That's a HUGE negative impact when using a two ccd chip on newest DX11 optimized drivers. I think this should be pretty easy to fix, hope some AMD responsible can take a look into this as soon as possible.
IMPORTANT NOTE using ryzenmaster and checking your CCX and individual cores some cores are golden sample sorta 'go a bit faster than other cores maybe'. so it may be referring to the optimal core instead of the first one you see or might expect. But the other reason may be the app might be 'running as a service' or in background or isnt maximized fullscreen exclusive while youre MONITORING SOFTWARE IS IN THE BACKGROUND.. you see idle tasks switch between cores and performance usage based on foreground and background to multitask and manage shared resources.
all dx9 dx11 arent multicore or multithreaded.. BUT they ARE single thread.
So in 60's if your giant slab of computer runs at its one setting all day every day it overheats.
So they made single thread software to be like 2% or 5% of whatever it SHOULD be running at so it doesnt ever get close to overheating.
They took a big fat pipe of bandwidth and CPU engine power and made it use nylon rope hairs and only ever a single one..
since most computers on the market at the time were fake ones and not computers at all as they had high latency and were designed to be affordable for warehouses databases and offices they were making them for inventory and cash registers and banks.. which being a government service and a required utility service was cheap as possible and had no money involved in the banking industry at all as it never ever turns a profit and runs at cost price bank accounts exist by LAW and you must have and use one by law as a crime prevention measure. so if the government and law force you to open a bank account then take the money that they're supposed to be keeping safe from criminals they go to prison. law says you cant have more than a few thousand dollars cash items or money in your house or on your person as it encourages criminals. So they need computers to track and manage it so they're all the cheapest lowest cost solutions that get the job done usually. many banks almost never upgrade their older custom hardware. So these cash register type computers that can still run commercial commerce software such as IBMs COBOL commerce code later scribbled all over and renamed to C and C++ (more of the same hair thin single thread of nylon woven rope) means that your CPU usage will SHOW extremely high on a single core and SAY that its using 70% .. but you will have millions and millions of times WORSE performance than when DX12 shows 8% on a single core.
a better question is how many of my radeon 5700xt's 40 compute cores are being used by the GPU in a game when it shows 10 to 20% in thousands of times beyond ultra quality octillions of resolution 3 kinds of ray methods mortal kombat aftermath 11 ultimate
You will want to have it run on ALL cores not just the one. try forcing to use direct3d renderer in directx12 mode or force the use of VULKANRENDERER. by launching it with arguments on the commandline maybe in a shortcut or in the steam launcher link or by creating a config.ini text file and specifying what renderer to use CAPABILITYVULKAN1000 VXR VMA and so on? enable your infinitycache and infinityfabric in there.. uhh specify simultaneousinstantaeousmultiplepasses i guess? enable exclusivemode? FS2? VRR ? augmentedenhancedultramaxfullspeed maybe?
enable maybe a few thousands of tessellation? or 256xtessellation or infinitytessellation.. use infinity resolution maybe?
uhh try using infinityEQsupersampling?
disable ryzenmaster in task scheduler perhaps? then try again.
uninstall all .net and windows desktop runtime and C++ from installed APPs and add remove components and go download the latest from microsoft website.
Man, this is just that the threads are not loaded on the good chiplet with the cores with the highest boost as any other software that uses a couple of strong threads. That is all.
And this happens since the introduction of the new DX11 driver (May Preview, 22.5.2, 22.6.1, and 22.7.1) in DX11 games (another one would be Fall Guys, for example). Rolling back to the drivers prior to those, the loaded cores are the ones with the highest boost (thanks to the CPPC information). It is not overheating or anything like that.
There is some change in the scheduling policy that makes people with two-chiplet Ryzen performing worse than it should (in my case almost 10% lower), just because the load goes to the slow cores in the slow CCD/chiplet.
On Intel terms, it would be like loading the game threads on the E-Cores, while letting the P-Cores unused.
I do not use Ryzen Master BTW.
New driver (22.8.1) tested, the bug is still there
22.8.2 driver tested. Unsurprisingly, nothing was fixed.
Drivers 22.9.1/22.9.2 tested with no changes
honestly not sure if this is related, but i get random frame drops in dx11 games with the drivers 22.5.1, 22.5.2, 22.9.2, 22.9.1 and 22.Q2 and 22.Q3.
i havent tried older drivers and i dont know if its worth it?
specs: 5600x and 6800.
Hi, starting from the 22.5.2 drivers, a new DX11 driver was introduced. Your issue is different than my issue as my issue only happens with 2-CCDs cpus (5900X/5950X). However, it could be related to the new DX11 driver.
In my case, I do not have really framedrops, the game threads just go to the slowest cpu cores, otherwise the games run fine. So it is difficult to know what could be your problem. If you have any kind of overclocking/undervolting in cpu/ram/gpu, etc.; you should test without it.
i found out that it seems to be a known issue, at least community wise. amd didnt lose a single word about it.
its a driver issue, most likely shader related and possibly hard drive read/write related. both would explain the frame drops.
i had a 3070 before that and i had 0 issues with any games i played on it. i dont have any oc apart from xmp.
As I do in almost driver version, I have tested the driver 22.10.1. I am happy to say that the bug is already fixed in some DX11 titles such as Assetto Corsa and Heaven 4.0. However, there are some of them which are not, such as Valley, 3DMark Firestrike, Rocket League or Fall Guys.
At least they are trying to fix it and I appreciate it
I confirm, it is still present in 22.10.1 in Team Fortress 2 (and any other game based on hl2.exe), so a fix would be much appreciated. It tanks fps by 20% or even more depending on map (as compared to manually setting affinity to the two best cores and their SMT threads in Task Manager)
youre aware infinity fabric and infinity cache means communication between cores is mathematician certified advertised on the box INFINITELY fast right? sorry forgot not INFINITELY fast.. its infinitely fast for the communication core to core bus is the FABRIC but then its INFINITELY FASTER again for all communications that fit within cache size because of infinity cache.. so its infinityxinfinity fast.. and youre saying.. a single core is slow somehow in the better than world record breaking super computer parts of your AMD ryzen anything? and as per my previous post directx11 and well all directx is single thread single core if youve nvidia or intel anything so thats maybe your problem.
nvidia still isnt directx 9.. i can prove it.. turn on light shafts lens flares tessellation(tesseraction) and higher than just 4th dimension tesseracts 5th and 6th and 7th dimensional maths.. and change your game world from 2d raster to 3d.. for the DIRECT3D.. you've maybe heard of it before? then when you realise your nvidia rtx isnt even truelight.. it cant lense flare they're like a photoshop 2d overlay picture of games.. its not a 3d game world.. AMD has ray traced since 60's and its AMD ray tracing and 4th and beyond dimensional tesseraction (tessellation) for lighting was in voodoo3dfx and in playstation 2 and stuff.. so umm.. i think your problem is intel and nvidia mate. that and try VULKAN.
New update: I tested the newest drivers 22.10.2 and the issue still persists.
The games that were fixed with the drivers 22.10.1 are still fixed, the others that I tested which were not are still not fixed.
New driver 22.11.1 tested, still not fixed.
Driver 22.11.2 tested, no change.
Well, I update the information that I have until now:.
- This bug appears under some (not all) DX11 and DX12 titles. One thread gets stuck on the slower CCD with two-chiplet Ryzen, reducing the boost capabilities of the cpu.
- It happens in Zen 3 and I know that it happens in Zen 4 (a forum mate tested it for me) with two chiplets.
- In DX11 became evident after the May_Preview driver with the new DX11 driver. Before that I never saw it in DX11 games. It is more evident in low-threaded games/benchs as if there are one or two main threads, it is easier to spot that the slower CCD has a very big load (sometimes the limiting one), limiting the overall boost of the cpu.
- In DX12 is less evident because lots of games are already well-threaded. However, in gpu-limited scenarios, especially on open worlds (MS Flight Simulator, Cyberpunk) what it seems like the assets/scenario loading thread, which at that time is the highest load in the system, appears on the slower CCD, while the rest of the cpu threads are doing a very low amount of work (I included pictures already).
- This bug does not appear under Vulkan or OpenGL (tested in several games and benches)
- I "played" with the power plan settings with the software QuickCPU, thinking that this thread behaviour could be provoked by some power plan settings. I tested with one thread in Cinebench. Well, with different power settings I could force the thread to go to the quickest cores on the faster CCD, I could force the thread to go to a core on the slower CCD (according to windows, that would be like the "efficient" cores) and I could make windows to ignore CPPC2 completely and that the thread moves randomly to any core available. This would do nothing in the games, that thread that it is stuck on the slower CCD remains there with any power plan settings. The only way to make a workaround is to make a new affinity mask to override whatever the driver or windows is applying.
- This does not happen with Nvidia cards. The mate that tested the bug with Zen 4 (7900X) changed from AMD to Nvidia and now the bug is gone.
- There are some games which are not affected like Assassins Creed: Origins/Odyssey, they may have their own affinity mask, I do not know.
- I saw this in Valley, Rocket League, Fall Guys, Firestrike, Cyberpunk 2077, MS Flight Simulator (DX11 and DX12), Superposition, less apparent in Deus Ex: MD, among others.
- I saw it also in Assetto Corsa and Heaven but they were fixed from the driver 22.10.1 on with Zen 3, but not with Zen 4.
- I can only guess that an assets loading thread (which sometimes is also inside the main thread of the game), a driver thread or something similar is forced to run directly or indirectly on the slow CCD.
- From my position and understanding I cannot go beyond that point of knowing what exactly is provoking this bug.
Same problem here
Driver version: any from preview May with DX11 optimizations until now
In DX11 games core load goes to the secod chiplet when a two chiplet AMD CPU is paired with a AMD GPU
Assetto Corsa 7950X and 6800XT
Assetto Corsa and 4090
This issue is been present for months now, I don't know if Microsoft is involved and should take a look into their scheduling system too, but it needs to be resolved. When forcing core load to CCD1 via task manager / process affiinity, there're noticeable performance gains.
Thanks in advance.
This issue still is there for Path of Exile, most of the time game thread is assigned to the second CCD, noticeble improvement if game is lassoed to the first CCD (5900x at fixed 4.8 ghz, tuned ram, 6900xt, 22.11.2 driver)
It'd be especially hilarious if they don't fix this before Zen4x3D is out, the driver would send the gaming thread to the CCD with less cache and that will decrease performance even more. Please fix it, AMD, it's not hard (it worked fine before)
If seen world of warcraft cpu load only putting load on ccd1 while in 22.11.2 it puts load on both ccd1 and 1 slow core on ccd2 not even picking the fastest one.
the game has a deadlock bug as well and does not report cpu usage in details tab in new task manager.
edit: wow can run dx11 as well as dx12 but on dx12 it uses ccd2 while on 22.5.1 it does not for example.
Another DX11 game to the list: Nobody Saves the World. In this case, it is not performance critical (it is capped at 120fps) but the load is on the wrong chiplet...
After 8 months of reporting the bug (easily reproducible) without any fix (only a couple of titles for Zen 3), I wonder what will happen when the two chiplet X3Ds are out..., I do not understand why we would need an Nvidia card to make the cpu work as it should. It is an AMD cpu for God´s sake.
Hey @Fantasma_N3D your msi RTSS profile is awesome! Can you please share with me where I can get it?
Hi, I have just answered you (I think) in oc.net.
the person that I contacted in oc.net was not you. Anyway I uploaded my RTSS template to dropbox so that anyone can take a look at it. However, I do not give any support to it and, if the hardware is different, the monitoring also may not work or partially not work and you have to adapt it yourself.
23.2.1 tested: no changes and with the hybrid X3D Ryzen almost here :facepalm:
During this time, a couple of games that show the wrong behavior (from scheduling perspective): Nobody Saves the World (DX11) and Recompile (DX11).
Games that are working apparently fine (from scheduling perspective): Marvel´s Avengers (DX12) and The Callisto Protocol (DX12).
I have the same behavior.
5900x with 6800XT
In my case i have optimized the cpu so that even with the slower ccd i get 4750 - 4850 MHz. But surely it could be a bigger problem with the 2 ccd 3D cache cpus.
The other thing are crashes in my main sim rFactor2 when loading to track.
This problem occours in every version higher than 22.5.1
23.2.2 tested: I was hopeful for the new release due to the DX11 fix but they may have fixed another thing, no changes so far.
I add another game with wrong scheduling behavior: FAR: Changing Tides (DX11)
From some mates with Zen 4 processors, they tested the drivers 23.2.1 and 23.2.2; this issue is fixed with two-chiplet Zen 4 cpus (7900X/7950X), at least, under Windows 11 (Win 10 untested).
However, Zen 3 remains, as far as I know (and I tested with Win 10) unfixed.....
No, it's not fixed, 23.3.2, 6900xt + 5900x on Windows 10 22h2
New drivers 23.3.1 tested: still not fixed on Zen 3.
AMD, please apply the same fix that was applied on Zen 4 processor with the same bug during the last couple of drivers.
23.3.2 tested: still not fixed.
23.4.1 tested, no changes.
For the time being, I did some kind of workaround, in case someone is interested:
With the anniversary of the introduction of this bug, I tested the drivers 23.4.2 and 23.4.3. No changes so far. More games tested with the bug: Batman Arkhan Knight, Genshin Impact, Horizon Zero Dawn, Portal, ScourgeBringer, The Talos Principle, Unravel 2,...
u need to type in add m.gpu and add m.cpu and m.usb and other things.
then give it another go. also dx11 isnt multicore multithreaded because intel nvidia disabled it.. your AMD light cell processor mathematically measuring weighing and assessing and processing all radiation data input audio or video or gamma rays or theta or zeta whatever you can name light and sound particles and waves data 3d like a point cloud in say .wav files to track oohh u know radiation particle wave formations. then give that a go on.. windows 95.. or some dos looking stuff the space and military used decades before windows im guessing.
your AMD device is a real computer but game and app developers LITERALLY cant math.. real computers like AMD do universal maths, constants and imaginary numbers and radiation sciences and true photographic RNA DNA REALITY 1:1.. mathematicians are the most precisely measuring and evaluating people in the universe.. how could their 1:1 RNA DNA universe true all forms of radiation (true audio) faster than reality freely syncing with existence and any input or output devices true analog ever not run a 'car sim'
New driver 23.5.1 and new chipset drivers (just in case) tested, no fix yet.