F5F Stay Refreshed Hardware Desktop High frame rate performance inconsistent throughout the scene.

High frame rate performance inconsistent throughout the scene.

High frame rate performance inconsistent throughout the scene.

Pages (2): 1 2 Next
T
TyLeck13
Junior Member
12
09-01-2018, 11:22 AM
#1
I set up a fresh OS and recall running Dying Light 2 at 280 FPS with the game bar on, using auto cores on Windows 11. After changing to Windows 10 and keeping the same settings, my FPS dropped to about 220. Switching back didn’t help. I’m puzzled since performance should match. I’ve updated everything, installed matching GPU drivers, and installed the same CPU. I really enjoy this CPU but worry I can’t unlock its full potential. Now I’m back on Windows 10 again; games like Cyberpunk still show the same score, yet Dying Light 2 no longer delivers those frames. This suggests other CPU-dependent titles might also suffer. My BIOS is up to date. My hardware specs: RTX 4090, 565.90 driver, 7950x3d, 5600MHz, 2x16GB Corsair Dominator XMP enabled, B650E-F, Strix Mobo, running on SSDs 980 Pro and 970 EVO plus.
T
TyLeck13
09-01-2018, 11:22 AM #1

I set up a fresh OS and recall running Dying Light 2 at 280 FPS with the game bar on, using auto cores on Windows 11. After changing to Windows 10 and keeping the same settings, my FPS dropped to about 220. Switching back didn’t help. I’m puzzled since performance should match. I’ve updated everything, installed matching GPU drivers, and installed the same CPU. I really enjoy this CPU but worry I can’t unlock its full potential. Now I’m back on Windows 10 again; games like Cyberpunk still show the same score, yet Dying Light 2 no longer delivers those frames. This suggests other CPU-dependent titles might also suffer. My BIOS is up to date. My hardware specs: RTX 4090, 565.90 driver, 7950x3d, 5600MHz, 2x16GB Corsair Dominator XMP enabled, B650E-F, Strix Mobo, running on SSDs 980 Pro and 970 EVO plus.

B
BOYLOLLOL
Junior Member
4
09-01-2018, 11:22 AM
#2
Yes, someone could share their results from a 4090 benchmark running Light 2 on 1440p at 1440p resolution with ray tracing and frame generation.
B
BOYLOLLOL
09-01-2018, 11:22 AM #2

Yes, someone could share their results from a 4090 benchmark running Light 2 on 1440p at 1440p resolution with ray tracing and frame generation.

F
58
09-01-2018, 11:22 AM
#3
Ensure a fresh Windows 11 setup with AMD chipset drivers for optimal performance. A clean installation is essential—downgrading and then upgrading isn’t viable. Even with a clean install, some users encounter minor problems; it’s best to restart and try again. Hardware Unboxed has a video covering this issue.
F
FreeThroatHugz
09-01-2018, 11:22 AM #3

Ensure a fresh Windows 11 setup with AMD chipset drivers for optimal performance. A clean installation is essential—downgrading and then upgrading isn’t viable. Even with a clean install, some users encounter minor problems; it’s best to restart and try again. Hardware Unboxed has a video covering this issue.

I
Im_Magical
Junior Member
6
09-01-2018, 11:22 AM
#4
Here are some clean reinstall guides for Windows. I found a few videos that walk through the process step by step. Let me know if you'd like links or more details!
I
Im_Magical
09-01-2018, 11:22 AM #4

Here are some clean reinstall guides for Windows. I found a few videos that walk through the process step by step. Let me know if you'd like links or more details!

E
EnderSponge_
Member
225
09-01-2018, 11:22 AM
#5
E
EnderSponge_
09-01-2018, 11:22 AM #5

B
Bloemkool33
Member
223
09-01-2018, 11:22 AM
#6
The most straightforward method to check for spillover onto CCD1 is using Processor Affinity in Task Manager’s Details tab (selecting the app). This lets you restrict selection to cores 0-15, known as CCD0, and applies changes immediately. Once set up, the CPU becomes quite straightforward, but confirming a real issue requires turning off CCD1 entirely in the UEFI and adjusting its settings via the AMD Overclocking menu to match this configuration: 1-1-1-1-1-1-1-1 (CCD0) 0-0-0-0-0-0-0-0 (CCD1). Problems with spillover explain why the 7800x3D often outperforms the 7950x3D, despite the latter’s higher boost clocks on CCD0. When spillover occurs, the chip behaves more like a 7950x than a 7800x3D. CPPC preferred core settings helped me bypass the 3D v-cache optimizer driver, which previously caused performance hiccups. It never became an issue for me afterward, but I only used Windows 11 Pro with my 7950x3D after ensuring direct-die cooling and fine-tuning. Disabling CCD1 exclusively was necessary once I reached optimal temperatures and wall-mounted tuning, as CCD1 would disrupt the 5.25GHz boost during gaming.
B
Bloemkool33
09-01-2018, 11:22 AM #6

The most straightforward method to check for spillover onto CCD1 is using Processor Affinity in Task Manager’s Details tab (selecting the app). This lets you restrict selection to cores 0-15, known as CCD0, and applies changes immediately. Once set up, the CPU becomes quite straightforward, but confirming a real issue requires turning off CCD1 entirely in the UEFI and adjusting its settings via the AMD Overclocking menu to match this configuration: 1-1-1-1-1-1-1-1 (CCD0) 0-0-0-0-0-0-0-0 (CCD1). Problems with spillover explain why the 7800x3D often outperforms the 7950x3D, despite the latter’s higher boost clocks on CCD0. When spillover occurs, the chip behaves more like a 7950x than a 7800x3D. CPPC preferred core settings helped me bypass the 3D v-cache optimizer driver, which previously caused performance hiccups. It never became an issue for me afterward, but I only used Windows 11 Pro with my 7950x3D after ensuring direct-die cooling and fine-tuning. Disabling CCD1 exclusively was necessary once I reached optimal temperatures and wall-mounted tuning, as CCD1 would disrupt the 5.25GHz boost during gaming.

T
tjbee
Member
77
09-01-2018, 11:22 AM
#7
Consider using auto to save time, but keep the cache for faster access. This way you benefit from both efficiency and speed during your next clean install.
T
tjbee
09-01-2018, 11:22 AM #7

Consider using auto to save time, but keep the cache for faster access. This way you benefit from both efficiency and speed during your next clean install.

H
hugog07
Member
107
09-01-2018, 11:22 AM
#8
You might want to test it, as it caused excessive stuttering so I disabled it. The 'cache' feature should only activate when needed, which is beneficial in certain situations. Usually, I saw it to achieve the desired behavior—running one app on CCD0 and others on CCD1.
H
hugog07
09-01-2018, 11:22 AM #8

You might want to test it, as it caused excessive stuttering so I disabled it. The 'cache' feature should only activate when needed, which is beneficial in certain situations. Usually, I saw it to achieve the desired behavior—running one app on CCD0 and others on CCD1.

W
Wumty
Member
195
09-01-2018, 11:22 AM
#9
It determines this by checking the cache settings during installation. If you enable caching, you should configure it to auto before installing the drivers. After setup, switching to cache is straightforward—just install the Windows normally and then update the chipset drivers. No need to change steps unless you want automatic optimization.
W
Wumty
09-01-2018, 11:22 AM #9

It determines this by checking the cache settings during installation. If you enable caching, you should configure it to auto before installing the drivers. After setup, switching to cache is straightforward—just install the Windows normally and then update the chipset drivers. No need to change steps unless you want automatic optimization.

B
BlueDragon2
Junior Member
17
09-01-2018, 11:22 AM
#10
It's a pattern I noticed in that setup. The system would handle about 20% load on CCD1 during gameplay using CCD0, confirmed by some benchmarks in Warframe where the performance gap between CCD0 and CCD1 was 50% (160 fps versus 240 fps in the toughest case). If that 20% usage came from the game, it wouldn't have matched the results of CCD0 alone.
B
BlueDragon2
09-01-2018, 11:22 AM #10

It's a pattern I noticed in that setup. The system would handle about 20% load on CCD1 during gameplay using CCD0, confirmed by some benchmarks in Warframe where the performance gap between CCD0 and CCD1 was 50% (160 fps versus 240 fps in the toughest case). If that 20% usage came from the game, it wouldn't have matched the results of CCD0 alone.

Pages (2): 1 2 Next