Enclosure & Hardware Discussions
Node Pro as eGPU....out of luck?
 

Node Pro as eGPU....out of luck?  

  RSS

rickey
(@rickey)
Active Member
Joined: 2 years ago
 

I'm onto build two and think I made a mistake.  2017 MBP 15" with a previously working Titan Xp in an Node box. This worked. I upgraded to the Titan V to increase the deep learning potential. I purchased the Node Pro for the larger power supply. At that time, I didn't notice the subtle difference in eGPU vs. PCIe boxes. 

After a little setback for drivers in Sierra, I moved up to high sierra. The system is partially working. The about system page doesn't show the GPU. System report shows the "Node Pro" on the TB chain at a 40gb/s bandwidth. There is no GPU under "Graphics/Displays", however, my CUDA app shows that a GPU is found. It shows CUDA driver 387.99 and GPU driver 378.10.10.10.25.104. So, the system is seeing the card, but not as a GPU I guess.

So, this brings me to the question. Is this as simply as the wrong box?  I will swap cards back to the old Node to see if it works, but ultimately, I would like both to work so my team can have an extra deep learning box. 

Or, is there a simple fix to have the GPU be a GPU in the PCIe box?

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
rickey
(@rickey)
Active Member
Joined: 2 years ago
 

Update: 

I switched my Titan Xp from the Node to the Node Pro.  I did a log off, attached TB3 cable (wait for screen flash) and logged on. The Titan Xp was detected without issue in the Node Pro.  I then installed the Titan V into the Node, and repeated the process. The screen never flashed, so I'm thinking the issue is with the webdriver version. The CUDA app detects the card but it isn't readily visible to the OS.  

Off to see what NVIDIA can do to help.  And also rebuild Tensorflow because in the process of all of this, I moved to CUDA 9.1 and cuDNN v7.1...i.e., broke my Tensforflow build. 

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


itsage liked
ReplyQuote
itsage
(@itsage)
Famed Member Admin
Joined: 3 years ago
 

There are differences in the TBT firmware of an eGPU enclosure vs. PCIe enclosure. While we can install a graphics card inside the Node Pro, AKiTiO promotes this enclosure as a PCIe expansion chassis with power delivery and DisplayPort outputs. We can use it whichever way we see fit but tech support for eGPU application is the cutoff.

Your issue doesn’t seem to be with the Node Pro. Nvidia web drivers may be the culprit due to how new the Titan V is.

Best ultrabooks for eGPU use

eGPU enclosure buying guide


ReplyQuote
Flint Ironstag
(@flint-ironstag)
Estimable Member
Joined: 3 years ago
 

I like this enclosure as well.  I would bet an updated driver unlocks the power of your card.  Be sure to let us know if Titan V is  a worthy upgrade from the Xp.

MP 6,1 | 4c | d700
MP 6,1 | 6c | d500


ReplyQuote
rickey
(@rickey)
Active Member
Joined: 2 years ago
 

NVIDIA is still trying to figure out the driver issue. They think it might be the volta architecture that is throwing things off for the mac.  The node pro is about 1000x quieter out of the box than the regular node, so I'm hopeful this will be a viable solution.  I'm also impressed at how less "hacky" the egpu feels in 10.13.3.  It feels like we are almost to a fully supported mode!

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
Flint Ironstag
(@flint-ironstag)
Estimable Member
Joined: 3 years ago
 

Glad to hear they're working on a fix.

MP 6,1 | 4c | d700
MP 6,1 | 6c | d500


ReplyQuote
(@wesley_beernaert)
New Member
Joined: 2 years ago
 

I can still just see NVIDIA web drivers (to be chosen above osx standard osx drivers) with my akitio node thunderbolt 3 to thunderbolt 2 adapter sine I'm using a mic 2015 macbook pro with iris 5200 pro and discrete amd R m 370 X . Works fine in bootcamp (using internal screen as of now since I don't have an external monitor (except one samsung 720 x 1033 tv upscaling to 1080 p and a full 1080p hd LG, but the pc only recognised it as an external generic p&p monitor so the resolutios aren't that great and actually half the fps fell off.. must have done something wrong, like morriring the screen instead of expansin it or so...than again , 1080p on the 50" up close doesn't look anything likee 1080p on your laptop..on the mac it' marverlous, the fog is dense (talking about alan wake now), no more aliasingn the lighting system is inredibe..... just to mention, there are still posibilities,  😉

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
Tico
 Tico
(@tico)
Eminent Member
Joined: 3 years ago
 

I'm thinking of getting a Titan V for my Sonnet Breakaway 550 (upgraded to 650 PS) to use with a iMacPro for a new deep learning project. I'm thinking the NVIDIA better with CUDA than the Vega 64 thats in the iMac Pro.  Currently I have a 2010 MacPro & GTX 1080.  It would be also used for other analyses through Matlab.
What problems if any are you having with a Titan V. Is there driver support for High Sierra
thanks

☀️θ∆ics🌎
MacBook Pro (Retina, 15-inch, Mid 2014),2.8 GHz Intel Core i7 16 GB NVIDIA GeForce GT 750M 2048 MB, Intel Iris Pro 1536 MB
MacBook Pro (Retina, 15-inch, 2018) I9, 2.9 GHZ 16gb Radeon Pro 560x 4 gb
Sonnet BA 650w (upgraded), Titan V & Sonnet BA 650w NVIDIA gtx 1080
Mantiz Venus RX VEGA 8 gb
iMacPro 64 gig. VEGA 54 using Sonnet BA


ReplyQuote
DANgerous25
(@danfalconar)
Estimable Member
Joined: 3 years ago
 

@tico: I have an iMac Pro. If you are happy using High Sierra 10.13.3 and eGPU (and the associated pitfalls, i.e. can't boot with it plugged in then have to plug in the log out/in) then you'll be fine - CUDA works and so does external display if you want it. I haven't managed to get my 1080TI working with 10.3.4 b2 for either CUDA or external display, the OS just doesn't detect the card. 10.13.4 b2 is good for AMD cards. I'm using a Vega FE LC.

late-2017 27" iMac Pro RPVega64 + [email protected] (AKiTiO Node) + MacOS 10.13.2
late-2014 27" iMac 5k R9 M295X + [email protected] (AKiTiO Node via TB3->TB2 adapter) + macOS10.13 & Win10


ReplyQuote
psonice
(@psonice)
Estimable Member
Joined: 2 years ago
 

@tico: AFAIK there are no drivers for the Titan V yet. That said, I'm using a 1080 TI for deep learning (using CUDA / Tensorflow), it's a massive speedup already. Roughly 10-25x faster for the things I've been doing! It's literally the difference between "Start training, come back in an hour" and "Start, and by the time you've tweaked the model ready for the next experiment, it's already done" 🙂

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


itsage liked
ReplyQuote
rickey
(@rickey)
Active Member
Joined: 2 years ago
 

As noted above, the lack of drivers for the Titan V is the problem. UNIX is up to CUDA 390.*. The mac isn't there yet. NVIDIA won't/can't comment on when the drivers will be available.

For deep learning, the eGPU works great. I have a titan xp that hums right along. You have to build tensorflow from sources. That is a fussy, time consuming process, but it can be done. 

This is a video demonstrating the speed on a relatively complex model.

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
Tico
 Tico
(@tico)
Eminent Member
Joined: 3 years ago
 

Thanks for the video. The group we are working with are involved in breast cancer detection rates using deep learning vs a traditional radiologist.  They are using 60k images & finding higher rates with deep learning. I'm just starting to apply these techniques to my research and see a lot of potential. We will be doing the work using super computer centers. I want to be able to explore & tweak the models using limited datasets. I think we are at the start of an exponential growth curve in the use of these techniques in medical imaging particularly now with almost entirely using digital film images. Access to relatively cheap & scalable GPU and storage capabilities makes a huge difference. Its light years away from the mid 80's when we were working with NMRI images.

☀️θ∆ics🌎
MacBook Pro (Retina, 15-inch, Mid 2014),2.8 GHz Intel Core i7 16 GB NVIDIA GeForce GT 750M 2048 MB, Intel Iris Pro 1536 MB
MacBook Pro (Retina, 15-inch, 2018) I9, 2.9 GHZ 16gb Radeon Pro 560x 4 gb
Sonnet BA 650w (upgraded), Titan V & Sonnet BA 650w NVIDIA gtx 1080
Mantiz Venus RX VEGA 8 gb
iMacPro 64 gig. VEGA 54 using Sonnet BA


ReplyQuote
rickey
(@rickey)
Active Member
Joined: 2 years ago
 

That's where we are going with our research and why I built a couple of eGPU systems. We have some IT requirements that make bringing up a rogue UNIX box hard. We have a couple of enterprise GPU clusters but there's much work that needs to be accomplished before we turn the training lose on those.

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
DANgerous25
(@danfalconar)
Estimable Member
Joined: 3 years ago
 

@Tico: What a great project! Oh how things have come on since my day... I remember writing my own neural net and learning system in Java for skin cancer diagnosis. No fancy software or hardware for the job, just pure code! But the input dataset was just a bunch of 1s and 0s for 17 different features from a laser, far simpler than an image like you're doing. I had to let the thing run overnight and hope that it was doing the right thing. Needless to say it didn't until the end, meaning it took a llllooonnngggg time to get it working properly!

Good luck with your application.

late-2017 27" iMac Pro RPVega64 + [email protected] (AKiTiO Node) + MacOS 10.13.2
late-2014 27" iMac 5k R9 M295X + [email protected] (AKiTiO Node via TB3->TB2 adapter) + macOS10.13 & Win10


itsage liked
ReplyQuote
Flint Ironstag
(@flint-ironstag)
Estimable Member
Joined: 3 years ago
 

Great discussion here.  I've always felt as though eGPU is going to be a major component of ushering in a new age of discovery.  All that power sitting under so many clever peoples' desks.  It's exciting.

MP 6,1 | 4c | d700
MP 6,1 | 6c | d500


itsage liked
ReplyQuote
itsage
(@itsage)
Famed Member Admin
Joined: 3 years ago
 

Absolutely amazing to learn how people are making use of eGPU.

Best ultrabooks for eGPU use

eGPU enclosure buying guide


ReplyQuote
cmm
 cmm
(@mois)
Trusted Member
Joined: 2 years ago
 
Posted by: itsage

There are differences in the TBT firmware of an eGPU enclosure vs. PCIe enclosure. While we can install a graphics card inside the Node Pro, ..

Which are those differences?

(This is also the topic of this thread,  https://www.reddit.com/r/eGPU/comments/92ndul/do_egpu_boxes_no_daisychain_have_special_firmware/ .)

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote