PCI riser weirdness

It seems that a lot of GPU problems revolve around specific versions of drivers. Though NVidia has their own support structure, you can often learn from information reported by others who fold.

Moderators: Site Moderators, FAHC Science Team

Post Reply
doctorzaius
Posts: 10
Joined: Sun Jun 07, 2020 6:41 pm

PCI riser weirdness

Post by doctorzaius »

Okay, my motherboard has the following:

3x PCIe 3.0 x16 slots (support x16/x0/x4, x8/x8/x4 modes)
3x PCIe 3.0 x1 slots

I have two GPUs (both RTX 2070) and I'd like to run one using a riser because that way both of them stay cool without screaming fans. There is some speed hit doing this, even running linux, but its not that big of a deal. I would also like to add a third GPU and this is where the problems start. Adding the third GPU means that I have to attach the riser to one of the 3.0x16 slots. And that works, except that performance drops by 5x or more. So a WU that would take 1 hour to complete on the x1 slot, takes 5 or more hours on the x16 slot. I have a feeling this may be a PCIe issue (and not a driver issue). But I figured I start here. Anyone seen this?
PantherX
Site Moderator
Posts: 7020
Joined: Wed Dec 23, 2009 9:33 am
Hardware configuration: V7.6.21 -> Multi-purpose 24/7
Windows 10 64-bit
CPU:2/3/4/6 -> Intel i7-6700K
GPU:1 -> Nvidia GTX 1080 Ti
§
Retired:
2x Nvidia GTX 1070
Nvidia GTX 675M
Nvidia GTX 660 Ti
Nvidia GTX 650 SC
Nvidia GTX 260 896 MB SOC
Nvidia 9600GT 1 GB OC
Nvidia 9500M GS
Nvidia 8800GTS 320 MB

Intel Core i7-860
Intel Core i7-3840QM
Intel i3-3240
Intel Core 2 Duo E8200
Intel Core 2 Duo E6550
Intel Core 2 Duo T8300
Intel Pentium E5500
Intel Pentium E5400
Location: Land Of The Long White Cloud
Contact:

Re: PCI riser weirdness

Post by PantherX »

Welcome to the F@H Forum doctorzaius,

From what I have read here, as long as you're running Linux, you can use 3x PCIe 3.0 x16 Slots without encountering a a significant performance impact. However, if you're using any of the 3x PCIe 3.0 x1 Slots, you will encounter significant slowdown.

Is this the current setup:
PCIe 3.0 x16 running at x8 -> RTX 2070
PCIe 3.0 x16 running at x8 -> RTX 2070
PCIe 3.0 x16 running at x4 with riser -> Third GPU

Also, is the riser a powered one or not?
Do you have sufficient power from the PSU to drive those GPUs?
What kind of CPU do you have?
What are the operating temperatures of the GPUs?
ETA:
Now ↞ Very Soon ↔ Soon ↔ Soon-ish ↔ Not Soon ↠ End Of Time

Welcome To The F@H Support Forum Ӂ Troubleshooting Bad WUs Ӂ Troubleshooting Server Connectivity Issues
doctorzaius
Posts: 10
Joined: Sun Jun 07, 2020 6:41 pm

Re: PCI riser weirdness

Post by doctorzaius »

Additional data: the riser is powered and the PSU is 1600 watts. Processor is Intel Coffee Lake. GPUs are cold. I think they have a target temp of 61C and they both operate around that temp. The remotely mounted GPU achieves that with intermittent fan use. The GPU mounted on the motherboard uses about 30% of fan capacity. Picture is worth a thousand words: https://imgur.com/a/4Ifm9ho
foldy
Posts: 2061
Joined: Sat Dec 01, 2012 3:43 pm
Hardware configuration: Folding@Home Client 7.6.13 (1 GPU slots)
Windows 7 64bit
Intel Core i5 2500k@4Ghz
Nvidia gtx 1080ti driver 441

Re: PCI riser weirdness

Post by foldy »

Maybe the 3rd GPU just got a work unit which takes more time? Or the 3rd riser is damaged? It should really work in theory - even putting 4 GPUs in.
doctorzaius
Posts: 10
Joined: Sun Jun 07, 2020 6:41 pm

Re: PCI riser weirdness

Post by doctorzaius »

I haven't installed the 3rd GPU yet. This happens just moving the 2nd GPU to what I assumed would be a faster slot. The big problem is that, as it is, there's only one little slot available. And if I can't use the bigger slots, there's no reason to even install the 3rd.
PantherX
Site Moderator
Posts: 7020
Joined: Wed Dec 23, 2009 9:33 am
Hardware configuration: V7.6.21 -> Multi-purpose 24/7
Windows 10 64-bit
CPU:2/3/4/6 -> Intel i7-6700K
GPU:1 -> Nvidia GTX 1080 Ti
§
Retired:
2x Nvidia GTX 1070
Nvidia GTX 675M
Nvidia GTX 660 Ti
Nvidia GTX 650 SC
Nvidia GTX 260 896 MB SOC
Nvidia 9600GT 1 GB OC
Nvidia 9500M GS
Nvidia 8800GTS 320 MB

Intel Core i7-860
Intel Core i7-3840QM
Intel i3-3240
Intel Core 2 Duo E8200
Intel Core 2 Duo E6550
Intel Core 2 Duo T8300
Intel Pentium E5500
Intel Pentium E5400
Location: Land Of The Long White Cloud
Contact:

Re: PCI riser weirdness

Post by PantherX »

Thanks for the picture! If the sliver PCIe Slot is slow, then look at your motherboard if it is sharing the PCIe lanes with an M.2 SSD or something else. I agree with foldy that it should work without issues.
ETA:
Now ↞ Very Soon ↔ Soon ↔ Soon-ish ↔ Not Soon ↠ End Of Time

Welcome To The F@H Support Forum Ӂ Troubleshooting Bad WUs Ӂ Troubleshooting Server Connectivity Issues
doctorzaius
Posts: 10
Joined: Sun Jun 07, 2020 6:41 pm

Re: PCI riser weirdness

Post by doctorzaius »

Thanks all. I replaced the little card that plugs into the PCI slot and everything is good. So, electrical problem. Which is weird because it was working--just really, really slowly. Thanks again.
bruce
Posts: 20910
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: PCI riser weirdness

Post by bruce »

I'm not sure exactly what you're describing, but if you have, say, a x4 connection with a bad pin somewhere, the connection might deprecate itself to, say, a x1 connection.
doctorzaius
Posts: 10
Joined: Sun Jun 07, 2020 6:41 pm

Re: PCI riser weirdness

Post by doctorzaius »

I think that is exactly what happened.
Post Reply