New Ryzen 7 1800X build impressions.

[Praz]

Hello

There are several errors in the motherboard manual regarding Q-Codes. Code 8 often occurs because the CPU/SOC voltages have not been brought online during the initial AMD microcode sequence.

[Darth Ender]

I say it's irrelevant because microcode loading occurs when you first bootup. So the cause of why my computer is showing that error has nothing to do with microcode loading because it will boot just fine and stay booted and run benchmarks under full load for however long i want. It's an error that happens after the error that i actually care to resolve.

The error seems to only occur when i'm not using the computer and it's idle for a long time, like overnight.

Unfortunately, i think it may be due to a power management function. I know for a fact that my 480 video card hates waking up from sleep. It will lock the computer ....and maybe cause this issue but i'm not sure. When i review what happened prior to the lock/reboot, nothing abnormal is logged.

Whenever i fail to disable dpms and screen blanking, i can guarantee that eventually the video card will not wake up and it will cause the entire computer to lockup. I'm now going to test my voltages with dpms disabled and screen blanking off and see if everything is good.

[MeanMachine]

Several Qcode errors in the manual indeed. Now I feel dead in the water. lol
It would be nice to know just what errors your referring to Praz. ?

I'm not getting Code 08 (never did) nor undocumented code 0d anymore and as you see from my previous post at #18, I was stable at 4184MHz with 1.504V on the core.
I brought it back to 4016MHz with 1.482V on the Core for safety and temperatures are acceptable according to HWInfo under load.

My biggest issue was being able to maintain that CPU Core frequency and getting my 3000MHz RAM stable at the same time. I've also sorted the Cstate sleep issue.
Enough of the OCing for now as I feel I've gone as far as I can go with Bios 1002. So, no further till the next stable Bios is listed at support and hopefully an improvement.
Will report back later with Gaming results.

[Darth Ender]

I'm interested in your c-state issue. What were the symptoms before you fixed it?

I'm fairly convinced my rebooting/locking up issues are revolving around some power management function not behaving well. Could be linux specific, could be some default bios setting that linux/windows just goes without overriding or changing. I dont stay in windows that often so i'm not sure if it has the same issues.

[Darth Ender]

Well, woke up this morning after a full day booted up and it had rebooted and failed with bios code 8 again.

Checking if the issue occurs in Windows now. I have windows setup to high performance, nothing sleeping power management. Not making any changes to my voltages and such. If windows ends up not rebooted for over a day, at least I can be fairly sure that it's linux specific that's triggering whatever issue is causing the machine to reboot in the first place.

[Darth Ender]

Well, windows has not had any issue sitting with performance mode and all power saving options i could find turned off for over a day now. No reboots and no code 8. So I have to assume that there is an issue with my linux kernel doing something that the ryzen cpu or board doesn't like. Probably sleeping some piece of hardware that has trouble coming back up and causing a reboot and then causing the code 8 from the crash-reboot.

Yay. Well, at least it's not hardware related. Just something not behaving the way it's expected. Not surprising from this platform.

[MeanMachine]

Good, glad you have the issue sorted. My sleep issue was resolved with an update to my mouse and keyboard drivers and I disabled any tree hugging feature I could find.
I was using the MS default drivers and system would reboot when using either to wake from sleep and probably due to signal state not interacting with PCH correctly. Not 100% sure what the fix was to be honest.

No I agree Darth Ender , it's not surprising but we have seen these type of issues before when a new platform is introduced. I have to say I feel the lack of information and errors in the MB manual makes me think there was insufficient time when Ryzen was released for ASUS to address the platform properly and it was rushed. My previous AMD platform was far more comprehensive with Bios information than this one. There's still much to do with Bios Micro Coding however I have the virtue of patience.
It would be nice to have a proper Bios manual made available in E Format when final Bios is sorted.

I've now loaded up a couple of my AAA games namely, "Elite Dangerous" and "Rise of the Tomb Raider" and these are some initial impressions.
I'm not going to compile graphs to compare CPUs between AMD and Intel but rather improved performance between this platform and my previous FX-9590 build.
Using Fraps to gauge my fps, I found Elite Dangerous running well as expected with a 10-20fps improvement.

Rise of the Tomb Raider I found little change between my previous GTX 980 sc and FX-9590 combo however, some settings in that build were turned off. Mainly MFAA, Antialiasing and Anastropic filtering.
Both games are smoother and I found the 1080 Hybrid card running much cooler than the GTX-980.
No bottlenecks in either games and the Ryzen build is much cooler than my previous build Overall.

Loading times and Windows performance in general are faster using the 960 M.2 drive as compared to my 850 EVO drive.
I guess time will give more accurate comparisons and I have to say at this point, I'm absolutely stoked with the performance so far.

[Darth Ender]

Good, glad you have the issue sorted. My sleep issue was resolved with an update to my mouse and keyboard drivers and I disabled any tree hugging feature I could find.
I was using the MS default drivers and system would reboot when using either to wake from sleep and probably due to signal state not interacting with PCH correctly. Not 100% sure what the fix was to be honest.

No I agree Darth Ender , it's not surprising but we have seen these type of issues before when a new platform is introduced. I have to say I feel the lack of information and errors in the MB manual makes me think there was insufficient time when Ryzen was released for ASUS to address the platform properly and it was rushed. My previous AMD platform was far more comprehensive with Bios information than this one. There's still much to do with Bios Micro Coding however I have the virtue of patience.
It would be nice to have a proper Bios manual made available in E Format when final Bios is sorted.

I've now loaded up a couple of my AAA games namely, "Elite Dangerous" and "Rise of the Tomb Raider" and these are some initial impressions.
I'm not going to compile graphs to compare CPUs between AMD and Intel but rather improved performance between this platform and my previous FX-9590 build.
Using Fraps to gauge my fps, I found Elite Dangerous running well as expected with a 10-20fps improvement.

Rise of the Tomb Raider I found little change between my previous GTX 980 sc and FX-9590 combo however, some settings in that build were turned off. Mainly MFAA, Antialiasing and Anastropic filtering.
Both games are smoother and I found the 1080 Hybrid card running much cooler than the GTX-980.
No bottlenecks in either games and the Ryzen build is much cooler than my previous build Overall.

Loading times and Windows performance in general are faster using the 960 M.2 drive as compared to my 850 EVO drive.
I guess time will give more accurate comparisons and I have to say at this point, I'm absolutely stoked with the performance so far.

So cpu frequency scaling in linux is currently done through an acpi driver for amd chips and I could never seem to figure out how to get linux to see custom p-states. It always seems to see some kind of default list of p-states. In the case of the 1800x, 2 p-states (top stock speed and something roughly half that). But i have no idea what the voltages or anything are related to those two states. They're not setup in the bios that way.

So when i recompiled my kernel, i didn't compile cpufreq in at all. And to add to that, I didn't compile in core power management features either. But i do have acpi compiled in as it's required for smp.

So far, everything is great in linux. I'm betting whatever these two p-states that linux is seeing with this acpi-cpufreq driver is the culprit. They're likely set with too low of a core voltage and idling too long overnight is causing the cpu to become unstable and crash/reboot.

No big deal to me. I'm using 110Mhz bclk and 36mult and temps are under 40C idle (according to linux sensors ...which I have not configured any offsets for). Max seems to top at under 60C.

This is essentially equiv in performance from what i've tested to when i had 100Mhz bclk and 40mult, but at a much lower temperature according to sensors. Like significantly lower.

[BlackX]

Why I can't set the Q-Code to show the CPU-Temp like on other ROG Hero boards?
Now its shows always a boring 24.

[MeanMachine]

Why I can't set the Q-Code to show the CPU-Temp like on other ROG Hero boards?
Now its shows always a boring 24.

Hi BlackX As far as I am aware, only the Impact MBs were able to show temps on the Qcode LED and I personally would much rather consult with my Bios for temperatures and fan control.
You mention a boring Qcode 24. I have no idea what that is as it's not documented. Are you not able to access Bios.?