Choice of graphics card

[nvaert1986]

Any update?

I mostly made support tickets at ASUS support, Sapphire support and made a issue on Gitlab (https://gitlab.freedesktop.org/drm/amd/-/issues/3809) regarding the hang during cold boot (which I also made with the 6.6 kernel but closed, due to the fact that I thought I had resolve that issue, but that wasn't the case). The hang in amdgpu during the bootup process causing the amdgpu driver to hang / crash is apparently caused by the psp according Alex Deucher (see the GitLab thread). Don't have any other information so far. Recently tried the 6.12.3 kernel and the issues still occur.

[Panix]

Any update?

[nvaert1986]

Okay. Is Gentoo on both PCs?

What is the Asus mobo? Also, the delay when the Asus PC boots - is just a delay - not a crash - just a long wait until it finishes the boot? This delay is every boot? If it's every boot - there might have been a 'mod' or change in the boot settings in the BIOS - you might want to check that out and even change it - set the number lower and see if it boots faster? If it's already quite low - then, yes, it could be a bios bug. 'Need to know the Asus mobo name so one could google/web search - but, my first inclination or thought is that boot number (seconds) is too high - maybe?

As for the ASUS and MSI PCs both crashing - due to some amdgpu driver issue - I dunno... I would try a different distro to make sure whether it's a repetitive error - that it's not distro interdependent. It *shouldn't* matter in theory but you never know - trying the latest Fedora or even OpenSUSE Tumbleweed will test the lastest/later kernel to confirm. If you can't even boot latter kernels, then you could try booting with a live usb flash drive.

I would also get a log or error msg for the crashes.

It sounds like both PSUs are fine - right, those a good PSUs.

- When the delay on the ASUS motherboard occurs, the system will not boot into X.org / Wayland interface and just hang indefinitely at the console stuck with a _ cursor. The amdgpu driver usually gets loaded (as the screen switches resolution), but just nothing else happens. This happens consistently every time this delay occurs. That's why I think it's a bug (as the MSI machine does not have this issue). When this issue occurs, my mouse and keyboard do not respond either. The system does not hang however, as I'm able to press the power button and then it'll just safely turn off the machine. The only way to resolve this issue is to physically unplug the power cord, wait a 1 minute and plug the power back in.

NOTE: When the graphical POST delay bug happens, the Microsoft Windows installer does show it's graphical screen, but the mouse and keyboard (USB devices) are not responding either, until I physically unplug the power and plug it back in after 1 minute.

This delay happens completely at random unfortunately so I have no idea, but I'd say every 1 out of 20 - 30 cold boots. I've tried updating the BIOS / UEFI, resetting it, tried turning PCI-E BAR on and off, tried turning ASPM on and off, discrete graphics on and off but nothing seems to resolve the issue. Even turning off memory context restore or turning fast boot on or off does not resolve the issue.

As for other distro's: I've tried using Arch and it shows similar results with the amdgpu driver (so I don't think it's distro related).

I've searched for this issue already on the ASUS forums and Google and several other resources, but couldn't find anything remotely useful of other people encountering a similar bug.

It feels like some of the chips are trying to load firmware (from UEFI) or initialize, something goes wrong in the initialization process and it will just continue to malfunction and that the driver remains loaded when power is plugged in (in some sort of stand-by state) and when unplugging the devices the system flushes it's power requiring the full initialization again, which causes it to start working again.

[Panix]

I already did a lot of troubleshooting, but to answer the questions to get a better picture:

- Only the ASUS (AMD Ryzen) motherboard has the issue where there's a delay of display in the POST for both AMD graphics cards (which I think is a BIOS / UEFI bug)
- The ASUS (AMD Ryzen) system sometimes hangs / crashes when the amdpu driver gets loaded when I'm using the integrated graphics too (I normally have this disabled in the BIOS / UEFI).
- Both the ASUS and MSI motherbord have issues with the system hanging or crashing when the amdgpu driver gets loaded on a variety of kernels from a cold boot
- Both the ASUS and MSI system have their own PSU's (the ASUS one has an Corsair RMx Shift 1000W and the MSI system a Corsair RM850, which are known to be good PSU's)
- The MSI systems seems to be running without an issue when the Intel Integrated Graphics driver gets loaded (though I did limited testing; I didn't load any games, because which games can you load on an Intel iGPU?.. but it boots fine)
- Swapping out the PSU's seems to have no effect
- I've tried using an older Enermax Platimax PSU and it seems to give the same results.
- The RDNA2 card is an MSI card and the RDNA3 card is an Sapphire card.

Okay. Is Gentoo on both PCs?

What is the Asus mobo? Also, the delay when the Asus PC boots - is just a delay - not a crash - just a long wait until it finishes the boot? This delay is every boot? If it's every boot - there might have been a 'mod' or change in the boot settings in the BIOS - you might want to check that out and even change it - set the number lower and see if it boots faster? If it's already quite low - then, yes, it could be a bios bug. 'Need to know the Asus mobo name so one could google/web search - but, my first inclination or thought is that boot number (seconds) is too high - maybe?

As for the ASUS and MSI PCs both crashing - due to some amdgpu driver issue - I dunno... I would try a different distro to make sure whether it's a repetitive error - that it's not distro interdependent. It *shouldn't* matter in theory but you never know - trying the latest Fedora or even OpenSUSE Tumbleweed will test the lastest/later kernel to confirm. If you can't even boot latter kernels, then you could try booting with a live usb flash drive.

I would also get a log or error msg for the crashes.

It sounds like both PSUs are fine - right, those a good PSUs.

[nvaert1986]

Hmmmm..... I had a friend who used Gentoo - but, I have never used it. I would try another distro but that's probably nothing to do with it.
So, step/experiment #2 - imho - would be to try taking the gpu out - do you have Linux or Gentoo on both machines? You have an Intel and AMD Ryzen PC, right? The Intel pc will have integrated graphics, for sure. If it's the ASUS mobo - then taking out the gpu would isolate the problem to either the pcie slot i.e. pcie x16 - or some other problem - if the boot problem continues. If it doesn't continue - then it's an issue with the amd (FOSS) driver/firmware etc. - something not hardware related.
Right?

Otherwise - a boot problem without the gpu - could be psu related, bios/stability problem - but, I believe you posted an amdgpu crash message - if that's related?

I already did a lot of troubleshooting, but to answer the questions to get a better picture:

- Only the ASUS (AMD Ryzen) motherboard has the issue where there's a delay of display in the POST for both AMD graphics cards (which I think is a BIOS / UEFI bug)
- The ASUS (AMD Ryzen) system sometimes hangs / crashes when the amdpu driver gets loaded when I'm using the integrated graphics too (I normally have this disabled in the BIOS / UEFI).
- Both the ASUS and MSI motherbord have issues with the system hanging or crashing when the amdgpu driver gets loaded on a variety of kernels from a cold boot
- Both the ASUS and MSI system have their own PSU's (the ASUS one has an Corsair RMx Shift 1000W and the MSI system a Corsair RM850, which are known to be good PSU's)
- The MSI systems seems to be running without an issue when the Intel Integrated Graphics driver gets loaded (though I did limited testing; I didn't load any games, because which games can you load on an Intel iGPU?.. but it boots fine)
- Swapping out the PSU's seems to have no effect
- I've tried using an older Enermax Platimax PSU and it seems to give the same results.
- The RDNA2 card is an MSI card and the RDNA3 card is an Sapphire card.

[Panix]

I'm using Gentoo as a distribution (so I can be very flexible when it comes to the kernel) and Plasma 6.1.5 as my desktop environment.
I've tried kernels from 6.6 up until 6.11 so far, where it's been a constant mixed bag. The 6.7 and 6.8 kernel had weird bugs and the 6.9 kernel showed significant performance losses. The kernels 6.6.x and 6.10.11 seem to be the most stable so far. Kernel 6.11 doesn't boot at all and always freezes when loading. I did try kernel 6.12-rc6 a few weeks ago and it booted, but haven't tried it since.

My experience (based on the GeForce 1070, haven't had any more recent GPU's, so can't judge more recent GPU's) is that if you stick with the LTS kernel, they seem to be extremely stable.

It's unfortunate when it comes to the AMD card (would love not to have these boot issues), and I just can't figure out why. I've even suspected the ASUS motherboard to have some sort of weird BIOS / UEFI bug, as I've had similar weird bugs I experienced on past ASUS boards, but then I'm talking 10 - 12 years ago so I'm not sure whether that's representative; rarely seen issues with MSI based systems, though I don't have any actual hard evidence for this.

Hmmmm..... I had a friend who used Gentoo - but, I have never used it. I would try another distro but that's probably nothing to do with it.
So, step/experiment #2 - imho - would be to try taking the gpu out - do you have Linux or Gentoo on both machines? You have an Intel and AMD Ryzen PC, right? The Intel pc will have integrated graphics, for sure. If it's the ASUS mobo - then taking out the gpu would isolate the problem to either the pcie slot i.e. pcie x16 - or some other problem - if the boot problem continues. If it doesn't continue - then it's an issue with the amd (FOSS) driver/firmware etc. - something not hardware related.
Right?

Otherwise - a boot problem without the gpu - could be psu related, bios/stability problem - but, I believe you posted an amdgpu crash message - if that's related?

[nvaert1986]

What distro are you using? I couldn't find any mention of it in your posts.

If you go with Nvidia - it's probably best to use a distro with a very recent kernel - it's most likely that these distros - like Fedora, Opensuse (Tumbleweed) and possibly, the latest Ubuntu - would have the best support with Nvidia + Wayland.

It would be interesting to see if you have a different experience - and if the black screen /freezes are solved.

But, there might be different/new issues - just be aware of it. There's lots of ppl who hate using nvidia gpus - but, I guess that might be because there's steps involved in getting the proprietary driver installed - the FOSS nature of most distros - requires manual installs - whether it's via CLI or enabling 3rd party repos and then 'clicking gui managers' of some kind - to install the binary driver - Ubuntu's method is probably easiest - Fedora is relatively easy - but, maybe a bit more work - and some distros like Debian - requires a lot more work - and afaik, still requires manual commands via CLI.

As far as AMD gpu problems with latter kernels - I haven't heard of that - that sounds peculiar. But, I don't have an AMD gpu - and I mostly research the current situation with amd gpus/drivers with regards to productivity software - in Linux, mostly - but, also Windows.

If all you do is game - an AMD gpu in Linux probably makes more sense (than Nvidia) but if there's crashes or screen issues - one idea is to research the 'error messages/log' you get and see what happens when you do a (google) web search on it?

I'm using Gentoo as a distribution (so I can be very flexible when it comes to the kernel) and Plasma 6.1.5 as my desktop environment.
I've tried kernels from 6.6 up until 6.11 so far, where it's been a constant mixed bag. The 6.7 and 6.8 kernel had weird bugs and the 6.9 kernel showed significant performance losses. The kernels 6.6.x and 6.10.11 seem to be the most stable so far. Kernel 6.11 doesn't boot at all and always freezes when loading. I did try kernel 6.12-rc6 a few weeks ago and it booted, but haven't tried it since.

My experience (based on the GeForce 1070, haven't had any more recent GPU's, so can't judge more recent GPU's) is that if you stick with the LTS kernel, they seem to be extremely stable.

It's unfortunate when it comes to the AMD card (would love not to have these boot issues), and I just can't figure out why. I've even suspected the ASUS motherboard to have some sort of weird BIOS / UEFI bug, as I've had similar weird bugs I experienced on past ASUS boards, but then I'm talking 10 - 12 years ago so I'm not sure whether that's representative; rarely seen issues with MSI based systems, though I don't have any actual hard evidence for this.

[Panix]

I used a NVIDIA GeForce 1070 Ti in the past. This was rock solid in my experience. I'll try getting newer / cheap NVIDIA card (as they seem to be extremely expensive and I'm open to any brand as long as the quality is good and it's stable, that's why I asked the question regarding the Intel based cards), to see what it does.
I'm using both an Intel 12th gen CPU and a AMD Ryzen series and this does not affect anything (aside from the kernel configuration) related to gaming. I consider both of them to be stablem.

My experience with configuring an NVIDIA GPU on Linux is relatively easy. Just install the nvidia-drivers package suppported by the GPU, reboot and done (and make sure you're using a supported kernel. I always tend to stick with the LTS kernels).

What distro are you using? I couldn't find any mention of it in your posts.

If you go with Nvidia - it's probably best to use a distro with a very recent kernel - it's most likely that these distros - like Fedora, Opensuse (Tumbleweed) and possibly, the latest Ubuntu - would have the best support with Nvidia + Wayland.

It would be interesting to see if you have a different experience - and if the black screen /freezes are solved.

But, there might be different/new issues - just be aware of it. There's lots of ppl who hate using nvidia gpus - but, I guess that might be because there's steps involved in getting the proprietary driver installed - the FOSS nature of most distros - requires manual installs - whether it's via CLI or enabling 3rd party repos and then 'clicking gui managers' of some kind - to install the binary driver - Ubuntu's method is probably easiest - Fedora is relatively easy - but, maybe a bit more work - and some distros like Debian - requires a lot more work - and afaik, still requires manual commands via CLI.

As far as AMD gpu problems with latter kernels - I haven't heard of that - that sounds peculiar. But, I don't have an AMD gpu - and I mostly research the current situation with amd gpus/drivers with regards to productivity software - in Linux, mostly - but, also Windows.

If all you do is game - an AMD gpu in Linux probably makes more sense (than Nvidia) but if there's crashes or screen issues - one idea is to research the 'error messages/log' you get and see what happens when you do a (google) web search on it?

[nvaert1986]

I thought I was the only one getting random black screens and display signal interruptions. Been having that issue since my RX480 (with FX8320 and Ryzen 1700) and now a RX6750XT (with Ryzen 5800X). Does not happen regularly and only once in a blue moon, so I learned to just live with it, but it is frustrating. Also for some reason my setup just refuses to work at all with DP 1.2 for some odd reason (have to switch my monitor to DP 1.1), but that could be my aging monitor.

Other than that, my experience with AMD GPUs has been satisfactory with what I do 90% of the time (games from old DX5 to newer DX12/Vulkan titles, internet, movie streaming and local playback, some work with Windows VMs). I don't even bother with the proprietary driver (maybe the above problems would be resolved if I did but who knows). The only nvidia cards I have are older models like legacy 700 series, and for those I just can't be bothered to get the proprietary drivers to work with Wayland and secure boot (and nouveau is unusable). Maybe current nvidia cards are better now with the new nvidia kernel module.

Have not tried Intel cards. From reviews it seems the A770 lands somewhere around RX7600XT performance (in Windows at least) and the current asking price is not bad. New Intel cards are said to come out soon, so it may be wise to wait a bit.

For me this happens every 3 - 4 cold boots (power-off and power-on of the system; not unplugging the system). It doesn't happen with reboots (sometimes I'm suspecting the ASUS motherboard; as I've seen several weird issues with ASUS in the past, though I'm just not sure). My screen either freezes or just goes blank (and eventually my monitor goes to stand-by) when loading the amdgpu driver.

My experience is that most games work fine (and can play for hours; so I don't think it's the cards), but that there's just some stability issues, where the kernel crashes with a notorious "[drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=xxxx, emitted seq=xxxx"" like message where my screen eventually switches just to stand-by and I have to reboot my system. That's something which I've never experienced on an NVIDIA GPU.

I'd like to see what an Intel card does in regards to stability, as I'd prefer an open source and better supported solution (if the support is actually better, as in my experience with AMD and asking to bi-sect is not very end user friendly; even though I'm not an end user and I'll do my best to see what I can do), I'm just on a limited budget and I'm just not sure about Intel cards in regards of compatibility and performance on Linux, as I can't seem to find that much information (that's why I asked my question here on the forums).

[nvaert1986]

Imho, the best thing for you to do is three-fold: 1) read the forums - in this section for ppl talking about their amd gpus - you will get a lot of opinions but I think most ppl here swear by amd gpus - for Linux gaming as well. 2) wait for specific replies. 3) Sell the older RDNA 2 card and buy an nvidia gpu. You won't know what's different or have a good idea until you try a nvidia card.

If someone recognizes and understands what you are talking about - and/or they have experienced the same thing - they can either relate or can assist you in what to do to fix it or workarounds.

If I had two PCs - I'd have one system with an amd gpu and 1 with an nvidia - just so I could compare but also, I am sure the experience would be different with them simply because of the 2 gpus.

I could be mistaken but I'm not sure the Intel cpu vs AMD ryzen cpu difference will be significant - at least for gaming experience - the Intel 13th gen was an improvement but the nvidia proprietary/wayland setup will probably still be quite different than using AMD's FOSS amdgpu setup. But, you won't know if you prefer one over the other or whether the experience is preferable with one until you configure one of the PCs with an nvidia gpu - I would try to get one closest to the amd gpu tier, though - so, it's a more fair comparison.

I used a NVIDIA GeForce 1070 Ti in the past. This was rock solid in my experience. I'll try getting newer / cheap NVIDIA card (as they seem to be extremely expensive and I'm open to any brand as long as the quality is good and it's stable, that's why I asked the question regarding the Intel based cards), to see what it does.
I'm using both an Intel 12th gen CPU and a AMD Ryzen series and this does not affect anything (aside from the kernel configuration) related to gaming. I consider both of them to be stablem.

My experience with configuring an NVIDIA GPU on Linux is relatively easy. Just install the nvidia-drivers package suppported by the GPU, reboot and done (and make sure you're using a supported kernel. I always tend to stick with the LTS kernels).