[zbug]

Hi Folks,
About to build my AM5 machine (upgrading from AM4 5800x3d which i'm passing on) with a 7800x3d and wondering if there are any recommended settings (similar to cppc etc for the AM4 variant that are detailed in the 5800x3d thread) to turn on/off in the bios? I'll be running a rog trix x670e-e

Can they all take -30 all cores easily like the 5800x3d?

Thanks!
Cheers

 

[rv8000]

Hi Folks,
About to build my AM5 machine (upgrading from AM4 5800x3d which i'm passing on) with a 7800x3d and wondering if there are any recommended settings (similar to cppc etc for the AM4 variant that are detailed in the 5800x3d thread) to turn on/off in the bios? I'll be running a rog trix x670e-e

Can they all take -30 all cores easily like the 5800x3d?

Thanks!
Cheers

I don’t know about the 7800 specifically, but my 7900x3d was prime stable at -20 CO across all cores, games were not happy though so I had to drop it to -15. The real stress tests seem to be load changes or anything that involves quick changes in power states/clocks.

 

[chewonthis]

Hi Folks,

Can they all take -30 all cores easily like the 5800x3d?

Thanks!
Cheers

No

 

[Nizzen]

cyberpunk 1080P Low, Red dead redemption 1080P lowest favor performance ticked. Unfortunately i'm going to capture a plague tale as no built in benchmark. Looking into atomic heart now. View attachment 2609324

For comparing:
Too bad you have 4080, and not 4090
EDIT: Looks like Framegeneration is ON on default preset here... Need rerun

1080p low preset default

Edit: Low preset 1080p with framegeneration off:

 

[harderthanfire]

Hi Guys!
Bad: I have a problem with 7950x3D on x670E EXtreme. For almost a week now I'm trying to make CCDs go with different voltages. I've regulated curve by hand and by HYDRA,My RAM is 5600 CL28-34-34-28. I'm able to score as good as SkaterBencher, but my CCDs are always playing the same DLDO_VID melody....
Worse: My CCD1 is running lower than CCD0
View attachment 2609381

And CCDs are always running almost identical VID
View attachment 2609382

I see domdtxdissar is able to run 7950X3D at CPU Package 159W. Mine cant go higher than 142W despite CCD1 cores just having free time.
I doesn't matter if I go by PBO, by regulated curve or by HYDRA, Everytime I have 142Wlimit and my CCD1 cores are just doing things at 60C

Any ideas?

Tried a more aggressive LLC setting to maybe reign in the vdroop?

 

[chewonthis]

For comparing:
Too bad you have 4080, and not 4090
EDIT: Looks like Framegeneration is ON on default preset here... Need rerun

1080p low preset default
View attachment 2609402

Edit: Low preset 1080p with framegeneration off: View attachment 2609402

I thought the settings were odd for defaults but then it says custom. I ll re run tonight. @ work.

 

[Wreckedge]

Tried a more aggressive LLC setting to maybe reign in the vdroop?

Up to LLC6 but of course I can try higher. Didn't thought it may take any difference with differentiation of CCDs voltages...
I'll be back.

 

[domdtxdissar]

Did any reviewers show the SP rating to prove that? Otherwise I doubt that is true. However, are 7950 X3D’s using binned chips that will be better quality than 7800X3D? 100%, that’s what you pay the extra money for. 5950X was the same too.

This is a invalid comparison
7800x3d have a other SP value than CCD0 on a 7950x3d because fmax is limited 5050mhz. (vs 5250mhz on the 7950x3d CCD0)
SP values between these two SKU's cant be compared in this regard.

And as my earlier data clearly have shown, SP values in a vacuum means very little, better to look at real voltage/frequency curve.

 

[Wreckedge]

@domdtxdissar @AvengedRobix @CiiD @shamino1978

Guys help me please with BIOS.....it's insane.


These are domdtxdissar results (yours ) :

Dont think this will help you much, but i have been optimizing my curve for AVX2 workloads like 3dmark profiler (AVX2 is harder than AVX512 for Zen4)
Currently i have it like this for my daily 24/7 settings
View attachment 2607020

The CCD1 on the 7950X3D is pretty hard to work with since you almost have to balance out two different voltage curves at the same time..
One for sub ~14 threads workloads where (hopefully) only CCD1 is active and you can get hit all the way upto 1.400 vcore in light workloads
And other one for allcore workds where the CCD0 is limiting CCD1 vcore from ~0.9 upto ~1.1 vcore

All that junk disabled

These are mine:

Similar a bit, but my 7950x3D is never going over 142W.
CCD1 stays at the same level as CCD0 with DLDO_VID

Please look at my settings maybe You can find something that I'm doing terribly wrong... same on 922 and 1004 BIOS

I'm able to reproduce same results with precise regulation of curve. Almost the same with PBO enabled and Curve for All Cores -15.

Any insight what maybe holding my CCD1 back from boosting to higher DLDO-VID and higher clocks?


Thanks

 

[lopri]

Hey friends,

I found a Z790 apex but I'm having regrets I think, I have no issues having to disable e-cores, OC to 5.8GHz and 7600mhz but does it even match a 7800x3d at that point?

I don't mind paying for the best gaming fps, I know 1% and 0.1% lows are slightly better on Intel but that does even matter if you're not a hyper competitive gamer.

I went all out because I can for the first time in years and it's fun to OC. But ehhh

It's totally fine. Only thing that you are missing out is lower power consumption. Nothing else.

 

[chewonthis]

For comparing:
Edit: Low preset 1080p with framegeneration off:
View attachment 2609405

Looks close. I had a bunch of crap open and it's a full bloat test OS aka worst case scenario.

I will say the loose timings... don't be fooled they are actually fast for 2x32 on amd but single ccd only. Comparable to my 28-37-37 2x16 same speeds/fclk.

 

[domdtxdissar]

@domdtxdissar @AvengedRobix @CiiD @shamino1978

Similar a bit, but my 7950x3D is never going over 142W.
CCD1 stays at the same level as CCD0 with DLDO_VID

Please look at my settings maybe You can find something that I'm doing terribly wrong... same on 922 and 1004 BIOS

I already told you the reason in private chat @ hwbot forum but seemingly you didn't like the conclusion.. (?)
(we were comparing powerdraw in cinebench, my 160w vs his 140w)

Don't really have anything more to add to this

 

[Wreckedge]

I already told you the reason in private chat @ hwbot forum but seemingly you didn't like the conclusion.. (?)
(we were comparing powerdraw in cinebench, my 160w vs his 140w)
View attachment 2609448
Don't really have anything more to add to this

Yup. you got me there: I do not like that conclusion. Something is holding my CPU PPT : mine 137.259 vs yours 155.466.
And I'm totally not on stock memory values - as you can see. 5600 Cl28-34-34-28 1T 1.4V

My DLDO_VID for CCD1 is always the same as for CCD0 - some about 0.93V while yours can go up to 1.1 and your CCD1 is holding higher clocks.
I did regulate CCD1 cores with much lower Curve voltage drop and the only thing that changed were temperatures that got up.

That is why I asked more people. After a week of all time changing presets I have done over 500 print screens, run much more tests and I cannot force CCD1 to go higher than CCD0.
In fact its clocking lower at All core tests than CCD0, despite ramping up in SC scenarios pretty well.

Should I understand that all my settings in your opinion are strictly correct and the only thing that is holding my CCD1 efficiency is CPU being not delidded? (while having 56C on CCD1 cores?)

 

[PensionMan]

Ok so....HZD is a little odd. I had the 7700x tuned really well for the best run with rebar off. This game may benefit from a better core clock chip. I will have to investigate more in this game....I really did not pay attention to reviews so no clue if this is inline with normal results. View attachment 2609312 View attachment 2609313

Fc5 unfortunately I don't have a balls out run saved with 7700x but it was a decent tune.......But holy hell on the x3d run..

View attachment 2609314

Just a sample run to compare above since i had 1.5 resolution scale on for some reason.

View attachment 2609331

View attachment 2609315

Impressive result and you use 64Gb ram at 6400, so ryzen have not stability issues with DR ?

 

[LtMatt]

cyberpunk 1080P Low, Red dead redemption 1080P lowest favor performance ticked. Unfortunately i'm going to capture a plague tale as no built in benchmark. Looking into atomic heart now. View attachment 2609324
View attachment 2609325

For comparing:
Too bad you have 4080, and not 4090
EDIT: Looks like Framegeneration is ON on default preset here... Need rerun

1080p low preset default
View attachment 2609402

Edit: Low preset 1080p with framegeneration off:
View attachment 2609405

Stock 7950X3D for comparison gentlemen, 1080P Low preset (no frame gen).

 

[Blameless]

Just got a 7800X3D 2311PGY. Haven't had any opportunity to test it yet, but I can confirm it posts fine with the F1 firmware on this new B650M Aorus Elite AX, which predates support for the CPU.

 

[chewonthis]

Impressive result and you use 64Gb ram at 6400, so ryzen have not stability issues with DR ?

Not until you stack 4x32 in

 

[Nizzen]

Stock 7950X3D for comparison gentlemen, 1080P Low preset (no frame gen).

View attachment 2609457

Looks like Cyberpunk liked HT=on, because I got higher min fps

 

[chewonthis]

Looks like Cyberpunk liked HT=on, because I got higher min fps
View attachment 2609472

How's red dead 2 mizzen? Isn't that an Intel benchmark currently? I threw a few games in on purpose that are not 7800x3d wins.....just for an open minded comparison.

Also plan to run 2 games that 7900x was trash at.

 

[Nizzen]

How's red dead 2 mizzen? Isn't that an Intel benchmark currently? I threw a few games in on purpose that are not 7800x3d wins.....just for an open minded comparison.

Also plan to run 2 games that 7900x was trash at.

Need to buy red dead 2 first. haven't tried yet. Only play BF games LOL. Rest of the games is just for benchmarking