Status
Not open for further replies.

maytiffins

Prominent
Oct 21, 2018
9
0
520
I am using a Radeon VII for compute alongside a 980ti which I am using for display. I have connected the R7 to a 1x riser however it is having error code 43 in device manager. The card was working fine in another computer and there were no errors
I have tried using the latest drivers (19.6.2) which brings me to this error, I have had some issues trying to install the 19.5.2 driver which resulted in blue screen on startup.

I don't know what could be causing this issue, but I will play around with the riser because it was working previously without it.

Other info:
OS: Win 10 1903
MB: Gigabyte AX370M-DS3H
CPU: 1700X
 
Solution
Yes I did consider that, until I realised that my Benq monitor could only use its maximum resolution (1080p @ 144 Hz) over DVI, and otherwise would be capped at 60 Hz.

I updated the chipset driver and BIOS to F31 which got the Radeon VII to work on the 3rd PCIe slot and starts without errors on the latest windows (1903) and AMD drivers (19.6.2)

After that I ran into a small issue of the number of available PCIe lanes, so I couldn’t run both SATA controllers on the motherboard so now I have no SATA ports left.

The only port that is on the same controller as the M.2 drive is directly underneath the 16x slot so the 980ti is barely in place. It nearly blocks the riser from going in the third slot, but the screws keep it place, just.
I have not been able to run anything on the Radeon VII while it has been on the riser as the error is on startup. I have a bequiet 650W power supply that was using 305-315W without the Radeon VII
 
Last edited:
There is zero chance of that PSU driving both GPUs in that system. I have a Radeon VII in my main game rig and it is a power hungry beast. Great GPU, but demanding.

Have you tried the Radeon VII connected directly into a PCIe x16 slot (no x1 riser cable)? Or swapping the GPUs you have (980ti connected to rider and Radeon as primary)?
 
  • Like
Reactions: digitalgriffin
I have been testing various arrangements, the Radeon VII works when plugged into the motherboard, but not the riser. The 980ti works with either. All those tests were with one GPU at a time. The Radeon VII also works with the 980ti on the riser while it’s on the motherboard.

I have got a 1000W Corsair PSU that I will be able to use if I can get the Radeon VII to work on the riser, as I need more than 1x to run games with the 980ti
 
For some reason it works when the riser is in the first (16x) slot. I’ll go through the BIOS to see what the options are that are making it stop working in the third slot
 
Yes I did consider that, until I realised that my Benq monitor could only use its maximum resolution (1080p @ 144 Hz) over DVI, and otherwise would be capped at 60 Hz.

I updated the chipset driver and BIOS to F31 which got the Radeon VII to work on the 3rd PCIe slot and starts without errors on the latest windows (1903) and AMD drivers (19.6.2)

After that I ran into a small issue of the number of available PCIe lanes, so I couldn’t run both SATA controllers on the motherboard so now I have no SATA ports left.

The only port that is on the same controller as the M.2 drive is directly underneath the 16x slot so the 980ti is barely in place. It nearly blocks the riser from going in the third slot, but the screws keep it place, just.
 
Solution
A properly mounted m.2 SSD (like with the proper standoff) should not interfere with a GPU (unless you have one with a heatsink attached to it). You might want to check that out.

Glad you sorted out the GPU config. What application are you using the Radeon VII as a compute device for?
 
The SATA controller for the M.2 drive has one SATA port, which I am using to connect to a HDD. The cable is in the way of the 980ti due to the placement of the port on the motherboard, not the M.2 drive.

I plan to use the card to do some rendering work and some AI implementations. The 16GB of HBM2 will really improve the workload.

Also I have some friends who have been using their Radeon VII to mine with ethash algorithms since the memory bandwidth gives some performance so I’ll be giving that a go as well.

I will still have to see if the PCIe bandwidth has any effect on the workloads, but at least I have the card working in the system now.
 
hi OP, just wanted to add I outwitted the (error code 43) just the other day

there is an amd pixel clock patch that made it go away. easy to find, do not have a link handy.

in my case I got some rx470 cards with modified bios... decided to make a mining rig now that card prices slumped... got em cheep!!

it also seemed tp mention something about 144hz monitor support, etc

anyway it worked for me device mgr flags went away and system lag resolved.
(cards are on risers ofc)


might be something to try out to see if the VII would work on a riser
 
Status
Not open for further replies.