AMD and Nvidia graphic cards at same time

[umran23]

@Gigamaxx is it possible to run intel and nvidia together? I want to use my Intel HD 530 primarily for video DECODE since nvidia web drivers lack the codecs in Mac OS and my system always hogging the CPU even in basic YouTube videos but want to use nvidia for my adobe suite needs.

 

[Gigamaxx]

@Gigamaxx is it possible to run intel and nvidia together? I want to use my Intel HD 530 primarily for video DECODE since nvidia web drivers lack the codecs in Mac OS and my system always hogging the CPU even in basic YouTube videos but want to use nvidia for my adobe suite needs.

I’ve heard about people using igpu vram allocation and multi monitor while having Peg as primary. For my Gigabyte board to do this I need to select peg mode then set ram allocation to 64mb, and select multimonitor. My Nvidia card at the moment isn’t a gt210so it doesn’t need Web Drivers. I don’t know if this will work for pascal cards but one user reported it working this way.

Peg mode 64mb ram allocation and multimonitor if you have those settings in bios.

 

[Scott1204]

In a dual card setup, can you disable an Nvidia graphics card? I have an RX580 that I’m installing with a 1080TI. The 1080 is for VR in Windows. I’d like the RX to be available in Mac.

If I don’t install any Drivers, would the Nvidia remain offline? Or would I have to do something more like disable a PCI slot? Running High Sierra.

 

[mwo21]

is it possible to run (in my case) a gtx 970 and a RX 460 at the same time. Basically utilizing CUDA and OpenCL at the same time. I figured the gtx would handle displays and the RX would just handle OpenCL for FCPX. Is this idea feasible if so any obstacles i would run into?

Tried it. Nvidia 1080 ti in slot 1, AMD Rx480 slot two. All displays connected only to 1080. Ran FCP X. No improvement in Bruce X benchmark at all. Running Luxmark, valley, geekbench... all the OpenCL scores are much higher than either alone. But, general use, the computer feels sluggish. I'm removing the RX480. In every way, except 1, the 1080Ti as the only card is much faster than even my dual RX480 set up. The one place RX 480 shines is with FCPX export. RX export is much faster than Nvidia.

 

[Scott1204]

Tried it. Nvidia 1080 ti in slot 1, AMD Rx480 slot two. All displays connected only to 1080. Ran FCP X. No improvement in Bruce X benchmark at all. Running Luxmark, valley, geekbench... all the OpenCL scores are much higher than either alone. But, general use, the computer feels sluggish. I'm removing the RX480. In every way, except 1, the 1080Ti as the only card is much faster than even my dual RX480 set up. The one place RX 480 shines is with FCPX export. RX export is much faster than Nvidia.

I'm experimenting with my setup. Agree 1080 is a much better card but my FCX performance is terrible. Can't even play 4K files smoothly. Plus, can't get iTunes movies to play due to HDCP issues.

What I want to do is use the 1080 for VR in Windows and the RX460 in High Sierra (if it works, will upgrade to a RX480).

In my case, the RX is in slot one, 1080 in slot two. System hangs a few seconds into boot. If I remove the 1080, the system boots normally. In Windows, both work fine.

But for High Sierra, I don't need both...just the RX. So, if there were a way to simply ignore the 1080, then this would work (in theory).

We're able to block certain USB ports from High Sierra. Can we block PCI ports?

@Gigamaxx
You have any ideas?

Thanks for any advice.

 

[Gigamaxx]

I'm experimenting with my setup. Agree 1080 is a much better card but my FCX performance is terrible. Can't even play 4K files smoothly. Plus, can't get iTunes movies to play due to HDCP issues.

What I want to do is use the 1080 for VR in Windows and the RX460 in High Sierra (if it works, will upgrade to a RX480).

In my case, the RX is in slot one, 1080 in slot two. System hangs a few seconds into boot. If I remove the 1080, the system boots normally. In Windows, both work fine.

But for High Sierra, I don't need both...just the RX. So, if there were a way to simply ignore the 1080, then this would work (in theory).

We're able to block certain USB ports from High Sierra. Can we block PCI ports?

@Gigamaxx
You have any ideas?

Thanks for any advice.

Try this 1080 in slot 1 RX in slot 2 use nv_disable=1 for boot arg it should get 7mb support but act as a helper card. Don't use any Nvidia fixes you want no support just the default display settings. Cable hooked to both cards to get screen menu. Or blind boot with only the RX connected. Don't use whatever green or rad de init command just try it without any RX fixes. This used to work in Sierra before we had fixes for RX cards. You should be able to boot into OSX and use RX card and in windows the RX won't get detected.

 

[Scott1204]

Try this 1080 in slot 1 RX in slot 2 use nv_disable=1 for boot arg it should get 7mb support but act as a helper card. Don't use any Nvidia fixes you want no support just the default display settings. Cable hooked to both cards to get screen menu. Or blind boot with only the RX connected. Don't use whatever green or rad de init command just try it without any RX fixes. This used to work in Sierra before we had fixes for RX cards. You should be able to boot into OSX and use RX card and in windows the RX won't get detected.

Great, thanks for the ideas! Will give it a shot and report back. Might be helpful for others wanting to use FCX with better performance.

 

[FrancisBitter]

Try this 1080 in slot 1 RX in slot 2 use nv_disable=1 for boot arg it should get 7mb support but act as a helper card. Don't use any Nvidia fixes you want no support just the default display settings. Cable hooked to both cards to get screen menu. [...]

This actually works, though sadly, it's the only procedure to make it work, got something similar with a 1070 and an RX 580. My problem with this now is, the NVIDIA needs to be in PEG0, the AMD in PEG1, only black screen otherwise. The performance however is by far not as amazing as only running the RX 580 by itself. PEG0 gets 16x PCIE lanes with a "dead card" as it's not used while the AMD on PEG1 only gets 4x PCIE lanes.

Is the number of lanes cause of this sluggishness or is it that the NVIDIA card breaks hardware acceleration just by being there?

 

[Scott1204]

I just tried installing my RX460 in PCI 2 and it worked (shockingly).

I tried this a few months ago and the system just black screened. Recently updated to 10.13.2 and I think that may have helped. System report shows the card, correctly identified. Not seeing any performance improvement at the moment.

When the 460 is solo, I can play iTunes movies (no HDCP errors). It never worked with my 1080Ti.

Wondering if there's a workaround to make it work now that both cards are in the system.

 

[Scott1204]

I tried putting the nvidia card is slot 1 (well, PCIX16_1) with the AMD in slot 3 (PCIX16_2) and I see no difference in behaviour.

Once it booted showing something on the two screens, but not where the login prompt were located. It was like it had detected a 3rd screen . Moving the mouse across and I could see it spanning from one screen to the next.

I wonder if there's a possibility to fully disable a PCI slot using clover, that would be the simplest: pretend the card isn't there

Curious about this too and asked in another thread & tried searching but never found an answer.