Overclock.net banner

1 - 11 of 11 Posts

·
Registered
Joined
·
1,236 Posts
Discussion Starter #1
Title. Long ago i decided that i won't care as long the performance is there but i wonder if theres someone out there that actually gets 4.4ghz at all (even at idle) in hwinfo effective clocks section. The max i've seen on my sku (early august 2019 ryzen, not those god-like march 2020 bins) is 4.325.
 

·
Registered
Joined
·
2,174 Posts
I have one from first batch but as during that time MB had several BIOS and AGESA versions things have been changing in between them. With first 3000 compatible BIOS and AGESA 1003abba (that's when I switched from 2700x) occasional boost (with PBO) went over 4.4GHz on one core and 4.3GHz on the rest of them but it came with a price of high (1.4v+) core voltage which my cooler soaked up.
Next BIOS with 1004b lowered max boost to just under 4.4GHz but default voltages dropped to (max 1.325v core) but benchmark scores went up.
Last BIOS 5601 and AGESA 1006 again allowed higher core voltage near 1.4v but 4.4ghz doesn't reach I suspect because temperatures near 70c. On occasions when temperature was kept at 65c max frequency went up slightly over 4.4GHz. Temperature seems to play large role when auto boost is concerned.
I have changed CPU cooler to present one about a week ago and had some problems with it so I had occasion to observe temperature to frequency interaction.
Temps hit 83c and max single core frequency dropped to under 4GHz with other cores barely over 3.5GHz.
Fixing cooler somewhat and with temps up to 73c. frequency hit 4.235GHz max.Fixing cooler to present state and on a hot day frequency at 70c is goes up to 4.398GHz with 1.327vcore although with negative voltage offset -0.03v which is highest I can go without compromising stability. CB r20 score returned back to normal 5000+. At 83c it was dismal 4300+.
Soooo... my old adage "Keep it cool fool" is vindicated. temperature is main key for Ryzen auto boost with ties to keeping Vcore in check.
As for manual all core OC, I never managed stable OC over 4.35GHz no matter which voltage. When I had courage to push it to 1.5-1.55v, 4.4GHz wasn't stable at all, it would pass manage couple of passes with CB r20 but scores dropped ! and other times locked up although temps never went over (relatively) safe 85c).
 

·
WaterCooling Enthusiast
Joined
·
395 Posts
I just bought a 3700X that will hit 4.425 on 2 cores and 4.4 on 4 others with PBO set to auto. It seems based on posts I'm seeing that a higher percentage of newer chips can hit 4.4 and 1900 Fclk.
 

·
Registered
Joined
·
1,236 Posts
Discussion Starter #4
Hwinfo effective clocks, guys. Mine even hit 4.45ghz but only in normal clocks section and early bioses
 

·
Registered
Joined
·
27 Posts
I am right there with you. I cannot "max" at 4.4Ghz to save my life. I tried



Clearing CMOS
enabling PBO
Tried it with it PBO "Auto"
Ryzen Balanced Power Plan
Ryzen High Performance
Updating bios and chipset to latest


Closest I can get in HWinfo is 4.35Ghz...


Are you saying in Ryzen master it will show differently?
 

·
Registered
Joined
·
294 Posts
....

Are you saying in Ryzen master it will show differently?
"Effective Clock" is an artificially developed clock speed report in later versions of HWInfo64 that considers the effect of things like imposed advanced sleep state. Here's a quick blurp from Martin (HWINfo author) on it:

https://www.hwinfo.com/forum/threads/effective-clock-vs-instant-discrete-clock.5958/

I think that it would be pretty hard to make effective clock equal max clock since Ryzen processors (especially Zen 2) go into advanced power saving sleep states so frequently and quickly. I've not tried a fixed multiplier overclock in a long time but as I remember Ryzen still went into sleep states even then. Is there any way to completely stop it?
 

·
Registered
Joined
·
2,174 Posts
"Effective Clock" is an artificially developed clock speed report in later versions of HWInfo64 that considers the effect of things like imposed advanced sleep state. Here's a quick blurp from Martin (HWINfo author) on it:

https://www.hwinfo.com/forum/threads/effective-clock-vs-instant-discrete-clock.5958/

I think that it would be pretty hard to make effective clock equal max clock since Ryzen processors (especially Zen 2) go into advanced power saving sleep states so frequently and quickly. I've not tried a fixed multiplier overclock in a long time but as I remember Ryzen still went into sleep states even then. Is there any way to completely stop it?
One way to solve it is thru BIOS, disable PBO, ERP and Power saving modes. Another one to try is in Power options > Razen Power saving > Minimum processor mode 100% in Windows.
In HWinfo, to see more fleeting highest maximum frequencies, Polling period can be adjusted to much lower number, 200 is enough.
 

·
Registered
Joined
·
294 Posts
MishelLngelo;28547634... Another one to try is in Power options > Razen Power saving > Minimum processor mode 100% in Windows. ...[/QUOTE said:
I don't have anything like that in my BIOS and tried everything else. I'm not worried about it because the description of how 'effective clock' is derived makes it seem very artificial anyway.
 

·
Premium Member
Joined
·
4,956 Posts
Mine seems to boost ok'ish.

PBO with limited PPT, TDC and EDC settings.
 

Attachments

·
Premium Member
Joined
·
4,956 Posts
Could you explain a bit more? You have PBO enabled? What setting is limited PPT, TDC and EDC?
Yes PBO+auto OC enabled

Limited as in not the max numbers you can put in.
PPT = 130W
TDC = 76A
EDC = 115A
Scaler = 4x
Auto OC = +100mhz

This way it can boost up to 4450mhz
If i just max PPT, TDC and EDC it will only boost to 4400mhz.

Also this way it runs full speed in CB20, Testmen5, games etc etc but limits AVX workloads like in P95 a little and heat in those AVX workloads is actually pretty low.
 
1 - 11 of 11 Posts
Top