F5F Stay Refreshed Hardware Desktop 4000MHz CL15 paired with 5800x configuration

4000MHz CL15 paired with 5800x configuration

4000MHz CL15 paired with 5800x configuration

W
WF_Catt
Posting Freak
761
04-13-2020, 12:00 PM
#1
I'm facing some challenges with my G.SKILL TridentZ 16GB RAM setup. I'm using it in the standard DOCP (XMP) mode, but there are a few strange behaviors I'm not sure about. After booting, the system tries to run at 4000MHz with an input frequency of 1800MHz and specific timing settings (15-16-16-36). However, once it starts, I notice some noise and distortion sounds.

I'm wondering if the recommended settings are too high, or maybe I need to adjust them. When I set the frequency to 2000MHz with a lower input frequency, the system boots normally but still shows voltage issues. In that case, the timing numbers don't match the expected 1:2 or 1:1 ratio. CPU-Z on FSBBig GrinRAM doesn't display the correct ratio, showing 3:54 instead.

For timing, the numbers 16-16-16-36 are tight, which seems fine, but the tRC (Bank Cycle Time) is at 94, which I think isn't ideal. My main concern is whether running it at 3600MHz with an input frequency of 1800MHz is the best path forward, or if I should stick with 4000MHz and 1800MHz. I'm also curious about the experience from others who tried similar configurations, like 3800MHz at 1900MHz FCLK.

What are your thoughts on this? Which setting do you recommend, and what potential problems might arise in the long term?
W
WF_Catt
04-13-2020, 12:00 PM #1

I'm facing some challenges with my G.SKILL TridentZ 16GB RAM setup. I'm using it in the standard DOCP (XMP) mode, but there are a few strange behaviors I'm not sure about. After booting, the system tries to run at 4000MHz with an input frequency of 1800MHz and specific timing settings (15-16-16-36). However, once it starts, I notice some noise and distortion sounds.

I'm wondering if the recommended settings are too high, or maybe I need to adjust them. When I set the frequency to 2000MHz with a lower input frequency, the system boots normally but still shows voltage issues. In that case, the timing numbers don't match the expected 1:2 or 1:1 ratio. CPU-Z on FSBBig GrinRAM doesn't display the correct ratio, showing 3:54 instead.

For timing, the numbers 16-16-16-36 are tight, which seems fine, but the tRC (Bank Cycle Time) is at 94, which I think isn't ideal. My main concern is whether running it at 3600MHz with an input frequency of 1800MHz is the best path forward, or if I should stick with 4000MHz and 1800MHz. I'm also curious about the experience from others who tried similar configurations, like 3800MHz at 1900MHz FCLK.

What are your thoughts on this? Which setting do you recommend, and what potential problems might arise in the long term?

S
196
04-13-2020, 08:44 PM
#2
Cl14 3733 indicates optimal performance with a 1:1 clock ratio.
S
StampyKittenNZ
04-13-2020, 08:44 PM #2

Cl14 3733 indicates optimal performance with a 1:1 clock ratio.

G
GabrielTonato
Junior Member
11
04-14-2020, 05:16 AM
#3
Work closely with the FCLK because of Ryzen's infinity fabric, which connects the CCDs. Running asynchronously doesn't harm RAM performance but reduces speed by 15-25%. It's uncommon for Zen 3 to hit 2000MHz FCLK, though it's possible depending on the chip. More often you'll reach 1900-1933MHz. You can lower the RAM speed to match and then improve timing. Also, the FSB ratio applies only to the front side bus, not the infinity fabric. That's why the numbers seem inconsistent.
G
GabrielTonato
04-14-2020, 05:16 AM #3

Work closely with the FCLK because of Ryzen's infinity fabric, which connects the CCDs. Running asynchronously doesn't harm RAM performance but reduces speed by 15-25%. It's uncommon for Zen 3 to hit 2000MHz FCLK, though it's possible depending on the chip. More often you'll reach 1900-1933MHz. You can lower the RAM speed to match and then improve timing. Also, the FSB ratio applies only to the front side bus, not the infinity fabric. That's why the numbers seem inconsistent.

N
Nixation
Member
222
04-14-2020, 05:28 AM
#4
Sure, let me clarify. These figures are around 1.5 to 1.5%, but I think there might have been a missing decimal point.
N
Nixation
04-14-2020, 05:28 AM #4

Sure, let me clarify. These figures are around 1.5 to 1.5%, but I think there might have been a missing decimal point.

A
anakindaur
Senior Member
576
04-15-2020, 09:25 PM
#5
Hardware Canucks detailed a 15% penalty for async usage. Additional tests indicate comparable outcomes, so you may explore further research yourself.
A
anakindaur
04-15-2020, 09:25 PM #5

Hardware Canucks detailed a 15% penalty for async usage. Additional tests indicate comparable outcomes, so you may explore further research yourself.

I
IMayBeDead
Senior Member
696
04-16-2020, 04:12 AM
#6
To operate c15 you need to disable gdm and likely choose 2T.
I
IMayBeDead
04-16-2020, 04:12 AM #6

To operate c15 you need to disable gdm and likely choose 2T.

1
1cocobanana
Junior Member
14
04-19-2020, 02:06 AM
#7
I did Google. That's why I had asked. Results seem to vary though. According to this site, it's not quite that big of a performance difference. 5% at best doing the math in my head.... https://www.overclockersclub.com/guides/...y_scaling/
1
1cocobanana
04-19-2020, 02:06 AM #7

I did Google. That's why I had asked. Results seem to vary though. According to this site, it's not quite that big of a performance difference. 5% at best doing the math in my head.... https://www.overclockersclub.com/guides/...y_scaling/