PCIE Bifurcation

Hello everyone

I just wanted to say thanks to everyone who has been contributing to this thread, there is so much information in here that can't be found anywhere else. So I stumbled upon this thread earlier this week looking for this exact solution.

First, my rig is an Ncase m1,asrock X99e-itx,1070 mini, 6800k. I'm looking to downgrade my gpu and make my machine a dedicated video editing and streaming machine. I want to install the new XFX RX 460 slim single slot and pair it with an internal single slot capture card and move the 1070 to a new rig. The problem with the RX 460 slim is that it does not have any power connection.

My question is what powered riser card would you recommend for the job? I've looked online but couldn't find what looked like a powered riser card. If anyone has a similar setup let me know how well this works? I could easily purchase an external capture card but the less cable management the better.

-ODB

Edit: I see the RSC-R2UT-2E8R has a 6 pin cut out in the PCB however it's PCIE 2.0.

Edit2: I found that there are powered riser CABLES that i could use for what i need. I found some good info on the process HERE. I'm pairing it with the ARC1-PERY423-C5 PCIE 3.0. I'll post on my progress mid Feb.

Someone else posted looking for a single slot 6/8pin powered gpus. There is an upcoming gtx 1070 made by GALAX that is going to be single slot. There is also the MSI 1060 6gt OCV1 that can be modified to a single slot since all of its outputs are on the left side of the mounting bracket.
 
Last edited:
Thanks... looking again this now makes perfect sense to me.
Although pcie standard only allows for 2 connections, not 3.
PCIe gen4 will even only allow 1 connector afaik. So in the future those riser combinations may really run into signal problems... I even had some rare signal errors with the original Raven Riser.


Maybe this is a power delivery problem.... all those risers I have seen are not powered, essentially pulling twice the juice off the mainboard...


Nope, it's not a power delivery problem nor a signal problem - I am able to run caffe (or folding@home) under Linux for hours with both GTX1080 GPUs pegged at 100%, and the CPU too ;-). Am using the Amerirack PCie v3 version of their x8x8 riser - the only one that works as far as I can tell.
This particular set of components (x99e-itx + 2x1080s + same riser) has never worked for me under Windows. Am pretty sure it is a DMA/IRQ issue, and in fact I am able to replicate it under Linux if I enable the hardware IOMMU (Vt-d). Am giving it another shot with the ASRock Taiwan guys and this time I am feeling quite hopeful they'll finally get to the bottom of it. Seeing this today at AnandTech (http://www.anandtech.com/show/11078...and-nightblade-pcs-with-intels-kaby-lake-cpus) only makes me more confident about the viability of the whole exercise:
"It is interesting to note that instead of using a larger (say, microATX) motherboard with two PCIe 3.0 slots, in the case of the Aegis Ti3 MSI decided to use a Mini-ITX mainboard and then use a riser card with a PCIe switch to bifurcate the only PCIe 3.0 x16 slot between two cards."
 
Nope, it's not a power delivery problem nor a signal problem - I am able to run caffe (or folding@home) under Linux for hours with both GTX1080 GPUs pegged at 100%, and the CPU too ;-). Am using the Amerirack PCie v3 version of their x8x8 riser - the only one that works as far as I can tell.
This particular set of components (x99e-itx + 2x1080s + same riser) has never worked for me under Windows. Am pretty sure it is a DMA/IRQ issue, and in fact I am able to replicate it under Linux if I enable the hardware IOMMU (Vt-d). Am giving it another shot with the ASRock Taiwan guys and this time I am feeling quite hopeful they'll finally get to the bottom of it. Seeing this today at AnandTech (http://www.anandtech.com/show/11078...and-nightblade-pcs-with-intels-kaby-lake-cpus) only makes me more confident about the viability of the whole exercise:
"It is interesting to note that instead of using a larger (say, microATX) motherboard with two PCIe 3.0 slots, in the case of the Aegis Ti3 MSI decided to use a Mini-ITX mainboard and then use a riser card with a PCIe switch to bifurcate the only PCIe 3.0 x16 slot between two cards."
Curiously, the mobo also has 4 sodium slots. Might be a custom board.
 
That's a good question. I'm using the RSC-R2UT-2E8R.

But this is a problem because this splitter's PCB is definitely too long to fit in the Dan A4.

Last I checked, the flexible Amerirack splitter, the only other one I've seen tested on this thread, was no longer being sold.

Do you know of another flexible (or low-profile) splitter?
Hi MaximumBurrito, the splitter u r using is RSC-R2UT-2E8R. Ur gpus running at PCI-E gen 2.0 or 3.0? I ask this cuz I saw that card only support PCI-E Gen 2.0 on the supermicro website.
 
Hi everyone!

Signed up because i've been working on this for a while now - Here are my results:

I am using the ASROCK X99E-ITX/ac with 3.63 bios, dual R9 Nanos and my DIY Riser specifically designed for the Raven RVZ-01 case.

My intent was to create a 2 Gamers 1 ITX System. (Virtualization) This is working spendidly.
So far I tried Battlefield 1 Singleplayer: 1440P Ultra 90FPS + 4K Medium 70FPS @ ~80% CPU.
(5820k @ 4.0GHz. Nanos Power-Limit+15%)
Crossfire is working as well. I didnt do much testing though.

My Riser:
I am using a IDT-9DBL02 Clock Buffer. I picked it because it has lower external component count and complies to PCIe Gen4
It is not powered, but the Asrock Mainboard seems to be able to provide enough juice to its PCIe Slot.
Maybe I will Design a powered one at one point.

I have two of them PCB's left because OSH Park always makes 3... so if anyone is interested...
I would also be willing to make the riser in a different height if someone would be interested...


Cheers!

Chris



View attachment 14807 View attachment 14808 View attachment 14809 View attachment 14810 View attachment 14812


Hi Chris, I got to say I'm very interested :D.
But I prefer a right turn, powered and dual slots for each pci-e slot riser card. (just can't find any single slot gtx1080 or 1070, and i realy don't want to mess up with those anoying DVI ports)
maybe I asked to much. >.<
Btw, nice work.
 
Nope, it's not a power delivery problem nor a signal problem - I am able to run caffe (or folding@home) under Linux for hours with both GTX1080 GPUs pegged at 100%, and the CPU too ;-). Am using the Amerirack PCie v3 version of their x8x8 riser - the only one that works as far as I can tell.
This particular set of components (x99e-itx + 2x1080s + same riser) has never worked for me under Windows. Am pretty sure it is a DMA/IRQ issue, and in fact I am able to replicate it under Linux if I enable the hardware IOMMU (Vt-d). Am giving it another shot with the ASRock Taiwan guys and this time I am feeling quite hopeful they'll finally get to the bottom of it. Seeing this today at AnandTech (http://www.anandtech.com/show/11078...and-nightblade-pcs-with-intels-kaby-lake-cpus) only makes me more confident about the viability of the whole exercise:
"It is interesting to note that instead of using a larger (say, microATX) motherboard with two PCIe 3.0 slots, in the case of the Aegis Ti3 MSI decided to use a Mini-ITX mainboard and then use a riser card with a PCIe switch to bifurcate the only PCIe 3.0 x16 slot between two cards."
Interesting. I wonder if the Mobo can be purchased separately
 
Hi Chris, I got to say I'm very interested :D.
But I prefer a right turn, powered and dual slots for each pci-e slot riser card. (just can't find any single slot gtx1080 or 1070, and i realy don't want to mess up with those anoying DVI ports)
maybe I asked to much. >.<
Btw, nice work.


Sorry i've been offline...

I have already added Power Pads to my design.
I consider building a right-angled one myself, however single slot... (so my dual gpus rest just above my motherboard/slim CPU waterblock)
https://www.ekwb.com/shop/ek-annihilator
My System will be so compact.:D:D:D:D:D:D

I think I will start next week. I would share my design with you or even solder one for ya. PCB are pricey though... I paid around 100 dollars for 3, and if you need more surface area they will get even more expensive...

BTW: Please do mess with those DVI Ports! It is great fun!:D
I would just solder the dual-dvi one out instead of using a dremel.... so much cleaner. Maybe one could even put in a single slot one afterwards.
Too bad EK doesnt make a waterblock for the Gigabyte ITX 1070...

Cheers
 
Sorry i've been offline...

I have already added Power Pads to my design.
I consider building a right-angled one myself, however single slot... (so my dual gpus rest just above my motherboard/slim CPU waterblock)
https://www.ekwb.com/shop/ek-annihilator
My System will be so compact.:D:D:D:D:D:D

I think I will start next week. I would share my design with you or even solder one for ya. PCB are pricey though... I paid around 100 dollars for 3, and if you need more surface area they will get even more expensive...

BTW: Please do mess with those DVI Ports! It is great fun!:D
I would just solder the dual-dvi one out instead of using a dremel.... so much cleaner. Maybe one could even put in a single slot one afterwards.
Too bad EK doesnt make a waterblock for the Gigabyte ITX 1070...

Cheers
Well thank u.
About the waterblock for Ga 1070 itx, I bought one. It is not as good as EK, since it is the only one I can find made for GA1070 itx, I....can live with that.
And Zotac just released GTX1080 itx ( PCB:177mm ), EK won't make a water block for that neither. Looks like they are not into making a waterblock for nvidia nano size graphics card. LOL
 
About the waterblock for Ga 1070 itx, I bought one. It is not as good as EK, since it is the only one I can find made for GA1070 itx, I....can live with that.

Which one is it then? Couldnt find one... is it single slot?
 
Which one is it then? Couldnt find one... is it single slot?
Single slot yes, but I don't know is it available in your area.
I'll PM you the brand and link, but I want share some pro&con with u first.
1. The quality of this waterblock is ……, well it only cost me 73 USD, what else can i say. >.< (I don't know the price u gona get)
2. Again, the brand is not attention to detail type, they usually make waterblock for everysingle graphics card they can get hands on. So this would be a good choice when u want a full cover waterblock for like GA1070 ITX or Zotac 1080 mini.
3. According to the owner, "the engineering sample (GA 1070 ITX) we use has a little difference." so what happened next is i have to dremel a little bit off the copper cuz it interfere with the fan header on the 1070.
4. U need to tell them cut the extra acrylic. U may ask "extra what?". Well, don't ask me, i don't know why a 170mm card has a 230mm(maybe?) long waterblock neither.
extra.jpg

5. at last, GA1070 ITX is really not a good choice if u try put two of this on your single slot bifurcation card, especially with acrylic top waterblock. Because the little backplate is really really hot.
 
Hi Chris, I got to say I'm very interested :D.
But I prefer a right turn, powered and dual slots for each pci-e slot riser card. (just can't find any single slot gtx1080 or 1070, and i realy don't want to mess up with those anoying DVI ports)
maybe I asked to much. >.<
Btw, nice work.

Riser.png


Done with my right turn Design.
I havent figured out the Edge-Connector to Slot 1 Dimension I will be going for yet.
fishell u interested? what dimensions would you need?
I could adapt it to your dimesions in no time and send you the .brd file.
 
Nope, it's not a power delivery problem nor a signal problem - I am able to run caffe (or folding@home) under Linux for hours with both GTX1080 GPUs pegged at 100%, and the CPU too ;-). Am using the Amerirack PCie v3 version of their x8x8 riser - the only one that works as far as I can tell.
This particular set of components (x99e-itx + 2x1080s + same riser) has never worked for me under Windows. Am pretty sure it is a DMA/IRQ issue, and in fact I am able to replicate it under Linux if I enable the hardware IOMMU (Vt-d). Am giving it another shot with the ASRock Taiwan guys and this time I am feeling quite hopeful they'll finally get to the bottom of it. Seeing this today at AnandTech (http://www.anandtech.com/show/11078...and-nightblade-pcs-with-intels-kaby-lake-cpus) only makes me more confident about the viability of the whole exercise:
"It is interesting to note that instead of using a larger (say, microATX) motherboard with two PCIe 3.0 slots, in the case of the Aegis Ti3 MSI decided to use a Mini-ITX mainboard and then use a riser card with a PCIe switch to bifurcate the only PCIe 3.0 x16 slot between two cards."

Update: I got a new UEFI from ASRock (3.64L) which finally allows both GTX1080's to work under Windows 10! I have been using the system for 2 days now and appears to be rock solid. Now solving the Vt-d issue under Linux is all that remains - I posted on the NVidia Linux driver forum, let's see if they are as helpful..
 
Update: I got a new UEFI from ASRock (3.64L) which finally allows both GTX1080's to work under Windows 10! I have been using the system for 2 days now and appears to be rock solid. Now solving the Vt-d issue under Linux is all that remains - I posted on the NVidia Linux driver forum, let's see if they are as helpful..
Could you tell me who you contact? I tried to contact the tech support on their website 20days before, but no one replied.:(
 
Update: I got a new UEFI from ASRock (3.64L) which finally allows both GTX1080's to work under Windows 10! I have been using the system for 2 days now and appears to be rock solid. Now solving the Vt-d issue under Linux is all that remains - I posted on the NVidia Linux driver forum, let's see if they are as helpful..

Would you mind sharing it? Or are you not supposed to?

Hi, Payne thanks, looks great! (Drooling........)

I am working on a third Design so I can put them on a big Panel and maybe safe some $$$.
An order with your size would be around 180$ on OSH-Park. Same goes for my Dimensions.
Be Patient! ;)


Do you speak the language?... I am having trouble navigating that website.
Maybe you can help me with an order there.

What kindof temps are you getting? That looks like a lot of hardware on a single 120mm radiator, if that's what you're using!

I went crazy and desoldered the CPU IHS. Now with my CPU Block directly on Die I get the following Temps:
i7 5820k @ 4.2 Ghz 1.2V
R9 Nanos Power-Targets each +20%

2x furmark, 2x prime95:
water= ~70 Celsius
GPU's= ~85 Celsius
CPU = ~100 Celsius
FAN at 90%

2x Battlefield 1 Singleplayer:
water= ~60 Celsius
GPU=~70 Celsius
CPU=~85 Celsius
FAN at 75%

20170211_150620.jpg
 
I am working on a third Design so I can put them on a big Panel and maybe safe some $$$.
An order with your size would be around 180$ on OSH-Park. Same goes for my Dimensions.
Be Patient! ;)


Do you speak the language?... I am having trouble navigating that website.
Maybe you can help me with an order there.


View attachment 17434
Did you receive the message I sent to you?
 
Would you mind sharing it? Or are you not supposed to?

Do you speak the language?... I am having trouble navigating that website.
Maybe you can help me with an order there.

View attachment 17434

Well, I do speak the language. But I think u can try to reach the youtube channel "Declassified Systems" first, their builds use that brand waterblock a lot. I think it would be a good idea that ask that channel where to buy, cuz it will cost a lot ship from China to Europe or USA,but I'll try to help u if u need.

I am working on a third Design so I can put them on a big Panel and maybe safe some $$$.
An order with your size would be around 180$ on OSH-Park. Same goes for my Dimensions.
Be Patient!
wow wow w8 a minute, I'll pm u later.
 
New gigabyte Z270 board announced with specific mention of PCIE Birfurcation ability on the product page.

(Link)

Pretty interesting... Of course you'd still have to find the right riser board for it though.
 
Just caught up with the thread after a bit of a break.

Seems there are a number of people with this working now and as I understand it there are two options for PCIe gen 3 support:

1) Bifurcation with ameri-rack ARC1-PELY423-C5V3 splitter, (using 3.64L bios for asrock board). This gives 1x16 -> 2x8 so half the bandwidth but with no added latency.
2) PLX chip with supermicro RSC-R2UG-A2E16-A splitter. This gives 1x16 ->2x16 so full bandwith for one card at a time but presumably has added latency from the PLX chip.

Also C Payne's custom splitters which presumably perform like option 1 above and MaximumBurrito's success with RSC-R2UT-2E8R which (assuming it's PCIe gen 2) probably has the worst performance.

Is it possible to get a fair performance comparison between the two options above to see which performs better in SLI?

I'm thinking the GTX 1080 Ti is looking good for this BTW :) no HDMI port so it's single slot capable without modification.
 
Seeing some folks here report that they were able to purchase from ameri-rack, but... no such luck for me. Boy am I pissed at them.

Ameri_rack_no_sale.png


Seems like flexible bifurcated risers are out of the question for me.

C_Payne : Do you have any of those PCB's remaining from your first run?

Thanks!
 
heb1001, yes, it is a PCI-E 2.0 splitter, and it didn't fit in the case anyway, so I replaced it with the Ameri-rack splitter.

psychic_quark, i was able to buy one after I talked with sales and reassured them that I wasn't going to steal their patent-pending design. They're just worried that someone else will produce their board, so you should be able to discuss that you're not going to reverse-engineer and re-sell it and you should be able to get it.
 
Ok, so it took some time (mostly waiting for the case to finally arrive), plus a hacksaw, a pair of garden shears, and two wine corks, but I was finally able to build my Dan A4 bifurcated dual Nano machine.

IMG_4902.JPG


IMG_4895.JPG


IMG_4893.JPG



This is small enough to fit behind my wife's 13" MacBook Air!

IMG_4899.JPG



Note that this is INCREDIBLY tight. It took multiple tries to get all the components into the case, and ended up requiring a very specific order of assembly to fit at all. Oh, and the Ameri-rack splitter had an edge that was running into the case, so it had to go.

IMG_4878.JPG


Aaaaand then the other side was running into the ethernet port housing on the Asrock board, so that had to go to.

IMG_4879.JPG


Uhhhh, and then it turned out that the other side still had too much board and was running into the RAM, so that's when the garden shears came out. The saw couldn't cut under the ends of the PCI-E slots like I wanted, but they could.

IMG_4882.JPG


Hey, no one has to see it! After that, it finally fit.

IMG_4884.JPG


So the moral of the story is that brutality can work when brains fail, I guess?

IMG_4887.JPG


Anyway, it works!

Capture.JPG

The skeptical reader may note that you could achieve a similar score with a single GTX 1080, but that's not as fun. More to the point, it doesn't quite work for my application: I'm developing some software where I want to be able to do graphics on one card and simultaneous computations on the other. Basically, this is a DIY Mac Pro.

This particular setup, with one card facing vertically, was jb1's idea all the way back in August 2015, so credit where credit's due.
 
Last edited:
heb1001, yes, it is a PCI-E 2.0 splitter, and it didn't fit in the case anyway, so I replaced it with the Ameri-rack splitter.

psychic_quark, i was able to buy one after I talked with sales and reassured them that I wasn't going to steal their patent-pending design. They're just worried that someone else will produce their board, so you should be able to discuss that you're not going to reverse-engineer and re-sell it and you should be able to get it.

Thanks for the tip, let me give this a shot ONE more time since i haven't heard from C_Payne yet.
 
This particular setup, with one card facing vertically, was jb1's idea all the way back in August 2015, so credit where credit's due.

Nice work getting this built!! I just received my A4 as well and was really tempted to try this idea, but having waited nearly 1.5 years I decided that I could wait a little longer and see what was around the corner with Vega-- I'm hoping that AMD produces an update to the R9 Nano.

Since you actually have this up and running, I've got a question for you

I purchased the same splitter as you did for this build, however, instead of getting one with a short length I got one that's 30cm long that would wrap around to the GPU side of the case. However, I'm concerned about crosstalk in riser since it's so long and each of the leads is not well isolated like the 3M one. Have you noticed any issues due to crosstalk in your shorter bifurcated riser? Which risers are you using to connect to the GPUs? It looks like one of them is the 3M one, but what is the black one?

Thanks!
 
Ok, so it took some time (mostly waiting for the case to finally arrive), plus a hacksaw, a pair of garden shears, and two wine corks, but I was finally able to build my Dan A4 bifurcated dual Nano machine.

View attachment 19310

View attachment 19311

View attachment 19312


This is small enough to fit behind my wife's 13" MacBook Air!

View attachment 19313


Note that this is INCREDIBLY tight. It took multiple tries to get all the components into the case, and ended up requiring a very specific order of assembly to fit at all. Oh, and the Ameri-rack splitter had an edge that was running into the case, so it had to go.

View attachment 19315

Aaaaand then the other side was running into the ethernet port housing on the Asrock board, so that had to go to.

View attachment 19316

Uhhhh, and then it turned out that the other side still had too much board and was running into the RAM, so that's when the garden shears came out. The saw couldn't cut under the ends of the PCI-E slots like I wanted, but they could.

View attachment 19317

Hey, no one has to see it! After that, it finally fit.

View attachment 19318

So the moral of the story is that brutality can work when brains fail, I guess?

View attachment 19319

Anyway, it works!

View attachment 19320
The skeptical reader may note that you could achieve a similar score with a single GTX 1080, but that's not as fun. More to the point, it doesn't quite work for my application: I'm developing some software where I want to be able to do graphics on one card and simultaneous computations on the other. Basically, this is a DIY Mac Pro.

This particular setup, with one card facing vertically, was jb1's idea all the way back in August 2015, so credit where credit's due.

You are INSANE!!!!!!!
This world need more insane people, like Dan Hensen and you!
Keep up the good work!
 
Thanks, couldn't have done it without the extensive development y'all put in!

Per jb1's comments:

Nice work getting this built!! I just received my A4 as well and was really tempted to try this idea, but having waited nearly 1.5 years I decided that I could wait a little longer and see what was around the corner with Vega-- I'm hoping that AMD produces an update to the R9 Nano.

Yeah I thought the same thing, but 1. there's no guarantee that AMD will issue an updated Nano (though I'd snap it up in a second if they did!), and 2. it would need to be absolutely NO bigger than the current one or there is no possible way it would fit in the case.

The wine corks propping up the bottom of the second card may give the impression that there is room under the card, but there really isn't: the 8-pin PCI-E power plug barely has room to clear the SSD bracket at the bottom of the case. Even if the bracket were removed completely, it wouldn't free up much space. Any future cards fitting into this case would really need to be Nano-sized or smaller!

I purchased the same splitter as you did for this build, however, instead of getting one with a short length I got one that's 30cm long that would wrap around to the GPU side of the case. However, I'm concerned about crosstalk in riser since it's so long and each of the leads is not well isolated like the 3M one. Have you noticed any issues due to crosstalk in your shorter bifurcated riser? Which risers are you using to connect to the GPUs? It looks like one of them is the 3M one, but what is the black one?

No issues with crosstalk. It's quite stable. I've ran Firestrike, Heaven, Valley etc. and the only thing that crashes it is Furmark, but I think that's an issue with the PCI-E bifurcated Crossfire setup: it just crashes on startup, not due to any instability.

The riser I used is a Lian Li PW-PCI-E38-1 380mm riser, which is rather expensive but seems to be quite high quality. The two things that ended up being crucial about this riser don't have to do with stability.

The card end of the riser is at an angle, which allowed me to thread it through to the other side of the case after cutting the GPU shroud provided with the case:

IMG_4896.JPG


and the cables of the riser are split into 4-lead chunks instead of bundled into a single cable, which allowed me to spread them out flat in order to close the top of the case:

IMG_4886.JPG


You can see the riser looping through from the other side of the case right next to the main 24-pin motherboard power connector.
 
During my BF1 testing CPU was sitting at around 80 Celsius and GPU's at around 60, with that Nocuta NF12 at 60% PWM.
That Fan packs a Punch and makes some serious Noise at 100%. At 60% it's quite OK... less then a stock nano.
Radiator is 45mm thick 120mm dont remember the Brand or Model... Cheap though... It was the only one i had around that did fit.

CPU Temps can be lowered I hope:
I upped my core voltage to 1.25 Volts for that quick overclock to 4Ghz... I have now found out that it does 4 Ghz at 1.1V using much less Amps.. ;)
Also I am pretty sure I mounted my CPU heatsink 90 Degree off... lets see if that helps.

Water Temperature Sensore Coming in soon.


Cheers
Is one 120mm radiator enough to cool overclocked two R9 Nano & CPU?
If one can handle, it will be possible to squeeze your setup using ek-annihilator & 25mm thick radiator inside A4-SFX.
 
Seems like flexible bifurcated risers are out of the question for me.

C_Payne : Do you have any of those PCB's remaining from your first run?

Thanks!

I do... however I ran out of x16 Slots to solder...

U interested?

Also my right turn PCB's came in:

Waiting for my Annihilator to come in... Will Post Pictures and test results then.1490076002241-1922142794.jpg
 
I do... however I ran out of x16 Slots to solder...

U interested?

Also my right turn PCB's came in:

Waiting for my Annihilator to come in... Will Post Pictures and test results then.View attachment 19866

Hi C_Payne. Thanks for replying, I did decide to take a second look at my layout and realized that inflexible risers would just not fit in, so please count me out.

BUT your x8 + dual m.2 bifurcation riser seems great for another build I am doing. Count me in for that if you ever decide to produce it.
 
Hey everyone, I've decided to make an account as I've also been searching for a bifurcation solution for a while now--although not for GPUs. Up until finding this thread, I kept running into forum posts of people asking about it and the post then being closed once someone told them that the BIOS must support it. My mobo definitely supports bifurcation (z9pe-d8 ws) but I haven't been able to find the physical hardware to let me plug in multiple cards.

So I have two (2) UAD-2 PCIe 2.0 x1 DSP cards. I basically want to be able to split one of the x16 slots on the motherboard into two x1 slots. Although since the bifurcation on the mobo only goes down to x4, I suppose something that did x8x4x4 or x4x4x4x4 would be fine. Having read through the thread as well as browsed countless websites, this http://aerocooler.com/rc1-pely423-c...flexible-riser-one-pcie-x16-to-dual-pcie-x16/ seems like it's my best bet but I can't get excited about spending $70 for something that's pretty un-ideal. (I've become disabled so I'm not as moneyed as I was when I bought all this stupid stuff :/ )

Secondly, and this is more of a shout-out to others that may come across this thread as they look for a solution to their problem, I feel like there is a large interest building in PCIe-to-m.2 solutions. Especially once Sandy Bridge-EP and Naples hit; lots of pci lanes and not enough room on the motherboard. There are options provided by HP and Dell but they both appear to be BIOS locked to the specific vendor's workstations.

C_Payne, I don't know what sort of design/production skills you have but it looks like you're doing some crazy advanced stuff. Since it looks like you're already doing stuff with multiple m.2 slots I guess I'll just say that more people than you expect might be very interested.
 
So I have two (2) UAD-2 PCIe 2.0 x1 DSP cards. I basically want to be able to split one of the x16 slots on the motherboard into two x1 slots. Although since the bifurcation on the mobo only goes down to x4, I suppose something that did x8x4x4 or x4x4x4x4 would be fine. Having read through the thread as well as browsed countless websites, this http://aerocooler.com/rc1-pely423-c...flexible-riser-one-pcie-x16-to-dual-pcie-x16/ seems like it's my best bet but I can't get excited about spending $70 for something that's pretty un-ideal. (I've become disabled so I'm not as moneyed as I was when I bought all this stupid stuff :/ )

Would you be interested in one of my Risers?
I have my first version laying around... I would give it away for $20 plus shipping... dont need it anymore.
I Could also make an new one, the right angled version for you for $50 plus shipping.

I dont know how performance dependent your dsp cards are... maybe this could be a solution:
https://www.aliexpress.com/item/PCI...lgo_pvid=a75011fc-2519-4522-ba9b-dfbc6de37179
And since it has a packet switch it should be compatible with all mainboards.
Downside is that both DSP would share their gen2 x1 Bandwidth.

Chris
 
Would you be interested in one of my Risers?
I have my first version laying around... I would give it away for $20 plus shipping... dont need it anymore.
I Could also make an new one, the right angled version for you for $50 plus shipping.

I dont know how performance dependent your dsp cards are... maybe this could be a solution:
https://www.aliexpress.com/item/PCI...lgo_pvid=a75011fc-2519-4522-ba9b-dfbc6de37179
And since it has a packet switch it should be compatible with all mainboards.
Downside is that both DSP would share their gen2 x1 Bandwidth.

Chris

Hey Chris,

Thanks for the reply. I've seen those PCIe to USB splitters around but I was never really sure that they'd work for me. Additionally, I know a bunch of this stuff is from random OEMs anyway, but AliExpress terrifies me as I have no idea what the worst possible failure scenario would be. If just the little board dies that's fine I guess, but if it fries the cards or the rest of the system I would literally cry. Additionally, I'm not sure where one would then put the PCIe cards; I can't find a chassis that isn't meant for industrial-grade PCI expansion.

As far as your offer, it seems like the best way to get my feet wet. The left-angled one would be ideal for my case, but I don't want you to feel like you're giving something away. I'm happy to pay whatever the fair value is +shipping (I assume somewhere between $20 and $50). I'll PM you and reply with the details when you're ready :).

Russell
 
I am in the process of setting up a small server with the Asrock x99e-itx and I wanted to ask if there is some experience with Vt-d and PCIe card passthough using ESXi using the bifurcation cables.
 
I am in the process of setting up a small server with the Asrock x99e-itx and I wanted to ask if there is some experience with Vt-d and PCIe card passthough using ESXi using the bifurcation cables.

I bought one of C_Payne's riser/splitters with the intent on using it with ESXi, passing through 2 HBA cards to 2 different VM's. I haven't gotten around to testing it yet, but will reply back when I do.
 
I am in the process of setting up a small server with the Asrock x99e-itx and I wanted to ask if there is some experience with Vt-d and PCIe card passthough using ESXi using the bifurcation cables.

I am using KVM and iommu works splendidly. Both of my GPU's end in different IOMMU groups.
I dont see why there should be issues with ESXI. Do you have the lates 3.64 bios?

Chris
 
Back
Top