Overclock.net - An Overclocking Community - Reply to Topic
Thread: *Official* Intel DDR4 24/7 Memory Stability Thread Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
02-24-2020 10:53 PM
zGunBLADEz
Quote: Originally Posted by Jpmboy View Post
Quote: Originally Posted by zGunBLADEz View Post
@Jpmboy now im summoning you for real
look what i manage to do in this second evga board what you think...
im wondering if is even stable at this point just put 17/19/17 let the board do the rest just for the time been..
Edit: pulled my usb gsat and did a run
Drooped one of the timings and tweaked a few more
17/20/17/[email protected]
Tfaw 16
Trfc 320
On that test
Looks real good to me. tRCD below 20 not stable? That tRAS is likely subbed since it is shorter than CL+RCD+RTP (all three operations have to complete within the ras ticks, and if ras is too low it cycles or, more importantly the bios microcode subs in a value to correct the timing error - which is not reported to the OS.
I managed stable 4200 with a 9900X (and 64GB, no less), but alas, using the same sticks and R6EO board, this 10980XE can't manage 4000, so I run 3733 c15 (which is pretty low true latency). The issue with 4200 and 4000 with this 10980XE is RTLs. Can't get truly stable low RTLs over time with >3733, and since these are the most important "timings" 3733 actually performs better overall. My 7980XE/R6 Apex is happily humming along at 4000c16 with very tights secondaries and RTLs. It's been in compute duty for 3 weeks now, non-stop with 3 Titan Vs. Zero reason to tweak anything on that rig. [IMG class=inlineimg]/forum/images/smilies/smile.gif[/IMG]
Some screen shots:
Nice tho but i see your dilemma it makes sense tho i can do 125-130gb reads as it is still limited even when i run high multi and higher mesh i guess once you are in the 50-55ns you can call it a day in x299 even if your ram/cpu/mobo cant do better.....

The problem i have with that 20 or +3 over cl at the beginning i didnt understand why i cant lower it down on par with the other 2s.. Till my board went to heaven with a permanent code FF....

When i started cleaning everything around.. I noticed one of my sticks on the gold contacts theres a few of them that are pilled off.... The gold plating felt partially off... its not burn its been like that since day 1.. i never check them when i bought them but im sure it was like that ot it felt off.... but I never cant have stable that particular timings above lets say 3700mhz is always 3 ahead of cl.. Sometimes when i start droping down rtls that particular stick dissapear out of the 4 he is a kind of a maniac by itself. So when i notice the gold contacts i did a mem run without that stick at that time on my asus x299m on tri channel and hold and behold the timing sticked on the stress test without errors..

I can rma but i don't know if gskill will return me a same set of sticks that perform the same... This 2 kits of bdie do very nice timings i bought them together.. Trfc i can drop them as low as 300trfc with 4200MHz under 1.45mV even do 1t at that speed as well.. So its a gamble for me at this point so i just deal with it like that, to me its not that important in this system.. It will drop the latency a tiny bit but if i get another kit and i cant lower trfc as low then its like the same in the end...



I can try to plate the gold back.. I have done plating in japanese swords parts before its not that hard. But the plating kit was over 100$ back in those days... dont know exactly now i supposed it's higher now..


I left it running on ramtest still running it was like 18000% when i left to work [email protected]/20/[email protected] trfc is at 320 im tightening that one and tfaw as low as 12 the rest of timings auto is putting low ones on this board i will see.


Edit: well it passed with flying colors
im going to make the tras 38 and see if i can lower that trfc more at least to 290-300 if possible after some other tweaks..

rtls are 65s/10s @ 4200 all of them i can boot at 4400 but loose 2 sticks in the process try once just to see want it to make sure first 4200 was steady
02-24-2020 10:03 PM
munternet
Quote: Originally Posted by bei fei View Post
Here is my starting point for my new memory. G.Skill Sniper X F4-3466C18D-16GSXW

This is my first attempt. Any hints or tips on sub-timings are welcomed and appreciated.
Not sure if you have looked at the tutorial at the start of this forum and there is also a tutorial link in my sig.
Pay to have a read of both if you haven't already and report back. Looks like there is much to do
Are those sticks B-Die?
02-24-2020 08:25 PM
bei fei Here is my starting point for my new memory. G.Skill Sniper X F4-3466C18D-16GSXW

This is my first attempt. Any hints or tips on sub-timings are welcomed and appreciated.
02-24-2020 06:36 PM
Jpmboy
Quote: Originally Posted by zGunBLADEz View Post
@Jpmboy now im summoning you for real
look what i manage to do in this second evga board what you think...
im wondering if is even stable at this point just put 17/19/17 let the board do the rest just for the time been..
Edit: pulled my usb gsat and did a run
Drooped one of the timings and tweaked a few more
17/20/17/[email protected]
Tfaw 16
Trfc 320
On that test
Looks real good to me. tRCD below 20 not stable? That tRAS is likely subbed since it is shorter than CL+RCD+RTP (all three operations have to complete within the ras ticks, and if ras is too low it cycles or, more importantly the bios microcode subs in a value to correct the timing error - which is not reported to the OS.
I managed stable 4200 with a 9900X (and 64GB, no less), but alas, using the same sticks and R6EO board, this 10980XE can't manage 4000, so I run 3733 c15 (which is pretty low true latency). The issue with 4200 and 4000 with this 10980XE is RTLs. Can't get truly stable low RTLs over time with >3733, and since these are the most important "timings" 3733 actually performs better overall. My 7980XE/R6 Apex is happily humming along at 4000c16 with very tights secondaries and RTLs. It's been in compute duty for 3 weeks now, non-stop with 3 Titan Vs. Zero reason to tweak anything on that rig.
Some screen shots:
02-24-2020 04:38 PM
KedarWolf
Quote: Originally Posted by munternet View Post
I did better on my Maximus X Hero with 4 sticks than I did with 2 sticks as far as max frequency goes
Yeah, Gigabyte and Asus use T-Topology on their four stick boards which will overclock better with four DIMMs than two.
02-24-2020 04:35 PM
munternet I did better on my Maximus X Hero with 4 sticks than I did with 2 sticks as far as max frequency goes
02-24-2020 12:10 PM
fleps
Quote: Originally Posted by bp7178 View Post
I lost interest with the link when I saw it was a 2017 press release.
What would you expect from a board that was launched at 2017?
And if you lost interest, then refrain from posting things that aren't true?

I get that we are all learning and trying to help each other here but it's also important to be responsible to what we say mate, so we don't mislead other people that might arrive here.

You may had a bad experience with a X Hero and 4 dimms, but that doesn't make the ultimate true, specially when there's actual facts to prove it works.

Best.
02-24-2020 07:39 AM
bp7178
Quote: Originally Posted by fleps View Post
- I don't think you even opened the link mate, otherwise you wont be saying rubbish. The memories from G.skill got validated with a 4h run / 400% HCI as 19-19-19-39 timings.
- Who said I'm using XMP?
I lost interest with the link when I saw it was a 2017 press release.
02-24-2020 07:18 AM
chrcoluk
Quote: Originally Posted by ashr View Post
Is there any reason why I've lost performance over time? Is it due to security patches (9900k)? I've tried all stock/auto settings and I'm getting the same lower performance.
The top result is from 04/2019 and the bottom one is today.
memory is an i/o operation and i/o is hit by the mitigations.

Download a tool call inspectre and hit the disable mitigations button, then reboot, I am curious how much of your performance you get back from that.

Note it doesnt disable all of the mitigations but does disable the worst offenders.

Also if you wanted to inspectre allows you to enable the mitigations again just as easily.
02-24-2020 07:13 AM
chrcoluk
Quote: Originally Posted by swddeluxx View Post
it is here

yeah that is kick ass man

I find it funny that all this time I thought I couldnt 4 sticks b-die 3200mhz because of motherboard, and after I swap cpu I could, so the IMC on cpu's can hurt.
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off