Overclock.net banner
781 - 800 of 1,784 Posts
Don't check the bclk on main page of cpu-z . enable virtualisation , then check your memory speed on the memory tab of cpu-z . it will report lower than it should be .
disable virtualisation and it'll be what you set in bios .

Checking in aida64 , go to tools - cache and memory benchmark . and fsb will be reporting at 97.7 .

if you run that aida64 memory benchmark, it will not score much lower compared to full speed memory though . 3200 vs 3133 is very close to each other of course .
Something wrong with your motherboard. Virtualization set at Auto and Disabled shows same bclk - 99.8. I'm not brave enuf to set it to enable. Currently, Virt is set at disabled. Optimized Default sets it to Auto.

Can't you just set it to Disabled?
 
Something wrong with your motherboard. Virtualization set at Auto and Disabled shows same bclk - 99.8. I'm not brave enuf to set it to enable. Currently, Virt is set at disabled. Optimized Default sets it to Auto.

Can't you just set it to Disabled?
I can easily set it to disabled . However , that would mean my windows sandbox will not be functional anymore .


Just updated to the 5216 bios and the issue persists .

Regarding bios 5216 :
-improves quite a bit on memory latency
-lost chipset temperature reading
-lost custom header temperature reading (windows)
-lost CPU_OPT fan speed reading (windows)

As a tip when you flash : when you enter the bios for the first time and reload your profile , go into advanced - USB and set the third option to enabled . For some reason it had set it to disabled for me , so i did not have any working usb devices . Took me half an hour and several cmos resets to figure it out :)
 
  • Rep+
Reactions: rdr09
I can easily set it to disabled . However , that would mean my windows sandbox will not be functional anymore .


Just updated to the 5216 bios and the issue persists .

Regarding bios 5216 :
-improves quite a bit on memory latency
-lost chipset temperature reading
-lost custom header temperature reading (windows)
-lost CPU_OPT fan speed reading (windows)

As a tip when you flash : when you enter the bios for the first time and reload your profile , go into advanced - USB and set the third option to enabled . For some reason it had set it to disabled for me , so i did not have any working usb devices . Took me half an hour and several cmos resets to figure it out :)
Thanks for the tip. Last night, tho, i set my oc profile and forgot the usb setting. Same thing happened and decided to take care of it this morning. Even setting optimized default won't work cos the default is disabled. lol. ASUS needs to fix this crap. Will be informing them shortly.

Anyways, the fix was to use another KB and Mouse. For some reason my existing ones don't get detected. Now, i set both my normal and oc profile with the USB enabled.

EDIT: Also, the latest BIOS does not change any behavior on my 3600. None that i notice.
 
  • Rep+
Reactions: Ipak
I can't wait so went ahead updated to BETA ABBA with ZERO issues.

Found here posted by @Reous provided by Wheity. Thank you!

https://www.hardwareluxx.de/community/f219/asus-prime-x470-pro-am4-1198563.html

Notes:
1. Make sure to set Optimized default first before flashing.
2. Make new profiles for your oc or other settings in BIOS. Do Not use your old profiles.


All Stock with RAM at 3333 MHz XMP.

https://www.3dmark.com/3dm/39399221?
0 issues but it doesn't seem to help with boost either. Are there any other changes in BIOS itself ?
 
0 issues but it doesn't seem to help with boost either. Are there any other changes in BIOS itself ?
You are using ABBA but you are not seeing the advertised boost? I do see it in two cores.

After using it a few days, been noticing how my fans are cycling up and down now. The min vcore for the 3600 using Win Bal (min proc state at 25%) is 0.9v. Used to be much lower like 0.4v. And now jumping to 1.45v sometimes higher.

The added voltage allows the cpu to reach max boost on two cores.

The sound of the fans ramping up is annoying, tho. So, i change the plan to Power Saver with min processor state to 90%. The minimum voltage stays at 0.9v but the max voltage is a bit lower. Fans are no longer cycling. Downside, cpu boosts lower. I never had issue with boost in the first place with ABB.

Not sure if the final version of ABBA will be any different.

EDIT:

Went back to Balance Power with min proc state of 5%. Power Saver only lets the cores boost to 2600MHz in gaming. lol

Have to figure out the Fan plugged to the AIO cos its making the most noise during gaming. Will play around some more to get the advertised boost playing with Windows power plan while achieving a quiet pc. Right now, tho, with this plan the ramping cycle is minimized at idle.
 

Attachments

You are using ABBA but you are not seeing the advertised boost? I do see it in two cores.

After using it a few days, been noticing how my fans are cycling up and down now. The min vcore for the 3600 using Win Bal (min proc state at 25%) is 0.9v. Used to be much lower like 0.4v. And now jumping to 1.45v sometimes higher.

The added voltage allows the cpu to reach max boost on two cores.

The sound of the fans ramping up is annoying, tho. So, i change the plan to Power Saver with min processor state to 90%. The minimum voltage stays at 0.9v but the max voltage is a bit lower. Fans are no longer cycling. Downside, cpu boosts lower. I never had issue with boost in the first place with ABB.

Not sure if the final version of ABBA will be any different.

EDIT:

Went back to Balance Power with min proc state of 5%. Power Saver only lets the cores boost to 2600MHz in gaming. lol

Have to figure out the Fan plugged to the AIO cos its making the most noise during gaming. Will play around some more to get the advertised boost playing with Windows power plan while achieving a quiet pc. Right now, tho, with this plan the ramping cycle is minimized at idle.
No, I'm using 5216 BIOS with 1003ABB only. Trying to find out if that beta BIOS has any other changes except for AGESA code. Best boost I had was with 5007 BIOS. worst with 5216. Most and steady boost is up to 4.25GHz on most cores, 5007 was pushing 1 - 2 cores to 4,37.
 
No, I'm using 5216 BIOS with 1003ABB only. Trying to find out if that beta BIOS has any other changes except for AGESA code. Best boost I had was with 5007 BIOS. worst with 5216. Most and steady boost is up to 4.25GHz on most cores, 5007 was pushing 1 - 2 cores to 4,37.
Oh, this Beta ABBA will make it boost alright. But, your temps will be higher cos of the voltage being higher. Also, your current oc profiles, if you have any, might need tweaking.

I have to apply a negative 0.03 offset on the vcore and keep the 4200GHz boost and voltage from jumping so high.
 
After experiencing high voltages when they are not really needed (1.5v+ at idle) but falling to 1.3v or less under a load which made CPU idle at high 40s and max over 70c at load (which kills the boost), I settled to manual 1.3v + LLC5 and 140% power. Now it boosts all cores to 4.25GHz but newer touches 4.3GHz.
1.3v with LLC at auto, kept 1.3v at idle (35c) but dropped to 1.28 at load, with LLC5 it stays at 1.3v all the time with up to 65c. My take on that is that algorithm controlling boost frequency, temperature and voltages is all wrong, all fighting each other.
BIOS itself needs a good workover, PBO is for instance duplicated, at two places at same time, I spent hours trying to find difference but it's on a level of statistical error. TDP setting doesn't do anything.
Only thing that is in plus is memory, it's very stable at 3600MHZ Cl16 but latency has dropped somewhat which brought memory score down a bit. Tried some RAM OC but it gets unstable (giving BSODs) at 3666MHz with any kind of load. Right now I just set DOCP at 3000 and memory to 3600MHz, FLCK 1800, 1.35v.
 
When I enable virtualisation in the bios , i notice there is a 10% performance hit in 3dmark firestrike(and several other benchmarks) .

The memory also reports a lower speed with virtualisation enabled .

I'm thinking there are several bugs in the latest bios . Like CPU_OPT fan monitoring stops working in windows , PCH temp monitoring same thing . You see them in the bios but in windows impossible to monitor them in bios 5216 .

Other ppl who notice the same behavior ?
 
10%? Are you doing the math right on that?
 
I can't detect any hit on performance by enabling SVM but depending on which VM you use windows may be affected. VM ware for instance leaves some services and threads still running even when not in use at the time.
 
I opened a thread at aida64 forums about it too : https://forums.aida64.com/topic/5178-asus-prime-x470-pro-2-issues/ .
from the forum : "I think with virtualization enabled, the APIC clock runs slower than the BCLK for some reason. Maybe because APIC is virtualized then as well. "
There certainly is some stuff going on behind the scenes when virtualization is enabled . And it negatively affects certain things .

@umeng2002 I call 18049 (SVM on) vs 19810 (SVM off) in 3dmark firestrike a 10% difference indeed .
As you can see , also 3dmark shows the memory running at 3133 instead of 3200 when virtualization is enabled .

SVM ON
https://www.3dmark.com/fs/20456121

SVM OFF
https://www.3dmark.com/fs/20457525

@MishelLngelo I have no idea how the choice of VM would affect my (for example) 3dmark results .. ?
Only VM right now which i have enabled is the Windows sandbox functionality btw .

Running latency memory test always results in a lower latency when virtualization is disabled . (you can check with ryzen calculator latency membench , or aida64 memory bench)
 
After careful deliberation I flashed 5220 BIOS. Haven't adjusted any PBO options yet and preliminary results are great, Some cores reach 4.391GHz which because of BCLK 98.8 can be considered that promised boost to 4.4GHz is practically achieved.
 

Attachments

Sounds great. Can't wait to get ABBA for my X470-F.
I actually achieved 4.400 with AGESA 1.0.0.2 and Windows Build 1909 so I'm looking forward to getting that back.

Will we ever achieve reported 100MHz BCLK by disabling SB Spread Spectrum in a modded BIOS?
 
Now I set PBO in BIOS to max settings. RM reports up to 4.401GHz on one core during Passmark benchmark single core test while CB r20 still manages only 3.971GHz, According to RM, my best cores are 02, 03 and 08 while windows keeps on cycling highest core 02 and 03. I tried setting it's affinity to those cores one at a time and results are same.
BTW, it's first time I see in RM that cores go to "Sleep" at idle. Anyway, off to try some more BIOS settings, that non-100MHz BCLK really screws up calculations. I don't know why they don't fix that. That may account for less than 4.4GHz boosts.
 
Now when PBO seems to be fixed, got to do something about memory scores that are pretty well at bottom of what I had before. Latency seems to be worst problem, it's about 70 while it used to be in low 60s. Can't go under Cl 16 for lower or money so I tried overclocking it. Got to over 3800 but on 3733 Cl16, Fclk 1900 but that didn't heÄŤp much.
My "normal" settings are DOCP 3000 and frequency 3600(Fclk 1800.
 
781 - 800 of 1,784 Posts