cancel
Showing results for 
Search instead for 
Did you mean: 

X670 / X870 resource thread

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.

896 Views
4,012 REPLIES 4,012

QuestUA
Level 8

Hey abit confused at this week main site will be updated by beta or final version? "b650e-f"

Maddeniels
Level 9

so, since the BIOS works a bit better, i for the first time since i have this PC (2 years minimum) was in an experimental mood, and i put my best thanos face on and thought to myself "fine, i'll do it myself!"
i played around with destiny 2, and the options of core parking.

i run a dual monitor set up, cause i'm ADHD like that. i had firefox running with twitch, and a 4K video next to it in VLC on one screen, and destiny 2 on the main screen. i played all those strikes so many times, that yesterday, i dozed off whilst playing, and watching a video, and i woke up at the end of the strike, and it took us maybe a minute more than when i'm fully awake. i'm a multitask master.

so for starters windows with no aid besides gamebar and the new driver from AMD. reliable in changing the CCD. but doesn't allow for both to really be used. it throws everything on the gaming CCD once a game is on, and if it peaks, it throws things on the parked CCD, which caused stutters, mostly just one noticable one every minute give or take, depends on the amount of action on screen, annoying. pretty smooth tho, when it works.

now the new mainboard solution for CCD parking. when it works, it definitly feels the smoothest, but it sometimes forgets that there's also a game running, and without restarting the game, it stays on the none X3D CCD, dunno what causes it, sometimes in orbit (which is basiclly a hub area in some sense) which is sad, cause it somehow even managed to throw some of the workload of the none gaming tasks, on the CCD which is literally just idling in the system, which is what you'd expect the system to do. why else would you have a second CCD in the system, when one could be focused on only gaming, and the rest is on the CCD which is far better for tasks. that lax workload should be no issue at all, as destiny is not really that CPU bound, and i have a 4090, so it's moreso GPU bound, at 3440x1440p.

process lasso. feels the worst, after having tried the other solutions, but it can keep everything on the right CCD cause you can literally chose where what runs. it's not unplayable mind you, it's just, that the new mainboard CCD solution when it works, feels buttery smooth, with barely any frametime spikes, when it doesn't forget that you're playing a game. lasso even in heavy action scenes, stays the exact same with minimal frametime wobbles, whilst both other solutions, are making the gaming CCD spike too high with too many tasks which are not games, and throw parts of the game on the non gaming CCD, when you'd expect it to throw tasks on there.
i currently have an ingame build perfect for stressing thee system, which is triggering so many effects, that the CPU is starting to sweat with process lasso, but it continiously stays below 90 in usage, and uses all lanes. both other solutions sadly, often even have 1 or 2 gaming lanes asleep on the gaming CCD, and instead of utilizing those, suddenly throw things on the non gaming CCD. in ryzen master, there's also on both CCD a best and a second best lane, and those were sometimes the ones sleeping on the X3D CCD XD

conclusion. this double chip design is not it for AMD. they either need to put a scheduling chip which can handle two lanes simultaniously on this chip, with an AI solution or something to keep the schedules in lane (cause when it is scheduled correctly, not even pure graphics insanity can overwork it), or stop the cap, and release a real 9950X3D, where both CCDs are X3D.

as it is, it's a bit of a false advertisment, cause the whole chip is more of a work chip, cause that works well on it, with a second mode for gaming, but highly restricted.

why am i being so picky? cause my intel 8086k didn't have any of those issues. it could easily carry all that stuff, with destiny 2, without having even a noticable impact, besides of course, being far weaker and having generally lower overall FPS.
as for the wobble, i'd chalk that up to the lack of a real scheduling chip which can feed both CCDs correctly, and one CCD being bothered by the other. i never had a none X3D chip, so i can't really say if those issues exist on those, but problem here is, a reviewer is of course always gonna try to create the most favorable testing enviroment, and every bench you see in videos, is after a fresh boot, to see the max power of the system.

i didn't try turning off the CCD yet, but there's like 500 videos about it online, just not in my usecase, but i doubt it would be much diffrent if the tasks make it peak with scheduling multiple tasks at once. the main problem here is how it is scheduled, and probably how many lanes there are for simoultanious tasks on both CCDs and/or how they interact.

which solution will i go for? tough choice to make. if i could turn of the CCD without having to do it in BIOS, i'd go for that or the onboard mainboard scheduler. but, for now it's probably gonna stay lasso, until the next update, cause like with all things gaming, consitency is key, and the frametime wobble in lasso, is so consitent, that you grow accustome to it. but switching back to buttery smooth is hella jaring.

i know, it's a bit off topic, but it also is on topic as most ppl have been vocal about wanting changelogs, so i guess since ppl throw their overclock metrics on here, a bit of a dirty quick review, also has some validity to be here, when it comes to testing a new feature.
i don't know how deep the BIOS can even aid a usecase like mine, without a dedicated solution on the chip itself. all i know is, that intel has their chips and chiplets under better control in multitasking. AMD definitly has to put in some over time here. i was actually hoping that the new 9950X3D is gonna have a solution like mentioned on it, and not just a "we put the X3D under the chip for better termals" solution in it. they said they have exciting news, needing a bit less juice, and a bit better termals, is not even remotly exciting, cause games don't stress the CPU into peak temperatures anyway. for me, the termals with a decent 360 AIO from ASUS ryujin 2, never even go over 75C, and that is only when the CPU is like under real pressure, with a fanprofile on normal.

yadda yadda strix x670e-f, Aorus waterforce 4090, some g.skill ram 64GB, obviously a 7950X3D.

also wanted to say i'm glad that the BIOS is becoming fun now. my best friend is gonna be hella excited once i hand him the mainboard, as soon as i get my new one. he's also not even remotly as picky as i am, after 13 years now in the pro PC space. guess his birthday is gonna come a bit early this year.

AlphaXray
Level 8

Would love to hear more stories of people with 7950X3D's and the new setting.  I'm waiting for the bios to come out of beta then I'll install it.

GabcamMT
Level 9

1001 going rather well here ....   is my performance on X870e Hero + 9800X3D similar  ?

GabcamMT_0-1736981047829.png

below slight improv .. scalar 5 instead of 3 .... and negative all core curve optimizer 15

GabcamMT_0-1736984778207.png

 

 

How is it possible to get such high results in cpu-z? My 9800X3D gets about 850 single core and that is using curve optimizer etc etc.

it's the avx benchmark ... look at the drop down next to benchmark ...

kx11
Level 10

If i connect my gpu to the 2nd PCI slot am i losing bandwidth since i got m.2's in all nvme connectors, the manual doesn't answer that

 

 

My board is x870e HERO

Its actually right there in manual, bandwidth doesnt matter as slot will be disabled if you populate both of this m.2 slots.Capture.PNG 

Well i removed M.2_3 for the sake of saving bandwidth but turns out if M.2_2 is populated PCIE16_1 will be X8 so i guess i can now use PCIE16_2 ??

74lobster
Level 10

Hi, I have this USB port error:

unknown usb device (device descriptor request failed) windows 11

Windows has stopped this device because it has reported problems. (Code 43)

A request for the USB device descriptor failed.

Someone else have tyhe same problem ?

How I can fix it ?

Thanks in advance.


usb.jpg

 

 

74lobster