F5F Stay Refreshed Hardware Desktop High temperatures at 3900x

High temperatures at 3900x

High temperatures at 3900x

K
Koollojoe
Posting Freak
830
11-25-2019, 03:39 AM
#1
Regular post, understand? Stop trying to impress me too much. I tested three coolers and two pastes. Stock, Scythe Ninja 5, H100i V2, plus paste with Scythe Ninja 5 and Ceramique 2. From my earlier checks, Ceramique 2 performed better than MX-4.

According to Ryzen Master settings: CCD0 4.2ghz, CCD1 4.0ghz, 1.1v core (core voltage set to 1.1v in BIOS). Temperatures were measured with HWinfo. Used Aida 64 CPU+FPU, Cinebench R20, and two simultaneous libx265 encodes at 1080p. Main runs were 10, slow, animation, CRF18.

Stock paste: 94c - 86c - 95c
Scythe Ninja 5: 83c - 73c - 84c (spikes up to 90c+ still happen a fair amount)
H100i V2: 99c - 91c - 99c
Scythe Ninja 5 paste: ~0.5c better than Ceramique 2.

I see others getting lower temps at higher voltages, so I’m a bit puzzled.
K
Koollojoe
11-25-2019, 03:39 AM #1

Regular post, understand? Stop trying to impress me too much. I tested three coolers and two pastes. Stock, Scythe Ninja 5, H100i V2, plus paste with Scythe Ninja 5 and Ceramique 2. From my earlier checks, Ceramique 2 performed better than MX-4.

According to Ryzen Master settings: CCD0 4.2ghz, CCD1 4.0ghz, 1.1v core (core voltage set to 1.1v in BIOS). Temperatures were measured with HWinfo. Used Aida 64 CPU+FPU, Cinebench R20, and two simultaneous libx265 encodes at 1080p. Main runs were 10, slow, animation, CRF18.

Stock paste: 94c - 86c - 95c
Scythe Ninja 5: 83c - 73c - 84c (spikes up to 90c+ still happen a fair amount)
H100i V2: 99c - 91c - 99c
Scythe Ninja 5 paste: ~0.5c better than Ceramique 2.

I see others getting lower temps at higher voltages, so I’m a bit puzzled.

G
Goku_1_1_1_
Junior Member
7
11-28-2019, 05:09 PM
#2
The voltage during testing is 1.1V.
G
Goku_1_1_1_
11-28-2019, 05:09 PM #2

The voltage during testing is 1.1V.

S
Sentio_Cege
Member
246
11-29-2019, 03:11 PM
#3
It stays consistent all the time at 1.1v. Don’t forget, the Scythe Ninja 5 doesn’t heat up at all and its temperature plummets to around 40°C when idle, almost instantly.
S
Sentio_Cege
11-29-2019, 03:11 PM #3

It stays consistent all the time at 1.1v. Don’t forget, the Scythe Ninja 5 doesn’t heat up at all and its temperature plummets to around 40°C when idle, almost instantly.

L
louis69
Junior Member
3
11-29-2019, 04:38 PM
#4
I thought about it carefully. What if we focus just on PBO? Probably overclocking the Ryzen 3000 caused some odd problems, but I might be wrong.
L
louis69
11-29-2019, 04:38 PM #4

I thought about it carefully. What if we focus just on PBO? Probably overclocking the Ryzen 3000 caused some odd problems, but I might be wrong.

R
randomabby
Senior Member
476
11-30-2019, 10:34 AM
#5
I notice temperatures often exceed 90°C during mine runs with NH-U12A (stock). 1.1V should significantly reduce those readings. In the hardware info, I see about 140W during stress tests and occasional spikes to around 95°C.
R
randomabby
11-30-2019, 10:34 AM #5

I notice temperatures often exceed 90°C during mine runs with NH-U12A (stock). 1.1V should significantly reduce those readings. In the hardware info, I see about 140W during stress tests and occasional spikes to around 95°C.

H
herobrine3959
Senior Member
443
12-02-2019, 05:58 AM
#6
Manual settings showed increased temperatures, voltages, and reduced clock speeds. TDC was 90a, EDC 114a, average power 141.5w. Temp was 87.4°C; other tasks likely influenced the result.
H
herobrine3959
12-02-2019, 05:58 AM #6

Manual settings showed increased temperatures, voltages, and reduced clock speeds. TDC was 90a, EDC 114a, average power 141.5w. Temp was 87.4°C; other tasks likely influenced the result.

T
The_Error132
Junior Member
43
12-02-2019, 02:08 PM
#7
It looks like lower voltage means higher current for the same power, which suggests a possible issue with the mounting or solder quality between the IHS and the die. From the Buildzoid video, I saw someone cooling a 3950X to around 90°C while it was running at 240W on an AIO unit.
T
The_Error132
12-02-2019, 02:08 PM #7

It looks like lower voltage means higher current for the same power, which suggests a possible issue with the mounting or solder quality between the IHS and the die. From the Buildzoid video, I saw someone cooling a 3950X to around 90°C while it was running at 240W on an AIO unit.

A
adem2001
Junior Member
4
12-04-2019, 03:16 AM
#8
I've reattached the cooler four times, checking the mounting force to find the optimal setting, but no significant variation was observed.
A
adem2001
12-04-2019, 03:16 AM #8

I've reattached the cooler four times, checking the mounting force to find the optimal setting, but no significant variation was observed.

M
MessiFan4evr
Member
119
12-05-2019, 08:21 PM
#9
What case and fan configuration are you using? During Cinebench R20 and Prime95 runs, temperatures stay below 73°C with fans under 900 RPM and the pump at full speed. The peak temps I've recorded are near 82°C during MMO sessions with 18 instances loading new data, especially after a long startup. These readings are similar to what I see with a 2700x setup. I haven't used a X570 motherboard yet, but I'm curious if it's due to my BIOS version or a particularly efficient chip.
M
MessiFan4evr
12-05-2019, 08:21 PM #9

What case and fan configuration are you using? During Cinebench R20 and Prime95 runs, temperatures stay below 73°C with fans under 900 RPM and the pump at full speed. The peak temps I've recorded are near 82°C during MMO sessions with 18 instances loading new data, especially after a long startup. These readings are similar to what I see with a 2700x setup. I haven't used a X570 motherboard yet, but I'm curious if it's due to my BIOS version or a particularly efficient chip.

A
aliali121212
Member
52
12-11-2019, 05:18 PM
#10
Seems like a poor connection with the cooler.
A
aliali121212
12-11-2019, 05:18 PM #10

Seems like a poor connection with the cooler.