Processors

[ElBatto]

Good day!

I put together a new PC 2 months ago. I currently have the following hardware:

- TUF Gaming X670E
- Ryzen 7 7700x
- Kingston Fury RGB 2x16 6000 GB EXPO version
- NZXT z63
- Corsair rm1000x power supply

At the beginning I had extreme stability problems. The computer just shut down without warning. Usually, however, only e.g. when playing games.

After a bios reset by taking out the bios battery, the system was again much more stable and the hardware was recognized correctly. For whatever reason.

Then I decided to test the hardware more extensively. Memtest86 via a bootable USB stick showed no problems with the RAM. With Prime95 and OCCT tests for the CPU everything ran well.

However, my CPU got up to 95°C in most games. Which is normal under full load, as I understand it.

I tried all of the above tests with EXPO 1, EXPO 2 and without EXPO and everything ran fine. The PBO setting was always set to "Auto" by default.

However, I would like my CPU to stay a bit cooler during operation. So I set the PBO setting to "Enhancement" and set the temperature limit to "Level 2 80 °C".

Now I have tried the above CPU tests again. As soon as the test starts e.g. OCCT shows many errors in the physical core 4 (virtual core 8 and 9). Always only core 4. And after a few seconds my computer just goes off.

Whether I have Expo 1 or 2 on or off makes no difference here. I can fully utilize the other cores individually during the tests, and everything runs smoothly. Only core 4 is spinning its wheels. Whether I set level 1, 2 or 3 in the PBO makes no difference.

I tested the following PBO settings and they are stable: Auto, disabled, AMD ECO Mode (Auto). I did not test the other Eco modes.

I would like to use the temperature limit. Anyone have an idea what this could be?

Oh yes, latest Bios (Vers. 1616) and latest drivers are installed.

Greetings El Batto

[davidoliver]

The Ryzen 7 series processors are typically denoted by numbers like 3700X, 3800X, or 5800X, among others. If you are experiencing PC crashes with a Ryzen processor, it could be related to other factors.

To troubleshoot PC crashes, especially when using Precision Boost Overdrive (PBO) enhancement, here are some general steps you can take:

1. Ensure adequate cooling: PBO can increase the power and performance of your processor, leading to higher temperatures. Make sure your CPU cooling solution, such as a CPU cooler or liquid cooling system, is sufficient to handle the increased heat generated by the CPU.

2. Check system stability: Run stress tests and benchmarking tools to check the stability of your overall system. Tools like Prime95, AIDA64, or MemTest86 can help identify any stability issues related to the CPU or memory.

3. Update BIOS and drivers: Ensure that you have the latest BIOS version installed for your motherboard, as well as up-to-date drivers for your CPU, chipset, and other hardware components. Sometimes, BIOS updates can improve compatibility and stability.

4. Adjust PBO settings: If crashes persist, you can try adjusting PBO settings in your motherboard's BIOS. You may have options to enable or disable PBO, adjust power limits, or modify other related parameters. Experiment with different settings and observe the system's stability.

5. Monitor temperatures and power consumption: Use monitoring software, such as Ryzen Master or HWMonitor, to keep an eye on CPU temperatures and power consumption. Excessive temperatures or power draw could be indicative of issues, and you may need to adjust cooling or power delivery accordingly.

[ElBatto]

Hello!

1. problems with the temperatures I can safely exclude. The water cooling is in any case powerful enough and also HWinfo64 and other monitoring software shows that all temperatures are fine.

2. the tests I have already run as described. That is why I write among other things. Memtest86 did not throw any errors via a bootable stick. Prime95 and OCCT show errors when PBO Enhancement L2 is active.

3. I have updated the Bios a few days ago to V1616. The errors existed before and still exist.

4. here it is now about. With activated PBO L2 I get first in the core 4 errors. Then the other cores 5, 6, 3 and 2 follow. The cores 0, 1 and 7 have never reported errors so far. If I now change e.g. in the Curve Optimizer core 4 to positive 2, I get no more errors with the state now. I tried this with the other cores in a similar way, but unfortunately it did not lead to the desired success.

After changing the CurveOptimizer the reboot did not work correctly. First the CPU LED lit up, then the DRAM LED and I had to manually disconnect the computer from power. After that, the computer boots into the safe mode of the bios and I can make settings. With PBO Enhancement L2 disabled, these errors do not occur. I'm not familiar enough with the subject to know which values I have to change so that these errors no longer occur.

5. temperatures are monitored and are ok. The power consumption is also within the limits. I have revently replaced the power supply, because I thought the errors come through my 800w power supply. Now I have a 1000w power supply, same problems. The cables are also all connected correctly.

I'm beginning to suspect that the CPU is not 100% functional. By the way, I also replaced the DDR5 RAM as a test, but it didn't change anything when PBO Enhancement L2 was active.

Currently I have PBO AMD Eco -> Auto activated again. Under this mode, as I said, everything runs wonderfully, even with activated EXPO 1.

However, I would like to use PBO Enhancement L2.

[L0rdOfDarkn355]

I might not have a solution for your problem. But I encountered similar issues with my new AM5 built.

It's only that I have a 7600x (Air Cooled with Dark Rock Pro 4) on a PRIME X670 Board with 2 x 32 Gb Kingston RAM.

I built the system for some gaming instead of using my 3 year old gaming notebook. Initially I did some (I'd say smaller) settings in BIOS like activating EXPO Profiles for RAM and setting Thermal Throttle Limit to 85 (not the PBO Thermal Limit though!) and did quite some testing of the system.

Different Tests in 3dMark, Full Ram Tests with Windows 11 Memory Diagnostic and memtest86Free (10 hours run or so), Cinebench, FurMark, Heaven, Prime95; everything without any issues...

But in games I encountered crashes (basically in the way that the games just close to the desktop out of nothing). Most of the times there where events to the crashes to be found in event view. Games were e.g. Jurassic Word Evolution or PC Building Simulator 2. Funny thing is, that on my older notebook I can not remember such crashes. Sure sometimes games do crash... but on the new pc it was basically every time when playing those games. Game Benchmarks like AC Odyssey/Valhalla, Horizon Zero Dawn where fine though. Sure... the notebook does by far not max out any of the graphics settings.

I am now back to optimized settings and a fresh windows setup with all drivers up to date and doing some further testing. I even left memory on default without EXPO profile or so. And did a second memtest86 run (this time with pro version) with default run (all tests except test 14) for 4 passes... no errors found.

I know decided to do some tests with OCCT after I changed PBO to Enhancement and set PBO Thermal Limit to 70% (i found some information saying, that the performance loss with Level 3 on 7600x would be neglectable).

With these settings I did a standard Cinebench All Core Run; and HWInfo was on 70° CPU. But OCCT CPU Stress Test immediately gave many many errors until short after PC went down with BSOD DRIVER_IRQL_NOT_LESS_OR_EQUAL.

Now I am back to BIOS optimized defaults and OCCT CPU Stress Test just finished without any errors... CPU temps where up to 93 or so according to OCCT... well...

Maybe I will do some Games now with BIOS defaults, because it seems that Benchmarks are ok... also different 3dMark Stress Tests (not the Extreme Ones because I am not on 4k anyway) went through... CPU Stresstests will easily hit the 95 degrees so maybe I experiment with Q-Fan... Armoury Crate is just sh... because apart from FanControl and Updater I would not need anything... I have a Black Fractal Design Case... what do I need LEDs for?

I still wonder if there is any faulty hardware in the system or if the AM5 systems still are not fully mature.

Ah yes... and I use a 1000W Gold Thermaltake PSU...

[ElBatto]

Yes, sounds very similar to me.

I don't seem to have the problems any more. But I will observe the whole thing for a longer period of time.

In the meantime, I had tried everything possible in the bios. Under standard settings it ran well for a few hours, but then I still had random pc failures from time to time. I had now reached the point where I was gradually sending back all the components to test them individually.

First, I sent the RAM to Kingston and they sent a new kit without any fuss (although I think the RAM was fine). Then I sent in the motherboard and they refunded the price. I then bought the same one, but with integrated WIFI, as the network card I had before generally liked to cause problems. But the errors were still there.

Then I started an RMA procedure with AMD on 19.06.2023. After a bit of back and forth, they approved the procedure and I sent in my CPU, and this Monday my new CPU arrived.

With the new CPU, I had no errors after 5 hours of OCCT CPU stress test, with standard settings and with PBO Enhancement level 2 (80 °C), with and without Expo. After 5 1/2 hours, however, my PC apparently just shut down once (OCCT Stress test was still running), but I think that's more likely due to possible Expo and continuous stress testing over the long time.

As long as I can play performance-intensive games unhindered over a long period of time, I'm fine with it and I'll look into the remaining problems in bit by bit.

[L0rdOfDarkn355]

Ok. Right now I am not into changing parts. However... I did another OCCT CPU Test after setting both Precision Boost Overdrive to "Enhancement" and Thermal Limit to "Level 2 (80) and also setting CPU Boost Clock Override to "Negative 200". With this the OCCT CPU Test processed 30 mins without errors. I will do further checks with these settings. I also did a BIOS update on my ASUS Board (version 1636). I don't know by now whether this update has any influence on the tests. But one of the improvements of this BIOS version was memory compatibility. So... who knows? I know, that my RAM was not on the earlier qvl lists (I checked the list befor buying but somehow I have selected the wrong one during ordering) although none of the benchmarks or memtest86 gave any hints of bad memory... Maybe ram compatiblity is related to this OCCT issues?

[ElBatto · Accepted solution]

I can now safely say that the errors were caused by a faulty CPU.

I have been using my PC regularly under full load for over a month now, currently always with PBO level 2 + Expo level 1, without even one error/crash.

OCCT also runs without error messages.

With the old CPU, I was also able to get my system a bit more stable by playing around with the Curve Optimizer, but never found the perfect settings. I mean that I could sometimes use the computer for up to 2-3 hours.

At the time, my system was most stable with AMD ECO activated and Expo switched off.

By the way, my original vendor tested the CPU with programs (e.g. Prime95) and then sent my CPU back to me with the comment "Everything is fine". The AMD RMA service accepted the CPU as faulty and replaced it, but I didn't get a replacement CPU in the meantime.

I think I understood that the OCCT CPU test, if you select Large for the file size, also loads/demands the RAM memory. The errors can then also be related to the memory rather than purely to the CPU. I think this is explained in the info window of OCCT when you hold the cursor over the settings.