F5F Stay Refreshed Hardware Desktop Significantly reducing the CPU's boost settings can greatly improve frame rates.

Significantly reducing the CPU's boost settings can greatly improve frame rates.

Significantly reducing the CPU's boost settings can greatly improve frame rates.

Pages (3): 1 2 3 Next
N
NightmarePasta
Junior Member
11
01-26-2023, 06:33 PM
#1
Hello everyone. I found something unusual and I’m looking for your insights. I have two Ryzen profiles set up—one optimized for gaming and another with lower clock speeds for better efficiency. Profile 1 runs at 2825MHz @0.975V, while Profile 2 is at 4475MHz @1.25V. There was a moment when I forgot to switch between them while playing, and my frame rate dropped from 85 FPS to 75 FPS. Someone, could you explain how a heavily underclocked CPU actually improved my performance? I’ve attached a YouTube video showing the setup and the results. Thanks in advance!
N
NightmarePasta
01-26-2023, 06:33 PM #1

Hello everyone. I found something unusual and I’m looking for your insights. I have two Ryzen profiles set up—one optimized for gaming and another with lower clock speeds for better efficiency. Profile 1 runs at 2825MHz @0.975V, while Profile 2 is at 4475MHz @1.25V. There was a moment when I forgot to switch between them while playing, and my frame rate dropped from 85 FPS to 75 FPS. Someone, could you explain how a heavily underclocked CPU actually improved my performance? I’ve attached a YouTube video showing the setup and the results. Thanks in advance!

E
EssieFlo
Member
174
01-26-2023, 08:48 PM
#2
A possible explanation: Disabling vertical sync boosts frame rate but increases CPU usage. Lowering the CPU frequency reduces heat, making it more stable and capable of handling the extra load without overheating. The rise in FPS likely resulted from cooler temperatures after throttling. Updated March 18, 2022 by An0maly_76 Added details
E
EssieFlo
01-26-2023, 08:48 PM #2

A possible explanation: Disabling vertical sync boosts frame rate but increases CPU usage. Lowering the CPU frequency reduces heat, making it more stable and capable of handling the extra load without overheating. The rise in FPS likely resulted from cooler temperatures after throttling. Updated March 18, 2022 by An0maly_76 Added details

C
cecedabro
Member
182
01-27-2023, 05:44 AM
#3
Observe that GPU consumption decreased alongside higher FPS. It appears the game adjusts to lower system power. I don’t know the title, so it’s unclear what this means.
C
cecedabro
01-27-2023, 05:44 AM #3

Observe that GPU consumption decreased alongside higher FPS. It appears the game adjusts to lower system power. I don’t know the title, so it’s unclear what this means.

G
Gerbinnos
Junior Member
4
01-27-2023, 06:29 AM
#4
Abnormal outcomes often stem from inadequate testing.
G
Gerbinnos
01-27-2023, 06:29 AM #4

Abnormal outcomes often stem from inadequate testing.

F
Fred10244
Posting Freak
937
01-27-2023, 06:35 AM
#5
The Afterburner visualization doesn't provide much clarity. We lack sufficient details. Could you include per-thread utilization so we can observe how CPU threads are being utilized? Ideally, if downclocking functions correctly, we should notice higher utilization compared to Profile 2, since the cores would be using more of their available capacity as resources become tighter. I also believe you're overlooking the impact of having only two cores at 2825MHz while others run at 600MHz. That imbalance could skew your results. My idea is that by focusing boosting on just the top two cores, the game might shift its resources exclusively to those two, causing threads to stay on them instead of moving between cores. This could lead to a performance hit. In Profile 2, if the main thread constantly switches among all six cores, the cost of task switching—especially with data moving between L1 and L2 caches—might degrade overall performance. Another aspect I'm thinking about is the Nvidia driver, which handles scheduling and runs in multiple threads. It's possible your Profile 1 setup pushes the driver onto slower 600MHz cores, still within its capacity, but now it also has to manage task switching. This could be similar to how disabling Hyper-Threading or enabling Game Mode in older AMD setups improved performance by simplifying memory handling and reducing overhead. Using Process Lasso to pin processes to specific cores might also help, as it can prevent unnecessary context switches. Essentially, just like turning off HT/SMT sometimes boosts gaming speed, and how AMD's "Game Mode" would disable some cores for a unified cache, it seems your game could benefit from soft-limiting it to just two cores and four threads. The remaining cores stay available, allowing the game to request more resources when needed without frequent switching.
F
Fred10244
01-27-2023, 06:35 AM #5

The Afterburner visualization doesn't provide much clarity. We lack sufficient details. Could you include per-thread utilization so we can observe how CPU threads are being utilized? Ideally, if downclocking functions correctly, we should notice higher utilization compared to Profile 2, since the cores would be using more of their available capacity as resources become tighter. I also believe you're overlooking the impact of having only two cores at 2825MHz while others run at 600MHz. That imbalance could skew your results. My idea is that by focusing boosting on just the top two cores, the game might shift its resources exclusively to those two, causing threads to stay on them instead of moving between cores. This could lead to a performance hit. In Profile 2, if the main thread constantly switches among all six cores, the cost of task switching—especially with data moving between L1 and L2 caches—might degrade overall performance. Another aspect I'm thinking about is the Nvidia driver, which handles scheduling and runs in multiple threads. It's possible your Profile 1 setup pushes the driver onto slower 600MHz cores, still within its capacity, but now it also has to manage task switching. This could be similar to how disabling Hyper-Threading or enabling Game Mode in older AMD setups improved performance by simplifying memory handling and reducing overhead. Using Process Lasso to pin processes to specific cores might also help, as it can prevent unnecessary context switches. Essentially, just like turning off HT/SMT sometimes boosts gaming speed, and how AMD's "Game Mode" would disable some cores for a unified cache, it seems your game could benefit from soft-limiting it to just two cores and four threads. The remaining cores stay available, allowing the game to request more resources when needed without frequent switching.

R
radfable
Junior Member
12
01-27-2023, 03:27 PM
#6
The issue seems linked to how the CPU adjusts performance when switching between profiles. Running low profile with low voltage and clock speeds can cause the CPU to cool down quickly, but transitioning to a high profile might trigger immediate throttling due to temperature or power constraints. It’s likely the system needs time to stabilize before it reaches its power limits, which explains the sudden slowdown.
R
radfable
01-27-2023, 03:27 PM #6

The issue seems linked to how the CPU adjusts performance when switching between profiles. Running low profile with low voltage and clock speeds can cause the CPU to cool down quickly, but transitioning to a high profile might trigger immediate throttling due to temperature or power constraints. It’s likely the system needs time to stabilize before it reaches its power limits, which explains the sudden slowdown.

T
TheMaxish
Junior Member
20
01-31-2023, 04:01 PM
#7
The video game is Assassin's Creed Valhalla, yet I tried it afterward using Rise of the Tomb Raider, and the outcomes matched. I’ll need to review some settings upon my return from work.
T
TheMaxish
01-31-2023, 04:01 PM #7

The video game is Assassin's Creed Valhalla, yet I tried it afterward using Rise of the Tomb Raider, and the outcomes matched. I’ll need to review some settings upon my return from work.

N
nighthawk7108
Junior Member
34
02-19-2023, 01:26 AM
#8
Bad testing refers to conducting experiments without proper controls or reliable data, leading to unreliable outcomes. Your experience with Cinebench R20 and R23 shows good stability, but ensuring consistent results requires careful setup and validation.
N
nighthawk7108
02-19-2023, 01:26 AM #8

Bad testing refers to conducting experiments without proper controls or reliable data, leading to unreliable outcomes. Your experience with Cinebench R20 and R23 shows good stability, but ensuring consistent results requires careful setup and validation.

C
Craftery
Member
207
02-24-2023, 02:45 AM
#9
Compare against stock is the initial step since excessive overclocking may damage performance. Next, use a stable benchmark to ensure consistency, as some titles naturally vary in FPS.
C
Craftery
02-24-2023, 02:45 AM #9

Compare against stock is the initial step since excessive overclocking may damage performance. Next, use a stable benchmark to ensure consistency, as some titles naturally vary in FPS.

C
Cupcake_Rose
Posting Freak
844
02-28-2023, 07:29 AM
#10
I believe the explanation is flipped. I was talking about how CPU speed changes with thermal throttling versus when it stays cool. Lower temperatures generally mean better performance since the processor isn’t forced to slow down. When vertical sync is active, it limits FPS for smoother visuals on lower refresh rates. Disabling it lets FPS rise but demands more CPU power. Boosting the CPU frequency can speed things up, but raises heat. Reducing the frequency brings temps down, easing throttling and letting the CPU manage vertical sync more efficiently. This means underclocking helps maintain cooler operation, which in turn supports smoother performance. For displays that support higher rates, turning off sync shifts some rendering work to the CPU, boosting FPS at the cost of increased usage. However, an overclocked CPU under stress won’t help much because it’s already trying to protect itself. Underclocking lowers heat and gives the CPU more breathing room to handle the sync demands.
C
Cupcake_Rose
02-28-2023, 07:29 AM #10

I believe the explanation is flipped. I was talking about how CPU speed changes with thermal throttling versus when it stays cool. Lower temperatures generally mean better performance since the processor isn’t forced to slow down. When vertical sync is active, it limits FPS for smoother visuals on lower refresh rates. Disabling it lets FPS rise but demands more CPU power. Boosting the CPU frequency can speed things up, but raises heat. Reducing the frequency brings temps down, easing throttling and letting the CPU manage vertical sync more efficiently. This means underclocking helps maintain cooler operation, which in turn supports smoother performance. For displays that support higher rates, turning off sync shifts some rendering work to the CPU, boosting FPS at the cost of increased usage. However, an overclocked CPU under stress won’t help much because it’s already trying to protect itself. Underclocking lowers heat and gives the CPU more breathing room to handle the sync demands.

Pages (3): 1 2 3 Next