Discussion in 'Processors and motherboards AMD' started by Hilbert Hagedoorn, Feb 3, 2021.
I see the same in 2.1 RC5. Profiles do not survive a PC reboot.
Hello, Any knows.
Whichone is this particular improvement?
"support for unique dual CCD 5600X and 5800X processors"
I have one of those CPUs (5600x) .
I would like to know if I can get an extra performance?
Reporting similar issue, but found a way to fix for now.
5900X as well and started getting the same thing in logs and severe instability (random restarts using DaVinci Resolve; gaming I did not try).
What worked for me was clearing config, setting values for profiles P1 and P2 manually and recalculating PX profile (because I didn't remember the exact values for MIX/AVX presets).
So far so good (~20 hrs and was able to work fine in Resolve), but haven't yet manually restarted the system to see if that holds.
asus x570 crosshair VIII dark hero (with EKWB cpu+vrm monoblock)
corsair vengeance pro 4000mhz 19-23-23-45
current stable(ish) settings
PX HIGH 4950hz 1375mv MID 4875mhz 1375mv LOW 4800mhz 1350mv
P2 VID-1260mV CPUusage-28% CCX1-4725Mhz CCX2-4575Mhz
P1 VID-1110mV CPUusage-79% CCX1-4475Mhz CCX2-4325Mhz
Memory DDR4-3600 18-21-21-41 CR1 1.4V
North Bridge Clock 1800Mhz
cinebench R23 28624/1616 17.71x 143.05w/57°C (cpu package watts/temp)
cinebench R20 11246/633 17.78x 148.84w/58°C
cinebench R15 4839/269 18.00x 235.07fps 140.85w/57°C
cinebench R11.5 52.84pts 131.24fps 139.12w/54°C
room temp 23.2°C
cinebench R23 29169/1623 17.97x 140.49w/57°C
cinebench R20 11207/633 17.71x 141.02w/59°C
cinebench R15 4854/270 18.01x 254.91fps 139.14w/57°C
cinebench R11.5 53.75pts 126.65fps 138.74w/55°C
room temp 25.1°C
a bit of explanation :
- the tuning feature for me always ends the same way, screen goes dark, pc reboots, ctr launches with no settings no data no memory nothing, never succeeded a run
- some of those values I got from the manual diagnostics/calculate profile I then raised towards numbers that I saw while using pbo etc prior to using CTR
- I recommed doing things simply, calculate profiles manually thhen start with those, CTR generates optimized performance/watts that should be more stable than what I use
- if you can't complete a diagnostic run something in your bios or memory settings is unstable
- the PX profile often gives me weird values so those too I "corrected" (I often have the low clock higher than the mid)
- all the bios settings I tried to change made it worse, especially LLC 100% crash if not set on auto, the only bios changes I have now are the memory settings
- I still have a problem now is that my gpu often "loses it's signal" monitor screen goes dark, apparently a known RTX 3000 bug but one that gets worse if I raise my memory to 3800/1900 18-22-22-43, I haven't given up on that front it boots it benches it's only with games or gaming benchmarks that I have a problem
I done few tests last night and today with CTR 2.1 on my build with 5950X on Asus ROG Crosshair VIII Hero X570, CPU waterblock is Aquacomputer Cuplex Kryos NEXT
I experienced I think 3 WHEA during the running CTR every time, I'm suspecting as well RAM or FCLK/IF, all settings I have set as per guide etc but still PC would crash,I tried just run Hybrid OC and literally I got WHEA error again and reboot
In my case CTR is unusable with 5950X, using PBO 2 with Curve optimizer and seems PC is quite stable and no issues
Hope this helps
I might have to try your settings jura11 I benchmarked 2 days all day long was working fine and today...crash-reboot nonstop I changed nothing
edit : nope got exactly the same problems using PBO enabled + curve optimize enabled
cinebench doesn't start or runs very slowly, black screen reboots etc...
I rolled back to previous motherboard version 3501 beta -> 3401 we'll see
What settings do you have enabled in BIOS?
Personally I would enable these settings:
These BIOS settings are not my but they should work for you, I'm using them with 5950X and PBO 2 with Curve optimizer (-20)
Hope this helps
I am on 2.1 RC5 also, and profiles are not enabled after reboot. I have to manually activate them again.
update of the day :
- thank you very much jura11 will try that later
- I fixed my black screen/reboot/no video signal crashes by raising my P2 (1260->1300mv) and P1 (1110->1200mv) voltages and it worked...this far, I should probably lower the P1 I'll try 1150mv I guess
ctr sometimes doesn't save your settings and weirder I saw it change them back live while I was changine a value in another cell
I suggest : deactivating all profiles 1st -> changing the values and switches like autoload -> saving all profiles -> exiting the app -> relaunching the app
that way it keeps the values I manually enter, the % especially have a tendancy to rollback to what you had previously
Thank you for your answer, I will try
Unfortunately that only worked for that time only.
I don't exactly remember what I did, but I noticed that running CoreTemp (www.alcpu.com/CoreTemp) started the same (PCI-MUTEX holding time has expired! Sending a command in an insecure way!) messages in CTR 2.1 and crash-reboots while in DaVinci. So this one's a suspect (might be coincidence), but I haven't had the time to test further.
Right now not having CoreTemp start doesn't help either - there are no messages, but I get crash-reboots anyway. I did try clearing config in CTR 2.1, setting all profiles manually or recalculating only PX several times. Haven't done a full new tune yet, don't know if that's really necessary.
And the profiles weirdness others are talking about is true for me as well now.
Have CTR 2.1 disabled for now.
Got an idea to clear config with "CTR CONFIG CLEANER" and then in CTR 2.1 too and go on testing different software and profile combinations, but I don't have time for that right now.
Don't know if it's relevant, but according to CTR results my 5900X is a golden sample with 4,17 efficiency, CPU VID/TEL 1065/1048.
I had thought to do the same. I reset the config with the tool, delete the file, do a full diagnostic and still the same error messages.
Suddenly, I have no more ideas. I have an Asus motherboard so I put LLC level 3 as recommended. I just have my RAM in D.O.C.P and activated the rezizable bar.
I tried using MIX and AVX presets. I have a bronze / silver sample
5950x asus crosshair VIII dark hero corsair vengeance pro or dominator memory
as I may have said in a previous post, touching LLC only gave me crashes, it made me crash way more often, it's a really a no-no for my PC
another thing I found yesterday on asus, PBO can be enabled in two different places on the tweaker page and also in the amd overclock part the one with the warning
when I reloaded my "nothing else than XMP" profile CTR wasn't activating voltages and clocks seem to be random..turns out reloading the profile left pbo activated in the amd overclock section it didn't roll back that change I loaded optimized defaults and chose the xmp auro overclock thing and thats it now, my bios is "optimized defaults" outise of that, CTR works "fine" now
well fine for me who's mostly gaming and encoding videos but....cinebench is often slow and the profiles don't seem to activate properly I get like 4950mhz @1.4v and 78°C ><
In my case I don't have any issues with the LLC, running at LLC 3 and no BSOD to report, but I resigned to use CTR for now, I'm using PBO 2 with Curve optimizer and in single threaded apps I can see boost up to 5250MHz
If you are getting BSOD I would disable C-States and depending on the PSU I would try Power Supply Idle Control = Low Current control or try it off
Yup on Asus you have in two places PBO, like one in Tweaker page and AMD Overclocking, personally I have enabled in both
What PPT,TDC, EDC do you have set in BIOS ? I have set 252 for PPT, TDC and EDC at 175 and temperatures in Cinebench are under 70's
Hope this helps
Running latest bios on my Asus Strix 550-E
5900x silver sample
On curve optimizer stable at -30 all cores but -25 on core 2 and core 9
But idle temps with curve optimizer were around 50C, so I thought I should give CTR a shot.
Issue: Even with optimized defaults, DOCP, CPU LLC3, Diagnostic run keeps on crashing the system.
Event log shows WHEA on either APIC 18 or 19 for every crash. It seems that CTR is trying to undervolt this CCX a bit much and ends up in a crash.
For now I am unable to use this tool I have heard so much about.
Please advise as you see fit.
UPDATE: 4th time is the charm, diagnostic completed successfully, now running tuning.
UPDATE2: Ran tuning, saved profiles. And the MT performance is just as good as my PBO2 CO, but single threaded performance is much much worse. Yes, I enabled hybrid OC, and saw 'rocket' next to PX profile when running ST tests, but performance was nowhere near my BIOS tweaks.
I think I am ok without CTR for now.
I did lots of tests again.
I tried with the AVX or MIX presets.
I have disabled PBO at all places in my BIOS of my Asus B550-F motherboard.
I tried with or without LLC 3.
I still have this message from LOG:
PCI-MUTEX holding time has expired!
Sending a command in an insecure way!
It wouldn't be bothersome if the crashes didn't come with this message. Sometimes I have crashes just when launching an application.
I have the impression that crashes occur during peak CPU usage.
If only I could figure out what this log message is, maybe I could fix problem there ...
Looking more closely, I see that the CPU often exceeds 1.4V while in the profiles there is no question of such voltages, is this normal?
PCI-MUTEX holding time has expired!
Sending a command in an insecure way!
Hi, i have a problem with CTR. Yesterday i made the disgnosis and said me for undervolting 4100 MHz and 1075 mV and i go to TUNE. after 10-15 min, it doesnt end correctly. Today a repeat the diagnosis, and values changed, say me 3800MHz and 1280mV (I dont remember exactly the values, but like that), it´s true that i changed two options of BIOS that i forgot to change CPPC and power c control. I dont understand why values are diferent and how can i do to make it better. Thank for reading this and i wait your answer.
D4mocl3s I would try disabling PX profiles and or lowering them, what I did personnally was to reduce PX clocks and voltages so that the PX low matches the P2 and tried to lower P2 and P1 percentages, that said I still have that weird half the time my cpu tries to run cinebench @ 4950Mhz 1.4v and gets really hot while bein super slow
no it's not normal there's something that doesn't "switch" right or switches back mid-usage between profiles, in a normal functioning way if you limit the amount of threads to your profiles % you should get the clocks and voltages programmed (in the settings of cinebench you can choose how many threads you want to run that's how I test my CTR % activation of P2 and P1)
@kakiharaFRS I am going to try by deactivating the PX profile but it is a shame to have to do without it because it is for this novelty that I went from CTR 2.0 to 2.1.
EDIT : It didn't change anything, still the same log messages!
Using Ryzen 3600, currently I set PX = P2 profile as PBO doesn't do better than 4250 MHz at 1350 mV.
What do I do with this? Was I right to just set PX = P2?
CTR BOOST TESTER RESULTS (test version)
CORE / FREQUENCY / VID / POWER / TEMP
C01 F 4037 V 1.406 W 10.19 T 63.11
C02 F 4036 V 1.402 W 9.93 T 63.44
C03 F 4048 V 1.406 W 10.3 T 66.68
C04 F 3998 V 1.394 W 9.82 T 63.56
C05 F 4068 V 1.41 W 10.44 T 64.5
C06 F 3992 V 1.392 W 10.07 T 64.32
CPU TOPOLOGY (ENABLED CORES)