Overclock.net banner
5,861 - 5,880 of 9,494 Posts
@domdtxdissar
Does the 9950x3d have a latency penalty in pyprime?
9800x3d 6000cl26 got ~6.6s
9950x3d 6400cl26 got ~8.2s
Not exactly the same tuning but ca 59 Aida latency…

Edit:
Set affinity to best cache core and got 6.555.

so frequency ccd is worse than cache ccd…
You are prob running PYprime on CCD1, force it over to CCD0 with affix or CPPC = cache
edit
See you have already figured it out (y)
 
i would first check the maximum stable FCLK now, try 2200 and 2233 if those dont work settle for 2167.
then i would check the maximum trfc. its either 170ns or 160ns on that kit (i have the same kit it does 160ns trfc)
then set VDD and VDDQ to 1,45v - 1,55v , raise timings and check if you can do DDR5-8400 or better
this will probable take some time a couple hours :) good luck
Thanks Kaliz, I'll try all of this tomorrow after work.
"This will probably take some time, a couple of hours," or many days :)
 
Is there any reason I am seeing lower GPU usage in Time Spy and Firestrike with both CCDs enabled (60-90% vs 100% with 1 CCD)?
 
I posted this in the Ryzen 7000 X3D thread as well but it's also relevant here and there are people actually posting in here... I've got a couple questions:

1. If I use Process Lasso to confine background tasks to the non-3D cache CCD and therefore the Game Bar is unable to park those cores, would that negatively impact gaming performance? Does the X3D CCD's superior gaming performance rely on none of the non-X3D CCD's cores doing anything with any other processes at the same time?

2. Many 9950X3D reviews seem to mention that the 7950X had a lot of challenges with its core parking that are apparently fixed with the 9950X3D. Are they just talking about the condition of the 7950X3D at the time of release? Because it seems to me that the AGESA and chipset driver updates have made the 7950X3D experience with core parking just as smooth when using it today.
 
I posted this in the Ryzen 7000 X3D thread as well but it's also relevant here and there are people actually posting in here... I've got a couple questions:

1. If I use Process Lasso to confine background tasks to the non-3D cache CCD and therefore the Game Bar is unable to park those cores, would that negatively impact gaming performance? Does the X3D CCD's superior gaming performance rely on none of the non-X3D CCD's cores doing anything with any other processes at the same time?

2. Many 9950X3D reviews seem to mention that the 7950X had a lot of challenges with its core parking that are apparently fixed with the 9950X3D. Are they just talking about the condition of the 7950X3D at the time of release? Because it seems to me that the AGESA and chipset driver updates have made the 7950X3D experience with core parking just as smooth when using it today.
They are talking about release. This CPU parks the same exact way as the other CPU did.
 
So the new core parking actually works great. They must of changed more than what meets the eye I guess. The CPU does not like CCD 1 being disabled. First picture is 1440P DLSS quality extreme preset with CCD 1 disabled. Second is with CCD 0 and 1 enabled. Both are PBO +200 with CO offset -25 on all cores. The game is CoD MW3.
 

Attachments

So the new core parking actually works great. They must of changed more than what meets the eye I guess. The CPU does not like CCD 1 being disabled. First picture is 1440P DLSS quality extreme preset with CCD 1 disabled. Second is with CCD 0 and 1 enabled. Both are PBO +200 with CO offset -25 on all cores. The game is CoD MW3.
Nice. Do you have any other settings changed in the bios related to the CCDs?
 
Hey all, got a quick question. Got my 9900x3d in today. Swapped it out with my 7950x3d.
Reset bios, did the Chipset driver cleanup and reinstall. Cores are parking and all that good stuff. Reset bios and all that.
As with my 7950x3d, I went back into the bios and set PBO to 85c limit. Set my EXPO on my memory.
My question is,
Is 80c ok on this CPU at a peak with 4 hours of Warzone Resurgence ? I switched AIO's today as well. Went from a Coolermaster Atmos that was cooling my 7950x3d with a peak of 69c with same simple bios settings during my warzone shenanigans- To a BeQuiet Silent Loop 3 420mm. When I would check Afterburner , my temps would be between 58c-66c ish while actually playing. But that peak of 80c has me wondering if I need to go back to my Atmos , or if I am just worried for nothing. Did mess around with the case fan speeds, and AIO fan speeds, and it didn't really make a difference besides added noise. Case Is a Flux Pro (yes I need to update my sig 😂)
tldr - should the 9900x3d be running warmer that my 7950x3d ? Thanks for any help or peace of mind on this
 
The Zen 5 X3D CPUs have a higher max temperature because the V-Cache CCDs have the cache on the bottom.
 
  • Rep+
Reactions: kfxsti
I posted this in the Ryzen 7000 X3D thread as well but it's also relevant here and there are people actually posting in here... I've got a couple questions:

1. If I use Process Lasso to confine background tasks to the non-3D cache CCD and therefore the Game Bar is unable to park those cores, would that negatively impact gaming performance? Does the X3D CCD's superior gaming performance rely on none of the non-X3D CCD's cores doing anything with any other processes at the same time?

Using process Lasso to manage core affinity could impact negatively the performance.
I mean a horde of game will check number of CPU threads then attribute in the game engine this number of threads at game start. When you manage a game to run affinity only CCD0, it try to run 16 threads into 8 threads (smt off example). This could impact negatively as i saw too much. That's why i stopped this method and was disabling CCD1.
I think the only proper way is to disable CCD1 when you play, even if you share application + game in same CCD, it will be better than this workaround...
 
Using process Lasso to manage core affinity could impact negatively the performance.
I mean a horde of game will check number of CPU threads then attribute in the game engine this number of threads at game start. When you manage a game to run affinity only CCD0, it try to run 16 threads into 8 threads (smt off example). This could impact negatively as i saw too much. That's why i stopped this method and was disabling CCD1.
I think the only proper way is to disable CCD1 when you play, even if you share application + game in same CCD, it will be better than this workaround...
I'm not talking about managing core affinity for games. I'm talking about managing core affinity for everything that isn't a game to run on CCD1 while the game runs on CCD0 which is managed by the Game Bar and the chipset drivers.
 
I posted this in the Ryzen 7000 X3D thread as well but it's also relevant here and there are people actually posting in here... I've got a couple questions:

1. If I use Process Lasso to confine background tasks to the non-3D cache CCD and therefore the Game Bar is unable to park those cores, would that negatively impact gaming performance? Does the X3D CCD's superior gaming performance rely on none of the non-X3D CCD's cores doing anything with any other processes at the same time?

2. Many 9950X3D reviews seem to mention that the 7950X had a lot of challenges with its core parking that are apparently fixed with the 9950X3D. Are they just talking about the condition of the 7950X3D at the time of release? Because it seems to me that the AGESA and chipset driver updates have made the 7950X3D experience with core parking just as smooth when using it today.
I use ProcessLasso on my 5900x. Instead of disabling a CCD, you should use ProcessLasso to direct your game to the 3D Cache CCD, and then direct other processes to the non-3D CCD. That way your game has one whole CCD almost completely to itself, and all of your other apps/processes have their own CCD.

So to answer: no it won't affect gaming other than if you disable a CCD, then your apps like Discord, etc., will run on the same CCD as your game. Better to use one CCD for games and the other for processes.
 
I use ProcessLasso on my 5900x. Instead of disabling a CCD, you should use ProcessLasso to direct your game to the 3D Cache CCD, and then direct other processes to the non-3D CCD. That way your game has one whole CCD almost completely to itself, and all of your other apps/processes have their own CCD.

So to answer: no it won't affect gaming other than if you disable a CCD, then your apps like Discord, etc., will run on the same CCD as your game. Better to use one CCD for games and the other for processes.
Right, but this is on a non-X3D CPU. X3D CPUs specifically park all of the cores on the plain CCD so that all processes run on the V-Cache CCD. My question is: If I manually assign my background processes like Discord, Voicemeeter, etc to my plain CCD while a game is running, does that negatively impact the V-Cache CCD's performance, since it is different from the normal dual CCD Ryzen CPUs.
 
I would like some R23 benchmark with only 3D CCD enabled from anyone, please!
Second this.
Same for CPU-Z, 3D mark timespy and similar benchmarks.

I have been doing this for years, since AM4 ;) , first CPU I did was a 5900X, where I matched voltage across all cores and kept rolling down and matching up voltages.

Check out OP of my curve optimizer thread. Don't use VID, info in OP of my thread. You want to use CPU VDDCR_VDD Voltage (SVI3 TFN). I would just keeping rolling down CO, you won't need Curve Shaper.
Do you wanna use CPU VDDCR_VDD Voltage (SVI3 TFN) or Vcore? Latter one is what's measured on cores, if I understand it correctlty
 
5,861 - 5,880 of 9,494 Posts