I just tried it out – disabling Core Performance Boost caps the frequency at the base frequency (4.5 GHz for a 7950x). The benchmarks from puget are a bit misleading perhaps since they are all multithreaded… On a single tread you lose a lot of performance like this… 5.5-5.7 down to 4.5 is a big gap, depending on the application. Even to 5.0-5.2 GHz multicore down to 4.5 is a pretty big dip.
IMO if the system is unstable with CPB enabled – given decent cooling etc. – the CPU should be RMA’d. Capped at 4.5 is too much of a limitation for many workloads.
Setting “Core Performance Boost” to Disabled criples my CPU by limiting it to it’s Base Clock (4.5 Ghz, see screenshot below)
When I booted BIOS for the first time this option (“Core Performance Boost”) was set to Auto (which seems to be the same as “Enabled”). If I “Load Optimized Defaults” (i.e. reset BIOS settings to “safe” defaults) it sets “Core Performance Boost” to Auto (which again seems to behave the same as Enabled and which allows CPU to run beyond Base Clock but still lower than Max. Boost Clock of “Up to 5.7 GHz” so that’s why I’m reluctant to call it OC. OC would be beyond 5.7 GHz)
Let’s wait for my 7600x to arrive, I’ll test and see. Will post results here
I agree with you @quilt , 4.5 limit is not ok for me. Regarding cooling, I use Noctua D15. So it’s not watercooling of course, but I consider it to be a very good air cooler which should provide room even for overclocking (which I don’t want nor plan to do)
Man, I’m already way over the budget on this system… buying a 2nd CPU (albeit a cheaper one - 7600x) just to confirm whether the 1st one is faulty. If I have to buy another RAM/Motherboard after that I’m gonna get broke (or build a 2nd PC)
I avoided Intel for this build due to stability/degradation concerns and look where that got me
I am using Pop OS on a AM4 platform. Using HDMI caused my DE to randomly freeze. I have a laptop using Pop and tested HDMI with a 4K monitor and no issues. I have recently discovered Nvidia server was installed and removed. My system has been up for 20hrs now no issue. I would check to make sure you do not have any Nvidia software installed on Linux since that is where your problem is occuring
Just to clarify, if you connect your monitor to iGPU (preferably HDMI on the back of your motherboard) you can use your operating system (Windows? Linux?) without any problems such as blackouts, freezes? What are your BIOS settings (anything special?). What’s your motherboard and RAM?
I checked on my own system and CBO is only offers auto or disabled options, with default being auto and no description whatsoever.
Reddit anectada matches you observation, that disabling it stops core from boosting above baseline frequency.
I would return that CPU upon confirmation that any other work fine with default settings. If amd defaults setting are this unstable on this unit, there is nothing constructive to do here.
Mine is rock stable once expo + training memory related instability was sorted, but I dint use igpu much.
EDIT: Is there some test scenario that reliably induces igpu crashes on your end, something quick thats not gaming? I can try it on my end.
furmark@1080 run like shit and grimdawn is slideshow, but that is expected. No crashes, but I can see choppines and delays on windows gui.
Yes I’m using the iGPU without issues (well some issues but those were due to the driver), currently on the USB4 port but previously on HDMI.
Motherboard is the proart x670e, memory is a GSkill 96GB 6400 kit that I run at 6000 (IMC limitation). Before I had a Kingston 32GB 4800 kit that also ran without issues (since fast dual rank kits were outrageously expensive when I built the system).
BIOS nothing special except for memory (manual settings since my CPU can’t do 6400 1:1, and 6000 1:1 is faster than 6400 1:2). Limited the CPU to 200W PPT since this is what my air cooler can dissipate without trottling; default 230W works as well but I preferred consistent slightly lower performance.
Yes also here auto (=default) is enabled. I would think that means it’s also AMD’s default and IMO should be stable on any non-defective CPU.
I assume you mean that you have monitor connected to Type-C USB4 port on the back? (which implies that you have that short DP cable connected from DP-out on your discrete GPU to the DP-in on your motherboard) ? If so, then you’re not actually using iGPU
No, I don’t have the DP cable connected. If you look at the manual (assuming this behavior didn’t change between 670 and 870), when no DP input is connected both USB4 have the iGPU signal, and if it is connected one USB4 has the iGPU signal, one has the dGPU signal
Today I received my 7600x which I ordered to troubleshoot this problem. Took out 7950x, placed 7600x into the socket. iGPU works perfecty. No blackouts, no freezes. Neither in Linux Mint nor in Windows 10. Exactly the same BIOS settings, the same HDMI cable and monitor. The same RAM and motherboard. The only difference is the CPU. Now I will try to RMA it. I swear to God, if AMD refuses to replace it I don’t know what I’m going to do…
Short rant: I recently bought an 18TB Seagate HDD and it wasn’t working. I returned it to the manufacturer and they told me “yeah, one of the components inside has been dislocated which could happen if there was a hit. This is not covered by the warranty” and they returned it to me without even trying to repair it. I’ve spent 400$ and got no working HDD nor my money back. I hope in this situation with CPU the things will go in my favor especially considering that I went the extra mile and bought a 2nd CPU just to be confirm that the 1st one was definitely faulty.
Wish me luck and thank you everyone, I really do appreciate all your help in dealing with this. Cheers! Love from Ukraine
It is also possible that the CPU was not seated well, or the cooler was not evenly tightened.
You could try putting the 7950x back carefully, examining the backside beforehand for anything that could prevent good contact… But I understand if you do not want to do that.
Overtightening or undertightening the cooler can also cause issues.
Oof, how was it packaged? I’ve ordered hard drives from different places and they have different packaging standards. Some use special HDD boxes with foam that snugly fits the drive, some just throw it in with typical paper filling in the plastic caddy. With some luck you can complain to the retailer…
Yeah, they should replace it. But who knows… You could still try reseating as @TObject suggests but I’d guess bad contact would show up otherwise too. One other option is power delivery (if you have a crappy PSU?), but otherwise it’s settled IMO.
Ok, I wouldn’t be able to go to sleep without trying it out. It would keep me awake =)
So I’ve put 7950x back into place and… issues came back immediately (blackouts, freezes, logouts). Then I put 7600x again and… all issues are gone. It works flawlessly.
At this point I think it’s safe to conclude that I’ve received a faulty CPU.
In the worst way possible. I’ve never ordered HDDs from Seagate before, only WD or HGST and those had solid packaging. Seagate came literally ‘as is’ in a simple anti-magnetic bag. I should’ve refused to accept it. Of course it might get damaged, it’s a hard-drive. There should be at least some packing/box to protect it during shipping. I was so mad about it, if fact - I still am.
Well, a new CPU (7600x) works fine and other components (two GPUs) are also good. PSU should be solid, it’s brand-new from Corsair (HXi 1500W). It’s almost certainly a bad 7950x, fingers crossed I’ll get a replacement.
So Seagate themselves are the seller… Very disappointing… Hopefully you have pictures of the packaging and can complain about that. I’ll note no never order directly from Seagate!
That should be easily enough power, so higher power use due to core boost should not be an issue.
