F5F Stay Refreshed Power Users Overclocking Undervolting 3800X

Undervolting 3800X

Undervolting 3800X

Pages (2): 1 2 Next
S
Sythe135
Junior Member
32
11-09-2019, 11:05 PM
#1
I have a 3800X with PBO enabled (motherboard limits, 2x scalar). The system shows 4600Mhz max boosts, while average boosts under load are around 4200. Ambient temperature is 24°C, idle at 29°C, and Cinebench R20 scores between 5025-5050 at 75°C.

I've applied a -0.10V offset in the BIOS, which lowered the temperature during heavy load by 5°C and improved Cinebench scores by 25-50 points. Max boosts decreased to 4475/4500. I appreciate the temperature reduction and the slight performance gain with PBO. Currently, I'm conducting stability tests (p95) and everything seems stable with this offset.

I've also noticed several discussions on forums warning against negative voltage offsets. Was anyone else seeking advice on this topic?
S
Sythe135
11-09-2019, 11:05 PM #1

I have a 3800X with PBO enabled (motherboard limits, 2x scalar). The system shows 4600Mhz max boosts, while average boosts under load are around 4200. Ambient temperature is 24°C, idle at 29°C, and Cinebench R20 scores between 5025-5050 at 75°C.

I've applied a -0.10V offset in the BIOS, which lowered the temperature during heavy load by 5°C and improved Cinebench scores by 25-50 points. Max boosts decreased to 4475/4500. I appreciate the temperature reduction and the slight performance gain with PBO. Currently, I'm conducting stability tests (p95) and everything seems stable with this offset.

I've also noticed several discussions on forums warning against negative voltage offsets. Was anyone else seeking advice on this topic?

B
Bazooker
Member
72
12-01-2019, 07:29 AM
#2
If you don't mind tweaking and really want to find the sweet spot for PBO with your rig then check out this thread:
https://www.overclock.net/forum/13-amd-g...boost.html
The experienced Ryzen overclockers have been complaining how PBO boosting works with Ryzen 3000 compared to how it did with 2000 CPU's in that does not follow the PPT/TDC/EDC settings with any coherency. As it turns out, there is a bug in AGESA 1004b PBO. They've found a way to at least get better PBO performance, even if it's still not exactly 'coherent' to the settings.
And lastly: do not forget that CPU boosting is...
B
Bazooker
12-01-2019, 07:29 AM #2

If you don't mind tweaking and really want to find the sweet spot for PBO with your rig then check out this thread:
https://www.overclock.net/forum/13-amd-g...boost.html
The experienced Ryzen overclockers have been complaining how PBO boosting works with Ryzen 3000 compared to how it did with 2000 CPU's in that does not follow the PPT/TDC/EDC settings with any coherency. As it turns out, there is a bug in AGESA 1004b PBO. They've found a way to at least get better PBO performance, even if it's still not exactly 'coherent' to the settings.
And lastly: do not forget that CPU boosting is...

T
TemkaPlay
Member
160
12-01-2019, 02:52 PM
#3
People have been lowering voltages on CPUs for years. This approach is typical since each CPU is unique and requires varying voltage/current levels for stability. Intel and AMD establish higher stock voltages to accommodate any potential issues, which might be slightly above the CPU's actual needs or significantly more than required. Adjusting offsets affects VID, not vcore. VID represents the voltage the CPU requests from the VRM, while vcore is the voltage it actually receives. SVI2 TFN indicates the voltage the CPU is using, regardless of its demands or supply.
T
TemkaPlay
12-01-2019, 02:52 PM #3

People have been lowering voltages on CPUs for years. This approach is typical since each CPU is unique and requires varying voltage/current levels for stability. Intel and AMD establish higher stock voltages to accommodate any potential issues, which might be slightly above the CPU's actual needs or significantly more than required. Adjusting offsets affects VID, not vcore. VID represents the voltage the CPU requests from the VRM, while vcore is the voltage it actually receives. SVI2 TFN indicates the voltage the CPU is using, regardless of its demands or supply.

I
iJedi007
Member
157
12-02-2019, 01:48 AM
#4
Minor negative adjustments have proven useful for refining it to 'the most'. I'm applying a -0.0125 offset for my PBO'd 3700x, for example. However, the issue arises with significant undervolts, and I must steer clear of fixed voltages unless you're highly experienced.
The challenge with excessive undervolting is known as 'clock compression'. The CPU continues to show high frequencies but actual performance drops, particularly under lightly threaded tasks. My advice is to always run a CB20 single-thread test to check for any impact. Strong threading matters most in gaming, so preserving that is essential.
I
iJedi007
12-02-2019, 01:48 AM #4

Minor negative adjustments have proven useful for refining it to 'the most'. I'm applying a -0.0125 offset for my PBO'd 3700x, for example. However, the issue arises with significant undervolts, and I must steer clear of fixed voltages unless you're highly experienced.
The challenge with excessive undervolting is known as 'clock compression'. The CPU continues to show high frequencies but actual performance drops, particularly under lightly threaded tasks. My advice is to always run a CB20 single-thread test to check for any impact. Strong threading matters most in gaming, so preserving that is essential.

T
theclemblackbg
Junior Member
31
12-02-2019, 05:53 AM
#5
Thank you for the input. I've experimented with -0.050 and -0.10, noticing temperature multicore gains with -0.10. It appears my offset is eight times higher than yours, which worries me. Is a -0.10V regarded as a significant undervolt? I plan to test it with a similar value of 0.0125 to compare.
T
theclemblackbg
12-02-2019, 05:53 AM #5

Thank you for the input. I've experimented with -0.050 and -0.10, noticing temperature multicore gains with -0.10. It appears my offset is eight times higher than yours, which worries me. Is a -0.10V regarded as a significant undervolt? I plan to test it with a similar value of 0.0125 to compare.

S
SpillQ
Junior Member
46
12-02-2019, 02:44 PM
#6
If I applied a -0.100 offset, I believe ST performance would drop significantly with my system... it did so at only -0.0500 when I pushed it down to -0.0125. From what I understand, negative offsets can help under heavy all-core loads by reducing temperature, as you mentioned, but excessive use was too damaging for single-threaded tasks. It won't cause harm, just won't handle high boost clocks well during light, bursty workloads—exactly the kind of scenarios many games present.

BTW: with PBO, you might want to adjust these values instead of just setting them to maximum: PPT = 333, TDC and EDC both = 230. It seems to work quite effectively.
S
SpillQ
12-02-2019, 02:44 PM #6

If I applied a -0.100 offset, I believe ST performance would drop significantly with my system... it did so at only -0.0500 when I pushed it down to -0.0125. From what I understand, negative offsets can help under heavy all-core loads by reducing temperature, as you mentioned, but excessive use was too damaging for single-threaded tasks. It won't cause harm, just won't handle high boost clocks well during light, bursty workloads—exactly the kind of scenarios many games present.

BTW: with PBO, you might want to adjust these values instead of just setting them to maximum: PPT = 333, TDC and EDC both = 230. It seems to work quite effectively.

X
Xofreu1603
Member
86
12-07-2019, 11:58 AM
#7
Thanks for your input.
I have run a quick test with Single Threaded cinebench 20. Here are my current results:
Voltage Offset
Cinebench ST
0/Stock
513
-0.0125
508
-0.0500
508
-0.1000
511
What is interesting here, is that I see a definite uptick at -0.10V, and that is strange. I also see a loss of 1.00% ST performance at -0.05, but a gain of 1.26% MT, as well as lower temperatures by 5C.
X
Xofreu1603
12-07-2019, 11:58 AM #7

Thanks for your input.
I have run a quick test with Single Threaded cinebench 20. Here are my current results:
Voltage Offset
Cinebench ST
0/Stock
513
-0.0125
508
-0.0500
508
-0.1000
511
What is interesting here, is that I see a definite uptick at -0.10V, and that is strange. I also see a loss of 1.00% ST performance at -0.05, but a gain of 1.26% MT, as well as lower temperatures by 5C.

N
179
12-09-2019, 08:56 AM
#8
That's quite unusual... wow... I've never adjusted that much of an offset before. Maybe I'll give it a shot now! But it's also important to note that motherboard VRM performance can significantly affect this, so your board will behave differently under various loads. That's why you need to adjust until you find the best setting for your setup.

Additionally, since you're someone who likes fine-tuning, you might want to experiment with those PBO settings and see how they impact things.
N
NicknameNumber
12-09-2019, 08:56 AM #8

That's quite unusual... wow... I've never adjusted that much of an offset before. Maybe I'll give it a shot now! But it's also important to note that motherboard VRM performance can significantly affect this, so your board will behave differently under various loads. That's why you need to adjust until you find the best setting for your setup.

Additionally, since you're someone who likes fine-tuning, you might want to experiment with those PBO settings and see how they impact things.

A
AapenStaartje
Member
164
12-16-2019, 07:07 PM
#9
I conducted a brief test and the outcomes match PBO-4 without voltage offsets.
R1, R2, R3
AVG: 5,005.33
DEV: 10.97
CUSTOM: 4999, 5006, 5016
MB MAX: 5018
This shows no major deviation from motherboard specifications.
I plan to experiment with various offsets and your suggested configurations as well as other combinations.
A
AapenStaartje
12-16-2019, 07:07 PM #9

I conducted a brief test and the outcomes match PBO-4 without voltage offsets.
R1, R2, R3
AVG: 5,005.33
DEV: 10.97
CUSTOM: 4999, 5006, 5016
MB MAX: 5018
This shows no major deviation from motherboard specifications.
I plan to experiment with various offsets and your suggested configurations as well as other combinations.

S
SoccerGirl1521
Junior Member
15
12-18-2019, 01:17 PM
#10
Your setup seems different from the standard configurations. The Miy 3700x improved my MT and ST scores notably—ST rose from around 4965 to about 5155, and MT increased from 490 to 512. My MT score fluctuates a lot; a morning cold boot brings it up to 5193, but by midday it dips to 5145 after three runs with a minute rest. Also, when I adjusted the offset to -0.1000, my MT score fell by 500 points, dropping from roughly 5155 to 4653. I didn’t focus much on ST testing since it’s not critical at that stage. This suggests a difference likely tied to VRM performance under load, affecting how our boards handle these metrics.
S
SoccerGirl1521
12-18-2019, 01:17 PM #10

Your setup seems different from the standard configurations. The Miy 3700x improved my MT and ST scores notably—ST rose from around 4965 to about 5155, and MT increased from 490 to 512. My MT score fluctuates a lot; a morning cold boot brings it up to 5193, but by midday it dips to 5145 after three runs with a minute rest. Also, when I adjusted the offset to -0.1000, my MT score fell by 500 points, dropping from roughly 5155 to 4653. I didn’t focus much on ST testing since it’s not critical at that stage. This suggests a difference likely tied to VRM performance under load, affecting how our boards handle these metrics.

Pages (2): 1 2 Next