F5F Stay Refreshed Hardware Desktop Limit for Ryzen 7950X3D is defined by PBO settings.

Limit for Ryzen 7950X3D is defined by PBO settings.

Limit for Ryzen 7950X3D is defined by PBO settings.

Pages (3): Previous 1 2 3 Next
D
Derp7575
Member
184
09-29-2023, 03:07 AM
#11
You said yourself that you lost 10% performance so it isn't for no reason. But if you are fine with it I won't complain or argue with you. There is no right or wrong here. As an owner of a 7700x and a 9800X3D I adjusted my fan curve to be a constant 30% instead of the stock fluctuation. 30% constant fan speed until 94c then 100% fan speed because THAT is where the CPU needs cooling. If you check the videos above. The CPU doesn't care if its at 54c or 89c, the AM5 is a bit different in this regard. The 30% fan speed is enough to keep my CPU under 90c during heavy gaming and also to keep my PC quiet while PBO and CO is enabled so I gain even more performance. We can't compare the Ryzen 7000/9000 to any other previous CPU from AMD or Intel. It takes some time to adjust to that 90c is actually okay. This is my fan cuve
D
Derp7575
09-29-2023, 03:07 AM #11

You said yourself that you lost 10% performance so it isn't for no reason. But if you are fine with it I won't complain or argue with you. There is no right or wrong here. As an owner of a 7700x and a 9800X3D I adjusted my fan curve to be a constant 30% instead of the stock fluctuation. 30% constant fan speed until 94c then 100% fan speed because THAT is where the CPU needs cooling. If you check the videos above. The CPU doesn't care if its at 54c or 89c, the AM5 is a bit different in this regard. The 30% fan speed is enough to keep my CPU under 90c during heavy gaming and also to keep my PC quiet while PBO and CO is enabled so I gain even more performance. We can't compare the Ryzen 7000/9000 to any other previous CPU from AMD or Intel. It takes some time to adjust to that 90c is actually okay. This is my fan cuve

B
bkisbeast1
Member
63
09-29-2023, 04:25 AM
#12
I don't suggest it because the 3D CCD runs about 20°C hotter under full load than the others. At 65°C it's quite low, but 75°C seems more reasonable. Only performance issues would appear in all core loads, and the 3D CCD would cap around 60°C max.
B
bkisbeast1
09-29-2023, 04:25 AM #12

I don't suggest it because the 3D CCD runs about 20°C hotter under full load than the others. At 65°C it's quite low, but 75°C seems more reasonable. Only performance issues would appear in all core loads, and the 3D CCD would cap around 60°C max.

N
NinaVanSteijn
Junior Member
44
10-15-2023, 10:08 AM
#13
I just want to share my thoughts without sounding like I’m being critical. It’s not about right or wrong—just my perspective.
N
NinaVanSteijn
10-15-2023, 10:08 AM #13

I just want to share my thoughts without sounding like I’m being critical. It’s not about right or wrong—just my perspective.

G
GamerGirl119
Member
60
10-15-2023, 11:39 AM
#14
Absolutely, thank you for your feedback. I understand adjusting the settings for higher temperatures. Perhaps setting the fan curves to 30% up to 80 degrees would work.
G
GamerGirl119
10-15-2023, 11:39 AM #14

Absolutely, thank you for your feedback. I understand adjusting the settings for higher temperatures. Perhaps setting the fan curves to 30% up to 80 degrees would work.

M
63
10-15-2023, 07:13 PM
#15
35k is solid for a 7950x3D in R23. The 7950x3D becomes TDP limited. The best thing I found to tune the 7950x3D was by reducing SoC voltage to 1.1V or less. This frees up about 6W from the IOD from ~18W to ~12W for the CCDs. When you're limited to like 150W, that 6W can matter. If you want a lot of the data, here's a whole thread that starts with CCD testing but leads into direct-die. People go nuts about temperatures on 5000x3D and 7000x3D but its really a non issue.
M
ManSpeeltGames
10-15-2023, 07:13 PM #15

35k is solid for a 7950x3D in R23. The 7950x3D becomes TDP limited. The best thing I found to tune the 7950x3D was by reducing SoC voltage to 1.1V or less. This frees up about 6W from the IOD from ~18W to ~12W for the CCDs. When you're limited to like 150W, that 6W can matter. If you want a lot of the data, here's a whole thread that starts with CCD testing but leads into direct-die. People go nuts about temperatures on 5000x3D and 7000x3D but its really a non issue.

C
ChocoMonster
Member
145
10-15-2023, 10:19 PM
#16
Performing well is key, delivering solid results while staying discreet. Let's keep things running smoothly and aim for that 85% efficiency. A bit of edge is good—just don't go overboard.
C
ChocoMonster
10-15-2023, 10:19 PM #16

Performing well is key, delivering solid results while staying discreet. Let's keep things running smoothly and aim for that 85% efficiency. A bit of edge is good—just don't go overboard.

Z
zanawesome
Junior Member
11
10-15-2023, 11:56 PM
#17
But if it isn't capped at 5.25GHZ, then it's definitely an issue. Just a joke, though—it's practically undoable given its TDP limits. I don't believe any undervolting could push a 7950x3D to hit full 150W at 5.25GHz across all cores. I managed 5GHz with heavy undervolting and direct-die, so maybe something like LN2 direct-die could work.
Z
zanawesome
10-15-2023, 11:56 PM #17

But if it isn't capped at 5.25GHZ, then it's definitely an issue. Just a joke, though—it's practically undoable given its TDP limits. I don't believe any undervolting could push a 7950x3D to hit full 150W at 5.25GHz across all cores. I managed 5GHz with heavy undervolting and direct-die, so maybe something like LN2 direct-die could work.

Y
Yesah
Junior Member
10
10-17-2023, 02:19 PM
#18
I adjusted the PBO temperature limit to around 89 degrees, which seems to have flattened all fan curve fluctuations until 80 degrees. It looks like these chips behave differently compared to my old i7 8700—while it ran cool but not extremely fast, the flat fan curves are quite unusual.
Y
Yesah
10-17-2023, 02:19 PM #18

I adjusted the PBO temperature limit to around 89 degrees, which seems to have flattened all fan curve fluctuations until 80 degrees. It looks like these chips behave differently compared to my old i7 8700—while it ran cool but not extremely fast, the flat fan curves are quite unusual.

E
140
10-18-2023, 12:46 PM
#19
It seems you're used to a CPU that stays cool even at low power, but now you're seeing temperature changes. The thermal paste layout looks good, yet the issue persists—what could be causing the heat buildup?
E
Egyptian_Gamer
10-18-2023, 12:46 PM #19

It seems you're used to a CPU that stays cool even at low power, but now you're seeing temperature changes. The thermal paste layout looks good, yet the issue persists—what could be causing the heat buildup?

F
FaZeChad
Junior Member
2
10-18-2023, 02:19 PM
#20
5000x3D and 7000x3D differ significantly with the 3D v-cache module added. It introduces roughly a 20°C buffer between the die. In HWinfo64, checking the per-core temperature column reveals this difference. The CCD0 and CCD1 readings are about 20°C apart, and CCD0 frequently reaches its thermal threshold. This behavior is inherent to the chip design, which remained unchanged until the 9800x3D release. The 9800x3D version behaves distinctly, possibly offering better performance than the standard Ryzen 9000 due to the need to closely match the die layout for the 3D v-cache implementation.
F
FaZeChad
10-18-2023, 02:19 PM #20

5000x3D and 7000x3D differ significantly with the 3D v-cache module added. It introduces roughly a 20°C buffer between the die. In HWinfo64, checking the per-core temperature column reveals this difference. The CCD0 and CCD1 readings are about 20°C apart, and CCD0 frequently reaches its thermal threshold. This behavior is inherent to the chip design, which remained unchanged until the 9800x3D release. The 9800x3D version behaves distinctly, possibly offering better performance than the standard Ryzen 9000 due to the need to closely match the die layout for the 3D v-cache implementation.

Pages (3): Previous 1 2 3 Next