F5F Stay Refreshed Hardware Desktop RAM overclocking on Ryzen 5000 with 4 dimms issue

RAM overclocking on Ryzen 5000 with 4 dimms issue

RAM overclocking on Ryzen 5000 with 4 dimms issue

Pages (2): 1 2 Next
A
AnnKa03
Member
120
04-20-2020, 10:12 AM
#1
Hello everyone, I'm currently facing some unusual issues while trying to boost my RAM's overclock. My setup includes an ASUS Rog Strix X570 E-Gaming Ryzen 9 5900X with PBO and curve optimizer MSI Suprim X RTX 3080. It runs at 2x G.Skill Trident Z Neo DDR4-3600, with clock speeds of 16, 16, 16, and 36 MHz, totaling 32 GB (with four dimms). I managed to achieve stable 2000 MHz without any problems. However, I haven't pushed beyond that speed yet. I've tested various configurations but keep encountering instability when trying to increase the frequency further. I've checked DRAM calculators and found that setting "SAFE" values above the 3600 MHz rating causes a safe boot loop. I also experimented with loose timings from another guide, but even with settings like 18-22-22-42 and Cwl at 18, I couldn't reach 3800 MHz on the RAM. Despite adjusting voltages and keeping everything else at default or auto, the system just won't boot properly. I'm not very familiar with RAM overclocking and have only read some guides. It seems like something might be missing in my configuration. Could it be that the four dimms aren't functioning correctly at higher speeds? I suspect the RAM itself could handle higher frequencies if optimized properly. I wonder if others have managed to push speeds up to 4400 MHz while maintaining original timings, especially with stock voltages. My watercooled system already shows elevated temperatures during load, which might affect stability. The DIMM temperatures are currently around 45-58°C, but I think they should be more critical for booting. There are two ways to set RAM overclock on the ASUS board: using AI Tweaker's DRAM timings section or the AMD Overclocking page. I've been using AI Tweaker with Auto mode, but I'm not sure if it makes a difference. Any advice or guidance would be greatly appreciated. Thanks!
A
AnnKa03
04-20-2020, 10:12 AM #1

Hello everyone, I'm currently facing some unusual issues while trying to boost my RAM's overclock. My setup includes an ASUS Rog Strix X570 E-Gaming Ryzen 9 5900X with PBO and curve optimizer MSI Suprim X RTX 3080. It runs at 2x G.Skill Trident Z Neo DDR4-3600, with clock speeds of 16, 16, 16, and 36 MHz, totaling 32 GB (with four dimms). I managed to achieve stable 2000 MHz without any problems. However, I haven't pushed beyond that speed yet. I've tested various configurations but keep encountering instability when trying to increase the frequency further. I've checked DRAM calculators and found that setting "SAFE" values above the 3600 MHz rating causes a safe boot loop. I also experimented with loose timings from another guide, but even with settings like 18-22-22-42 and Cwl at 18, I couldn't reach 3800 MHz on the RAM. Despite adjusting voltages and keeping everything else at default or auto, the system just won't boot properly. I'm not very familiar with RAM overclocking and have only read some guides. It seems like something might be missing in my configuration. Could it be that the four dimms aren't functioning correctly at higher speeds? I suspect the RAM itself could handle higher frequencies if optimized properly. I wonder if others have managed to push speeds up to 4400 MHz while maintaining original timings, especially with stock voltages. My watercooled system already shows elevated temperatures during load, which might affect stability. The DIMM temperatures are currently around 45-58°C, but I think they should be more critical for booting. There are two ways to set RAM overclock on the ASUS board: using AI Tweaker's DRAM timings section or the AMD Overclocking page. I've been using AI Tweaker with Auto mode, but I'm not sure if it makes a difference. Any advice or guidance would be greatly appreciated. Thanks!

M
mfesta100
Junior Member
12
04-20-2020, 09:49 PM
#2
It might be that the silicon lottery ended by using two pairs instead of one full set of four. I’d consider overclocking with just one pair to check for improvements.
M
mfesta100
04-20-2020, 09:49 PM #2

It might be that the silicon lottery ended by using two pairs instead of one full set of four. I’d consider overclocking with just one pair to check for improvements.

S
sianxy
Member
54
04-21-2020, 12:55 AM
#3
the basic spec is just the minimum you're aiming for, i'd begin at 1.15v expecting around 1.175v, in the end it's a silicon lottery whether your imc can hit 2000 with 4sticks. the quickest route seems to be driving the voltage up to about 1.45 and checking how far you can push the 16-16-16. my old sticks managed 3866 with some extra headroom (though not 4000). you'll need a fan in those dimmer settings, or at least some airflow.
S
sianxy
04-21-2020, 12:55 AM #3

the basic spec is just the minimum you're aiming for, i'd begin at 1.15v expecting around 1.175v, in the end it's a silicon lottery whether your imc can hit 2000 with 4sticks. the quickest route seems to be driving the voltage up to about 1.45 and checking how far you can push the 16-16-16. my old sticks managed 3866 with some extra headroom (though not 4000). you'll need a fan in those dimmer settings, or at least some airflow.

H
hampus_1
Member
72
04-25-2020, 07:17 AM
#4
Are there any settings above 1.1v that are acceptable? So far I've seen guidance suggesting a top voltage of 1.1v for safety. I also understand this makes sense if I aim to set the FCLK to 2000 MHz, though it's still working well with 3600 on RAM. That’s a solid observation. If I can spare some time, I might experiment with two sticks and check if it unlocks faster speeds. In the long run, running four sticks could likely offer better performance compared to using just two at higher FCLK.
H
hampus_1
04-25-2020, 07:17 AM #4

Are there any settings above 1.1v that are acceptable? So far I've seen guidance suggesting a top voltage of 1.1v for safety. I also understand this makes sense if I aim to set the FCLK to 2000 MHz, though it's still working well with 3600 on RAM. That’s a solid observation. If I can spare some time, I might experiment with two sticks and check if it unlocks faster speeds. In the long run, running four sticks could likely offer better performance compared to using just two at higher FCLK.

L
Linkiechu
Member
145
04-27-2020, 05:43 PM
#5
1.2v represents the highest safe voltage, 1.1v is essentially the lowest for 4x8 3600 (some require 1.18 just to be safe, though that's the most problematic case I've encountered), but you're welcome to verify the source link provided.
L
Linkiechu
04-27-2020, 05:43 PM #5

1.2v represents the highest safe voltage, 1.1v is essentially the lowest for 4x8 3600 (some require 1.18 just to be safe, though that's the most problematic case I've encountered), but you're welcome to verify the source link provided.

V
VodLer_
Junior Member
37
04-27-2020, 08:31 PM
#6
Unless you keep the CL16, it might not offer much advantage even with better FLCK, as explained by Jayz2Cents who showed that boosting clocks at the cost of CAS doesn’t always improve performance in practice.
V
VodLer_
04-27-2020, 08:31 PM #6

Unless you keep the CL16, it might not offer much advantage even with better FLCK, as explained by Jayz2Cents who showed that boosting clocks at the cost of CAS doesn’t always improve performance in practice.

P
PJtheSeel
Junior Member
36
05-02-2020, 03:03 PM
#7
I'll verify and attempt increasing the soc voltage. 4x8 3600 remains stable at 1.0 V logic level.
P
PJtheSeel
05-02-2020, 03:03 PM #7

I'll verify and attempt increasing the soc voltage. 4x8 3600 remains stable at 1.0 V logic level.

R
RulwenJr
Posting Freak
786
05-02-2020, 05:46 PM
#8
The voltage should be near 1.4 volts, with a 1.35V boost for the 3866 model. Running at 1.35V on most boards won’t work.
R
RulwenJr
05-02-2020, 05:46 PM #8

The voltage should be near 1.4 volts, with a 1.35V boost for the 3866 model. Running at 1.35V on most boards won’t work.

N
Noodle27
Junior Member
5
05-02-2020, 07:04 PM
#9
I reached 1.5 V during testing with the cl16 3866. No result was observed. I will attempt adjusting the supply voltage immediately and provide an update.
N
Noodle27
05-02-2020, 07:04 PM #9

I reached 1.5 V during testing with the cl16 3866. No result was observed. I will attempt adjusting the supply voltage immediately and provide an update.

A
amyeenhoorn
Member
121
05-02-2020, 10:42 PM
#10
Tested 3800 Mhz at various voltages but failed to boot (no startup). Voltage range: 1.1 - 1.2V (0.05V steps). VDDG CCD voltage: automatic and 1.075V. IOD voltage: automatic and 1.075V. cLDO VDDP voltage: automatic and 1.05V. RAM operates at 1.45V using original timings.
A
amyeenhoorn
05-02-2020, 10:42 PM #10

Tested 3800 Mhz at various voltages but failed to boot (no startup). Voltage range: 1.1 - 1.2V (0.05V steps). VDDG CCD voltage: automatic and 1.075V. IOD voltage: automatic and 1.075V. cLDO VDDP voltage: automatic and 1.05V. RAM operates at 1.45V using original timings.

Pages (2): 1 2 Next