F5F Stay Refreshed Hardware Desktop Unusual behavior during high-speed performance tuning

Unusual behavior during high-speed performance tuning

Unusual behavior during high-speed performance tuning

M
Markitos174
Junior Member
11
07-23-2019, 05:53 PM
#1
Hey everyone, I'm dealing with some odd behavior on my 3600x. I'm using a Gigabyte X370 AORUS Gaming K7 with the newest BIOS, a Deepcool Captain 240 AIO water cooler, and 16GB of Samsung B-die RAM. My PSU is a modular Corsair 650W unit—either gold or platinum, rated for 80+. At first, I set PBO (advanced, boosted by 10x, +200Mhz) in the BIOS because I wanted better performance for Cyberpunk. But after testing, I noticed my core speeds hovered around 4.1Ghz while playing, which didn't match the expected 4.0Ghz max from Cinebench.

I tried Ryzen Master to push the speeds higher, setting a cap of 4600MHz on the CoreX and applying AutoOC. I kept PBO enabled. The results were mixed: Cinebench briefly reached 4.05–4.1Ghz for a few seconds before dropping back to 4.0Ghz. This made me wonder if the issue lies with the motherboard, especially since it's an X370 model and not optimized for PBO features.

I also experimented with RAM overclocking, default speed 2133, and ran various tests—voltages, LLC settings—but found another inconsistency: BIOS dynamic voltage always showed 1.1V regardless of changes. The AIO was placed at the top of the case, now set to intake mode with an open case and no filter.

After some adjustments, I re-seated the AIO and reapplied thermal paste, but it didn’t help. When I cleaned the paste after reseating, the pattern looked normal. However, I had past issues with the intake filter on my R7 1700.

Anyone have ideas? Could be a motherboard limitation, cooling setup, or BIOS quirk. Hope this doesn’t lead to any serious problems!
M
Markitos174
07-23-2019, 05:53 PM #1

Hey everyone, I'm dealing with some odd behavior on my 3600x. I'm using a Gigabyte X370 AORUS Gaming K7 with the newest BIOS, a Deepcool Captain 240 AIO water cooler, and 16GB of Samsung B-die RAM. My PSU is a modular Corsair 650W unit—either gold or platinum, rated for 80+. At first, I set PBO (advanced, boosted by 10x, +200Mhz) in the BIOS because I wanted better performance for Cyberpunk. But after testing, I noticed my core speeds hovered around 4.1Ghz while playing, which didn't match the expected 4.0Ghz max from Cinebench.

I tried Ryzen Master to push the speeds higher, setting a cap of 4600MHz on the CoreX and applying AutoOC. I kept PBO enabled. The results were mixed: Cinebench briefly reached 4.05–4.1Ghz for a few seconds before dropping back to 4.0Ghz. This made me wonder if the issue lies with the motherboard, especially since it's an X370 model and not optimized for PBO features.

I also experimented with RAM overclocking, default speed 2133, and ran various tests—voltages, LLC settings—but found another inconsistency: BIOS dynamic voltage always showed 1.1V regardless of changes. The AIO was placed at the top of the case, now set to intake mode with an open case and no filter.

After some adjustments, I re-seated the AIO and reapplied thermal paste, but it didn’t help. When I cleaned the paste after reseating, the pattern looked normal. However, I had past issues with the intake filter on my R7 1700.

Anyone have ideas? Could be a motherboard limitation, cooling setup, or BIOS quirk. Hope this doesn’t lead to any serious problems!

P
Prodi19
Member
66
08-02-2019, 07:42 AM
#2
It could be that the temperature readings aren't coming through clearly. Using a thermal camera might help identify the issue, and a laser temperature meter could also provide useful data. That would be my initial approach. If you have another method for checking temperatures physically, feel free to share. Edit: I may have misunderstood, but this is still a helpful way to verify.
P
Prodi19
08-02-2019, 07:42 AM #2

It could be that the temperature readings aren't coming through clearly. Using a thermal camera might help identify the issue, and a laser temperature meter could also provide useful data. That would be my initial approach. If you have another method for checking temperatures physically, feel free to share. Edit: I may have misunderstood, but this is still a helpful way to verify.

E
Erual
Member
66
08-02-2019, 08:34 AM
#3
Για να μπορείς να έχεις να άναμε, καθά πρέπει να ήσει η κληρότητα του λέξεού. Είναι ο θεσμός τον καλομέτρο? Τοι είναι, οι ράσις στην άνθρωπική για πολλά δεν είναι έχουν καλομέτρο. Είναι η θεσμό καλη με την καθαρρία των ράσις?
E
Erual
08-02-2019, 08:34 AM #3

Για να μπορείς να έχεις να άναμε, καθά πρέπει να ήσει η κληρότητα του λέξεού. Είναι ο θεσμός τον καλομέτρο? Τοι είναι, οι ράσις στην άνθρωπική για πολλά δεν είναι έχουν καλομέτρο. Είναι η θεσμό καλη με την καθαρρία των ράσις?

R
ReborntoKill
Posting Freak
821
08-04-2019, 01:32 AM
#4
It seems like it might be a thermal regulation issue where the system slows down to maintain stability, or possibly a hardware-related quirk that causes unexpected performance drops.
R
ReborntoKill
08-04-2019, 01:32 AM #4

It seems like it might be a thermal regulation issue where the system slows down to maintain stability, or possibly a hardware-related quirk that causes unexpected performance drops.

B
British_Walrus
Junior Member
13
08-09-2019, 03:50 AM
#5
Boosting voltage raises temperature, and Cinebench will quickly consume a 240mm AIO. My R7 1700 ran on a 240mm AIO, so pushing headroom isn’t much with that setup. A custom loop keeps me at 4.2ghz across all cores, which matches the performance of 3600 series chips.
B
British_Walrus
08-09-2019, 03:50 AM #5

Boosting voltage raises temperature, and Cinebench will quickly consume a 240mm AIO. My R7 1700 ran on a 240mm AIO, so pushing headroom isn’t much with that setup. A custom loop keeps me at 4.2ghz across all cores, which matches the performance of 3600 series chips.

R
RageGlitch
Posting Freak
771
08-09-2019, 04:19 AM
#6
I have limited knowledge of other Matisse CPU models, so it's hard to say if your temperatures are typical. Full synthetic loads often push temperatures higher, but a 3000-series member could help confirm if this is normal or an issue.
R
RageGlitch
08-09-2019, 04:19 AM #6

I have limited knowledge of other Matisse CPU models, so it's hard to say if your temperatures are typical. Full synthetic loads often push temperatures higher, but a 3000-series member could help confirm if this is normal or an issue.

J
JagerTeunvdC
Junior Member
4
08-11-2019, 06:41 AM
#7
I also purchased this AIO using my 1700 model, and I was able to reach up to 3.85 at temperatures near 85 degrees. That’s typical for that generation and core count, particularly considering the voltage was about 1.4 at that time.
J
JagerTeunvdC
08-11-2019, 06:41 AM #7

I also purchased this AIO using my 1700 model, and I was able to reach up to 3.85 at temperatures near 85 degrees. That’s typical for that generation and core count, particularly considering the voltage was about 1.4 at that time.

B
banshee45
Senior Member
726
08-15-2019, 03:04 AM
#8
After spending more time on this, it became clear I needed to re-flash the BIOS. Now the system is running smoothly at 4.025-4100 across all cores during synthetic benchmarks at 70°C, while Cyberpunk achieves 4.275-4.3 at 66°C. Even with all-core overclocking, reaching 4.2 GHz still requires going to 1.375 MHz. Appreciate the support!
B
banshee45
08-15-2019, 03:04 AM #8

After spending more time on this, it became clear I needed to re-flash the BIOS. Now the system is running smoothly at 4.025-4100 across all cores during synthetic benchmarks at 70°C, while Cyberpunk achieves 4.275-4.3 at 66°C. Even with all-core overclocking, reaching 4.2 GHz still requires going to 1.375 MHz. Appreciate the support!

K
Kay_macedo
Member
66
08-15-2019, 05:06 AM
#9
Great to hear!
K
Kay_macedo
08-15-2019, 05:06 AM #9

Great to hear!

G
GETSOU
Member
220
08-22-2019, 07:13 AM
#10
Appreciate the correction. GorujoCY has a valid point about the thermal camera. I usually carry one around, especially for routine checks on mining equipment. I used an Hti 301 model—it connects directly to your phone. It’s a bit pricey, but the performance is top-notch. It made sense since the machines were worth more than that. If you prefer something more affordable, a Seek Compact works just fine. Just keep in mind, the software readings aren’t fully trustworthy; it’s always wise to cross-check with an independent source.
G
GETSOU
08-22-2019, 07:13 AM #10

Appreciate the correction. GorujoCY has a valid point about the thermal camera. I usually carry one around, especially for routine checks on mining equipment. I used an Hti 301 model—it connects directly to your phone. It’s a bit pricey, but the performance is top-notch. It made sense since the machines were worth more than that. If you prefer something more affordable, a Seek Compact works just fine. Just keep in mind, the software readings aren’t fully trustworthy; it’s always wise to cross-check with an independent source.