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.

1,027 Views
2,380 REPLIES 2,380

lordofwater
Level 9

i hope in one x670e Tuf hero in the future!!!!

AegisFLCL
Level 8

@Shamino @Silent_Scone 

The x670/b650 strix e & f boards seem to be having an issue with training in 1:2 mode at speeds above 7800. Lots of users over at oc.net, including myself, are reporting this.

 

Ive been able to run 7800 after a lot of timing/voltage tweaking up to 30k coverage ~12hr in kahru on multiple posts at the same settings, but next boot or any subsequent boot all of the sudden youll just fail at 1k coverage, 5k coverage etc. Its nearly impossible to tune with these boards. Are there any improvements in the pipeline or insight as to why the same exact settings are providing different stability results boot to boot?

*dimm temps max around 48-49c after 12 hours with 120mm fan blowing directly on them.

Need more info. Memory kit? What speed are you actually trying to stabilise, 8000MT? Hard to comment given the landscape without trying to replicate it with that board. They were never validated to run those speeds. 7800 is no joke.

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

F5-7200J3646F24GX2-TZ5RK

Working on 7800, 8000 is impossible as same settings can run for 8 seconds or 50 minutes so theres no telling what to adjust.

IMG_5096.png

63c on first dimm seems to be a readout error, dimms sat at 48-49c most of the test.

*edit - Theres also some weird behavior where the board wont post with vddp set to auto, or anything higher than 1v @ 7800/8000 (auto defaults to 1.15v when in 1:1 or speeds 6000-6400).

I can include a txt export of bios settings if that would be helpful.

I don't have that board so wouldn't be much point right now. You're pushing that kit quite far. Does auto vddp post @ any frequency below 7800MT, then? I'm unsure where the limits are on the lower-end chipset boards. If there's enough noise from the community for 7800-8000MT kit validation on AM5, will have more of an idea.

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

With VDDP @ Auto

6000-6600 - post into windows vddp @1.15

6800-7000 - post into windows vddp @ 1.05

7200 - no post hangs on 17, requires clear cmos

X670 Strix E and F exhibit very similar behavior. 7000+ requires VDDP to be manually set to 1v in the AI Tweaker menu; ive gotten it to post at 7800 by adjusting vddp values in AMD oc menu but its very unreliable and can fail to post on subsequent boots (not that this guarantees stability anyways).

A handful of other higher end b650 boards from other vendors seem to be having less issues with post to post stability at the same settings. My gut feeling is during the training process timings we dont have access to are being set differently leading to potential instability; but thats just a guess and im no expert.

There are several active users over at OCN in the DDR5 overclocking thread reporting the same experience and varying stability boot to boot; including a prominent overclocker that is using the same kits across several boards that is able to get 8000 24hr+ stable (but not on the X670 Strix E he has).

Shenny
Level 11

As a side comment, I am wondering why so many people insist on running their RAM on 8000+. Correct me if I am wrong but this could worsen the performance in many scenarios because the CPU memory controller is no longer in 1:1 ratio with the memory frequency. As AMD have stated, 6000 is the sweet spot for Zen 4 chips, maybe up to 6400 if you are lucky with the silicon lottery. Of course, there are memory intensive scenarios that will benefit from the higher memory speed but if I am buying a new AM5 platform now, I wouldn't aim at these memory kits, which are also more expensive on top of that.

Because people like to brag about useless numbers. On X3D at least, there's no benefit in running above 6000. There's almost no noticeable improvement between stock/6000 from what I tested on my 7900X3D.

Then they come here complaining about 8000 not working just because Intel can do it :))

___________________________________________________________________________________________________________________
AMD Ryzen 9 7950X3D | ROG Crosshair X670E Extreme [1905] | Sapphire Nitro+ Radeon RX 7900 XTX Vapor-X | G.Skill F5-6000J3040G32GX2-TZ5NR | 2TB Samsung 990 PRO | 2TB Samsung 980 PRO | 2TB Samsung 970 EVO Plus | Samsung Odyssey Neo G9 | ROG RYUJIN III 360 ARGB | ROG Thor 1200P | ROG Chakram X | ROG Strix Flare II | ROG Theta 7.1 | ROG Cetra True Wireless | ROG Balteus Qi | ROG Eye S | ROG Aura Terminal | ROG Throne | ROG Spotlight | ROG Strix Arion | Windows 11 Pro

That’s not really true, nor is it why I brought it up in the first place. My concern, and several other people on OCN seem to be running into issues where one post you can run 30-40k coverage in kahru and other stability tests, but subsequent posts you can fail at varying coverage rates; its an issue of consistency and as far as we can tell it could be the motherboard hardware/bios.

7800 c36 is roughly the same latency as 6400 c30 while providing 8-10gb/s more throughput on read/write with much higher copies. While that doesn’t universally translate to better performance everywhere there are plenty of applications in which you will see a benefit. 7800+, and to some degree 7600, can overcome the latency hit of 1:2 and desync’d fclk.


@AegisFLCL wrote:

several other people on OCN seem to be running into issues where one post you can run 30-40k coverage in kahru and other stability tests, but subsequent posts you can fail at varying coverage rates; its an issue of consistency and as far as we can tell it could be the motherboard hardware/bios.

This can be par for the course when pushing platform limits, though. There is quite a bit of disparity between what is electrically valid at [POST] and passing memory stress tests from within the operating system. If conducting a warm reset or powering the AC on and off is enough to create instability, the signal margins aren't sufficient enough at the applied overclock.

I'll see if we can get some clarity on VDDP behaviour.

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090