cancel
Showing results for 
Search instead for 
Did you mean: 

X670 resource

Shamino
Moderator

ill use this thread to collect some new test bioses for the boards, maybe also to explain some less understood options

to disable cores ccd go here and choose ccd xx bit map down core.
each ones stand for an enabled core
best to disable from the back, ie:
110000
instead of 0011000
after selection press downcore apply changes or discard if made mistake

ocpak/octools

FAQ:
7950x not boosting pass 5.5G -> check that CStates is not disabled
Detailed Explanation on CState Boot Limiter


Test BIOSes:

new:
X3D OC Preset for those MB with asynch BCLK Support: (for simple slight perf boost for X3D)
97792

DOCP/EXPO Tweaked: (for simple timings tightening)
97793

strixe-e 1515 

strixe-f 1515 

strix e a 1515 

crosshair hero 1515 

crosshair gene 1515 

crosshair extreme 1515 

creator 670 1515

creator b650 1515

strix 650E I

strix 670 itx

 

 

for crosshair and strix e-e:

explanation of segment2 Loadline:

dualseg.jpg

customize a heterogenous loadline for a dual segment workload range.

example above shows loadline=L6 when current is in range of 0~40A, and Level4 when current is above 40A.

 

 

 


Adds for x3d

dynamic ccd priority switch with core flex, os / driver agnostic so win10 win11 ok

97403

97404

Algo as follows:
If condition reached and ccd0 specified, then check current mem/cache activity > threshold and hysteresis reached, if fulfilled then switch
If condition reached and ccd1 specified, then check current mem/cache activity <=threshold and hysteresis reached,, if fulfilled then switch
Default hysteresis =4

Can combine multiple algos for ccd priority so combinations are wide

works on non x3d too but of course senseless on it. detailed explanation here.

1,501 Views
2,457 REPLIES 2,457

Aeolux wrote:
CPU: Ryzen 9 7950x3D
RAM: Corsair 64GB (32GBx2) 6000Mhz CL36-38-38
GPU: Nvidia RTX 4090 Founders Edition
PSU: Corsair HX1200 Platinum
SSD: Raid0 2TBx2 WD Black SN850x
MB: Asus ProArt x670E Creator Wifi

Profile: EXPO2 (I tried EXPO1 as well)

BIOS 0922

I'm having 80 seconds BIOS time (from the Task manager), I've tried with the Memory context option, but suddenly BSOD.




I hope someone at Asus is sorting this out and reevaluate currently available EXPO Ram , really step backwards after such a successful AM4 eco system for so many years

It is very interesting. I have been monitoring competitor forums and I have not seen similar issues to this so it seems it is likely a BIOS related problem. I just hope they sort it out quickly. I have parts sitting on a bench. No point in assembling for them to not work.Â*

bladerunner6 wrote:
I hope someone at Asus is sorting this out and reevaluate currently available EXPO Ram , really step backwards after such a successful AM4 eco system for so many years


It's not the RAM, it's the BIOS - 0805 was stable at least but limited in terms of what could be enabled without stability problems. Even a simple AI overclocking was unstable. Everything since has been problematic... to say the least. I've had my Crosshair Hero since Oct 2022 and I've never once been able to get anything except a basic BIOS config working. Be pissed at Asus for releasing broken BIOS updates and using paying customers like you as a test bed.

I know X670-E is of course very new, but I can't even turn off memory training without a BSOD. WTF did we pay for here? Why isn't Asus increasing resources to get this sorted out rapid?

theteamcaptain wrote:
It's not the RAM, it's the BIOS - 0805 was stable at least but limited in terms of what could be enabled without stability problems. Even a simple AI overclocking was unstable. Everything since has been problematic... to say the least. I've had my Crosshair Hero since Oct 2022 and I've never once been able to get anything except a basic BIOS config working. Be pissed at Asus for releasing broken BIOS updates and using paying customers like you as a test bed.


The 0805 BIOS can be made to be stable, and it actually performs very well.
The core issue with the AI Overclocking is that it enables Medium Load Boostit, which is the real issue.
Medium Load Boostit can be made stable though. Its core issue is that it will boost each core up to some internally calculated clock limit that results in Vdroop becoming excessive under light workloads, and bam, instant reboot or BSOD.

The fix is actually fairly easy. Set PBO to enable. Set Medium Load Boostit to Enable. Disable boost clock over-ride.
Now go into the per-core boost clock limit menu, and for everything that's on CCD0, set that to 5750MHz, and for everything on CCD1, set that to 5550MHz. If you have a particularly bad overclocking CPU, keep dropping those values by 25MHz at a time until stability is found, but 5750/5550 should work for most CPUs
Go into the core optimizer, and set that to all core, and don't drop it below -4. I mean, you can, but your mileage may vary. Setting per-core PBO values that vary widely seems to exacerbate the Vdroop issue, so it's much better to stick to all core CO, and a value of -4 is generally safe for most as a starting point, but probably best to start at 0, and drop by 2 at a time and test for stability each time to see how low you can go with it.

...and that's basically it. Save and reboot, and you should have 0805 stable now.

MLB is awesome on 0805, it's just that it's made unstable by the overly high boost limit. Fix that, and you're golden on 0805.

Castillan wrote:
The 0805 BIOS can be made to be stable, and it actually performs very well.
The core issue with the AI Overclocking is that it enables Medium Load Boostit, which is the real issue.
Medium Load Boostit can be made stable though. Its core issue is that it will boost each core up to some internally calculated clock limit that results in Vdroop becoming excessive under light workloads, and bam, instant reboot or BSOD.

The fix is actually fairly easy. Set PBO to enable. Set Medium Load Boostit to Enable. Disable boost clock over-ride.
Now go into the per-core boost clock limit menu, and for everything that's on CCD0, set that to 5750MHz, and for everything on CCD1, set that to 5550MHz. If you have a particularly bad overclocking CPU, keep dropping those values by 25MHz at a time until stability is found, but 5750/5550 should work for most CPUs
Go into the core optimizer, and set that to all core, and don't drop it below -4. I mean, you can, but your mileage may vary. Setting per-core PBO values that vary widely seems to exacerbate the Vdroop issue, so it's much better to stick to all core CO, and a value of -4 is generally safe for most as a starting point, but probably best to start at 0, and drop by 2 at a time and test for stability each time to see how low you can go with it.

...and that's basically it. Save and reboot, and you should have 0805 stable now.

MLB is awesome on 0805, it's just that it's made unstable by the overly high boost limit. Fix that, and you're golden on 0805.


This is really interesting, thank you! I think I will take some time this weekend to better understand your setup, flash back to 0805 and see if I can make it work for me. It would be great to get some more performance.

Hi,

old topic but I want to share experience with MLB.

MLB is nice, but trash.

If you are obliged to set -5 all cores to run it stable in all situations, you destroy Multithread performances.

For example, I don't know if you tried Aida64 FPU Julia test ? This is a MLB killer test.

I am fully stable in all games and tests as y-cruncher (1-7-0), corecycler, Realbench, MW2, Overwatch with follwing CO on my 7950X on StrixE-E with following CO and MLB enabled: -13, -8, -8, -13, -13, -13, -10, -13, -18, -18, -18, -18, -18, -18, -18, -18.

But only in this Aida64 FPU Julia test, this is instant reboot or PC freeze with this CO curve and MLB enabled.

I am obliged to retweak CO curve and even -8 all cores create these issues in this test with MLB enabled.

So globally, I prefer -13, -8, -8, -13, -13, -13, -10, -13, -18, -18, -18, -18, -18, -18, -18, -18 with MLB disabled fully stable everywhere (Even Aida64 FPU Julia) with nice multithread performances, than -7 all cores with MLB enabled, on which at the end I will have same boost in games because -7 all cores is much more voltages than -13, -8, -8, -13, -13, -13, -10, -13, -18, -18, -18, -18, -18, -18, -18, -18.

gcas23 wrote:
It looks likes this BIOS (and other ones which sets 1.35 on SOC) happens only to poeple using RAM that requests 1.35v VDD, could it be that theese BIOSes actually applies DRAM VDD to SOC? Is this expected behavior?


Looks like you have a point because my ram work with 1.4v

https://imgur.com/QfjibXE

Shamino wrote:
https://www.dropbox.com/s/056ot5553uot814/ROG-STRIX-X670E-E-GAMING-WIFI-ASUS-0088.zip?dl=0

https://www.dropbox.com/s/3uo7rj3c4grso6x/ROG-STRIX-X670E-E-GAMING-WIFI-ASUS-0087.zip?dl=0

there needs to be some positive feedback before i know which is the right direction,.


Hey @shamino,

Maybe embarrassingly, have just about every Asus x670E board out (that I'm aware of), I have the extreme, hero, a (few, don't ask) gene's, X670E-I., obviously I have no need for the strix or proart but still covers most boards. Alll proc's expect the 7600x but will probably get that shortly. As i mentioned earlier, I am competitively overclocking (ln2, I'm common for ya safedfisk 😉 , hwbot:lntwool ), I also have one of those board in my daily, on water. A few are running chilled on bench the rest are ln2. I can test across all boards and use cases(normal user ambient, normal user water, and us addicted ln2ers so ln2 mode, etc) and provide rapid feedback across all procs and mb's. The only thing I'll ask in return is 10 minutes of your time to tell me what all the damn aberveations or undesript fields in the OC Tool. But send me anything you got i will test any or all use cases

Shamino wrote:
this version is almostly exactly like 0805 , hopefully we dont need to go back there so much:
https://www.dropbox.com/s/lkxkguwc1s59xc6/ProArt-X670E-CREATOR-WIFI-ASUS-0088.zip?dl=0



7950x, ProArt X670E, Samsung DDR5 4800 32GBx2

1) first i tested 0087. couldn't POST after bios flashing.
and i retried with one memory module(32GB) installed, the BIOS installation continued and I was able to complete and boot it.
but 0087 was unstable. only works with default setting. cannot POST with AEMP (4800->5200) or PBO (temp limit 90') setup.
(on AEMP or PBO setting. POST failed with one memory module(32GB) and that reboot safe mode. POST impossible with two memory module(32GBx2))

2) next i tested 0088. memory training had some problem with two module installed, but it works POST possible.
i was able to set up AEMP(4800->5200) and PBO(temp limit 90') without any error. and windows boot was also successful.
I think this bios version is very close to 0805. i will look into other stability issues as i use game and app.

GyuhoLee wrote:
7950x, ProArt X670E, Samsung DDR5 4800 32GBx2

1) first i tested 0087. couldn't POST after bios flashing.
and i retried with one memory module(32GB) installed, the BIOS installation continued and I was able to complete and boot it.
but 0087 was unstable. only works with default setting. cannot POST with AEMP (4800->5200) or PBO (temp limit 90') setup.
(on AEMP or PBO setting. POST failed with one memory module(32GB) and that reboot safe mode. POST impossible with two memory module(32GBx2))

2) next i tested 0088. memory training had some problem with two module installed, but it works POST possible.
i was able to set up AEMP(4800->5200) and PBO(temp limit 90') without any error. and windows boot was also successful.
I think this bios version is very close to 0805. i will look into other stability issues as i use game and app.


i tested 0088 with that setup for hours. mostly stable on gaming and app. but i found some bug of PBO.
i may have misconfigured but, it does not work AMD Eco Mode.
to be more specific, AI Tweaker / Precision Boost Overdrive / AMD Eco Mode and cTDP Limit option cannot works.
but PBO thermal limit, and curve opimizer options works without any issue.