a bridged TB enclosure gets the same bandwidth:

Using single-slot unbridged TB enclosures

http://forum.techinferno.com/implementation-guides/8579-2013-13-macbook-pro-2-x-titan_z%4016gbps-tb2-akitio-thunder2-osx10-10-%5Bnesone%5D.html#post116923

http://forum.techinferno.com/implementation-guides/7326-2013-13-macbook-pro-gtx780ti-sli%4016gbps-tb2-sonnet-ee-sel-win8-1-%5Bsquinks%5D.html#post100318

Using multi-slot, bridged TB enclosures

http://forum.techinferno.com/implementation-guides/8585-2014-15-macbook-pro-2x-r9_280x%4016gbps-tb2-netstor-na211tb-win8-1-osx-%5Bgoalque%5D.html#post117016

Link to comment
Share on other sites

If you have 2x 16Gbps TB2 channels, then you'd use two enclosures to do it as has already been demonstrated. However, it's likely that your two ports are just one 16Gbps link in which case using a bridged TB enclosure gets the same bandwidth:

Using single-slot unbridged TB enclosures

http://forum.techinferno.com/implementation-guides/8579-2013-13-macbook-pro-2-x-titan_z%4016gbps-tb2-akitio-thunder2-osx10-10-%5Bnesone%5D.html#post116923

http://forum.techinferno.com/implementation-guides/7326-2013-13-macbook-pro-gtx780ti-sli%4016gbps-tb2-sonnet-ee-sel-win8-1-%5Bsquinks%5D.html#post100318

Using multi-slot, bridged TB enclosures

http://forum.techinferno.com/implementation-guides/8585-2014-15-macbook-pro-2x-r9_280x%4016gbps-tb2-netstor-na211tb-win8-1-osx-%5Bgoalque%5D.html#post117016

The system claims each port is 20 Gbps

"Port: Status: No device connected

Link Status: 0x101

Speed: Up to 20 Gb/s x1

Current Link Width: 0x1

Receptacle: 1

Link Controller Firmware Version: 0.14.0

Port:

Status: No device connected

Link Status: 0x101

Speed: Up to 20 Gb/s x1

Current Link Width: 0x1

Receptacle: 2

Link Controller Firmware Version: 0.14.0"

Is there a terminal command to check if these share the same Thunderbolt chip?

Link to comment
Share on other sites

The system claims each port is 20 Gbps

"Port: Status: No device connected

Link Status: 0x101

Speed: Up to 20 Gb/s x1

Current Link Width: 0x1

Receptacle: 1

Link Controller Firmware Version: 0.14.0

Port:

Status: No device connected

Link Status: 0x101

Speed: Up to 20 Gb/s x1

Current Link Width: 0x1

Receptacle: 2

Link Controller Firmware Version: 0.14.0"

Is there a terminal command to check if these share the same Thunderbolt chip?

There is OS X version of lspci which prints very useful information about PCI buses and devices. As an example, here is my output in tree format (-t option). Both the NA211TB / GTX 780 and AKiTiO / GTX 980 plugged in MBP's separate TB2 ports. As you can see they share the same root.

post-28870-14494999821457_thumb.png

You can get even more information

07:00.0 System peripheral: Intel Corporation Unknown device 156c

Subsystem: Unknown device 2222:1111

Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+

Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <tabort- <mabort-="">SERR-

Latency: 0, Cache Line Size: 128 bytes

Interrupt: pin A routed to IRQ 16

Region 0: Memory at a0b00000 (32-bit, non-prefetchable)

Region 1: Memory at a0b40000 (32-bit, non-prefetchable)

Capabilities: [80] Power Management version 3

Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold+)

Status: D0 PME-Enable- DSel=0 DScale=1 PME-

Capabilities: [88] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable+

Address: 00000000fee00000 Data: 407d

Capabilities: [c0] Express (v2) Endpoint, MSI 00

DevCap: MaxPayload 128 bytes, PhantFunc 0, Latency L0s <2us, L1 <4us

ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset-

DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-

RlxdOrd+ ExtTag+ PhantFunc- AuxPwr- NoSnoop+

MaxPayload 128 bytes, MaxReadReq 512 bytes

DevSta: CorrErr+ UncorrErr- FatalErr- UnsuppReq+ AuxPwr+ TransPend-

LnkCap: Port #0, Speed 2.5GT/s, Width x4, ASPM L0s L1, Latency L0 <2us, L1 <4us

ClockPM+ Suprise- LLActRep- BwNot-

LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- CommClk+

ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-

LnkSta: Speed 2.5GT/s, Width x4, TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt-

Capabilities: [a0] MSI-X: Enable- Mask- TabSize=16

Vector table: BAR=1 offset=00000000

PBA: BAR=1 offset=00000fa0

Capabilities: [100] #8086

0a:00.0 VGA compatible controller: nVidia Corporation Unknown device 13c0 (rev a1) (prog-if 00 [VGA controller])

Subsystem: nVidia Corporation Unknown device 1116

Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+

Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <tabort- <mabort-="">SERR-

Latency: 0

Interrupt: pin A routed to IRQ 18

Region 0: Memory at c0000000 (32-bit, non-prefetchable)

Region 1: Memory at <ignored> (64-bit, prefetchable)</ignored>

Region 3: Memory at <ignored> (64-bit, prefetchable)</ignored>

Region 5: I/O ports at 8000 [disabled]

Capabilities: [60] Power Management version 3

Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)

Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Capabilities: [68] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable+

Address: 00000000fee00000 Data: 4086

Capabilities: [78] Express (v2) Legacy Endpoint, MSI 00

DevCap: MaxPayload 256 bytes, PhantFunc 0, Latency L0s unlimited, L1 <64us

ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset-

DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-

RlxdOrd+ ExtTag+ PhantFunc- AuxPwr- NoSnoop+

MaxPayload 128 bytes, MaxReadReq 512 bytes

DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-

LnkCap: Port #0, Speed 5GT/s, Width x16, ASPM L0s L1, Latency L0 <512ns, L1 <4us

ClockPM+ Suprise- LLActRep- BwNot-

LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- CommClk-

ExtSynch- ClockPM+ AutWidDis- BWInt- AutBWInt-

LnkSta: Speed 2.5GT/s, Width x4, TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt-

Capabilities: [100] #10de

Capabilities: [13c] #112

c3:00.0 VGA compatible controller: nVidia Corporation Unknown device 1007 (rev a1) (prog-if 00 [VGA controller])

Subsystem: eVga.com. Corp. Unknown device 3787

Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+

Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <tabort- <mabort-="">SERR-

Latency: 0

Interrupt: pin A routed to IRQ 16

Region 0: Memory at a1000000 (32-bit, non-prefetchable)

Region 1: Memory at <ignored> (64-bit, prefetchable)</ignored>

Region 3: Memory at <ignored> (64-bit, prefetchable)</ignored>

Region 5: I/O ports at 2000 [disabled]

Capabilities: [60] Power Management version 3

Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)

Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Capabilities: [68] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable+

Address: 00000000fee00000 Data: 408b

Capabilities: [78] Express (v2) Endpoint, MSI 00

DevCap: MaxPayload 256 bytes, PhantFunc 0, Latency L0s unlimited, L1 <64us

ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset-

DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-

RlxdOrd+ ExtTag+ PhantFunc- AuxPwr- NoSnoop+

MaxPayload 128 bytes, MaxReadReq 512 bytes

DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-

LnkCap: Port #4, Speed 5GT/s, Width x16, ASPM unknown, Latency L0 <512ns, L1 <4us

ClockPM+ Suprise- LLActRep- BwNot-

LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- CommClk-

ExtSynch- ClockPM+ AutWidDis- BWInt- AutBWInt-

LnkSta: Speed 2.5GT/s, Width x8, TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt-

Capabilities: [100] #10de

Capabilities: [100] <chain looped=""></chain></tabort-></tabort-></tabort->

Link to comment
Share on other sites

Thanks for your reply! This is very interesting:

With the modification of the web drivers would the adapter be able to power a 750Ti without any other modifications? Can you link me to a working example of one on the forum?

Link to comment
Share on other sites

... would the adapter be able to power a 750Ti without any other modifications? Can you link me to a working example of one on the forum?

This is about the closest to your request: http://forum.techinferno.com/implementation-guides/8569-2011-11-macbook-air-gtx750ti%4010gbps-tb1-akitio-thunder2-osx10-10-%5Bjohndshea%5D.html#post116853 .

Only recently we had the AKiTiO 25W slot power specification demystified. Prior to we had many users reverting to powered risers to increase power, as that link shows, which actually degrade the PCIe signal. Maybe you can be the first to show use of an 120W uprated AC adapter powering a GTX750Ti?

Can see other AKITiO implementations at http://forum.techinferno.com/diy-e-gpu-projects/6578-implementations-hub-tb-ec-mpcie.html#Thunderbolt

Link to comment
Share on other sites

Maybe you can be the first to show use of an 120W uprated AC adapter powering a GTX750Ti?

I'd sincerely like to be, but i'm concerned by this discussion: Can my power supply support a GTX 750 ti? [solved] - Graphics Cards - Graphics Cards

As I understand the user pcgaming98 is saying that 12V railed power supplies (like the one you linked on eBay) do not provide enough Amps to run a GPU. What do you think about this?

Yes, with the additional 120W PSU you can run a 750Ti that comes without an additional 6-Pin PCIe Power connector.

Are you sure there wouldn't be amp problems like the forum post on Toms hardware says?

Also the GTX750Ti apparently requires 300W though it says that it only draws 60W. Can you clarify this for me? It's all quite new!

Interesting that user johndshea almost got it working in the thread you linked though.

Link to comment
Share on other sites

I'd sincerely like to be, but i'm concerned by this discussion: Can my power supply support a GTX 750 ti? [solved] - Graphics Cards - Graphics Cards

As I understand the user pcgaming98 is saying that 12V railed power supplies (like the one you linked on eBay) do not provide enough Amps to run a GPU. What do you think about this?

Interesting that user johndshea almost got it working in the thread you linked though.

GTX750Ti was measured to draw 66W at full load at NVIDIA GeForce GTX 750 Ti 2 GB Review | techPowerUp . The AKiTiO needs about 10W to drive the TB circuitry,so you're looking at 76W peak load

johndshea would have had instability due to overdriving the supplied 60W AC adapter.

The previously linked 120W AC adapter will do the job so long as it delivers on it's power specs.

Link to comment
Share on other sites

Can you clarify on the 300W 60W confusion that I edited into my previous post please?

Thanks for all the info by the way!

Are you sure there wouldn't be Ampage issues though? Apologies if its a stupid question.

Those 300W figures are powering a whole system so make little sense in our case where we are just powering the video card + TB board. I've already specified a cost effective TB2 enclosure + 120W AC adapter to power a GTX750TI (without PCIe power connectors). No power or amperage issue so long as the 120W AC adapter is a quality item delivering upon it's specs.

Link to comment
Share on other sites

The card is specs are 300W to power a whole PC with that card and it's rated to 300W to be able to provide enough power for the card on the 12V rail.

As long as you don't overclock the card way too much, you will never exeed the 75W of the slot by any time (maybe short peaks). So the 120W PSU is just a very safe way to go ;)

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.