Overclock.net banner
10701 - 10720 of 12824 Posts

·
Overclock the World
Joined
·
3,142 Posts
Still having error 2, 6 and 12
Should i use 2T instead?
Error #2 and #6 mostly relates to "not enough current"
Pure 6 at the start is about 20mV VDIMM missing
#2 is a long balancing and timing fight "something times out according to TM5 descriptons"

#12 and #0 are slightly timeouts, slight powering issues, mostly related to lack of VDIMM

On your point i think you just overvolt it - but i don't know
Haven't seen people struggle with CL15 thaat much, on such low MT/s
Do you remember your PCBs ?
CL16 on 3800 needed 1.36-1.38vDIMM, CL15 around the 1.46 mark
1.5+ is already quite a bit , as people get CL14 to run at 1.48v @ 3800MT/s - well but some also need 1.6v there :D

Generally, you only fight now with your powering issues, soo we'll see what can happen
If you don't resolve them, your primaries won't drop magically
Now i'm having some fun...
Fun indeed :D
You'll be stuck on this for a bit

Why don't you run 16-20-20-20-40-60 ?
Would be soo much easier & either work away tRCD 20 or up to 3800MT/s
Balancing high tRRD is not easy
5-7-4-12 runs,
5-7-5-14 does
6-8-5-14 also

other combinations need balancing elsewhere
Stay on lower ones till you are sure they are completely fine
Your errors continue to say "there are issues with them"
And your tFAW continues to be bad , even tho i told you to figure out tCCD_L from the bios and use this = tRRD_S * tCCD_L = tFAW

Seeing you can actually post and run this with 1.42v - 1.5v is far too much & you have issues elsewhere. You where overvolting the whole time as it seems
Give more procODT a try. I just now notice you are on Matisse ~ far less cLDO_VDDP , far less VDDG_IOD (wonder if i haven't told you)
And your tRTP or tWR is wrong. If you really use *8 here , then you need to continue this scaling = tRTP 8, 16. Same for tWR, 16,32 and so on
more values work ~ but i have to re'mention it
 
  • Rep+
Reactions: byDenoso

·
Registered
Joined
·
121 Posts
On your point i think you just overvolt it - but i don't know
Haven't seen people struggle with CL15 thaat much, on such low MT/s
Do you remember your PCBs ?
Yeap, is a A1 PCB, Hynix DJR


Generally, you only fight now with your powering issues, soo we'll see what can happen
If you don't resolve them, your primaries won't drop magically
Got it, i'll work hard on it.


Why don't you run 16-20-20-20-40-60 ?
Would be soo much easier & either work away tRCD 20 or up to 3800MT/s
Balancing high tRRD is not easy
Actually i've tested 16-20-20-40-60 and it worked stable, but i want to continue the jorney to "Find the missing MB/s/latency".


And your tFAW continues to be bad , even tho i told you to figure out tCCD_L from the bios and use this = tRRD_S * tCCD_L = tFAW
my tCCD_L is 5ns (7 clock i guess?) so tFAW would be 7xRdds?


Seeing you can actually post and run this with 1.42v - 1.5v is far too much & you have issues elsewhere. You where overvolting the whole time as it seems
Give more procODT a try. I just now notice you are on Matisse ~ far less cLDO_VDDP , far less VDDG_IOD (wonder if i haven't told you)
And your tRTP or tWR is wrong. If you really use *8 here , then you need to continue this scaling = tRTP 8, 16. Same for tWR, 16,32 and so on
more values work ~ but i have to re'mention it
I can lower CLDO_VDDP until 820mv, but i can't lower SOC and VDDG_IOD voltages lower than that because my Matisse just randomly restarts.
And worst of all, My CPU cant FCLK 1900, i've tried of several methods with no sucess.
 

·
Overclock the World
Joined
·
3,142 Posts
my tCCD_L is 5ns (7 clock i guess?) so tFAW would be 7xRdds?
Probably even gigabyte has an SPD-Z tool (named differently) that deciphers Intel XMP ~ inside the bios
It likely is different each frequency, but you can use it - yes
Either 4* or tCCD_L *
I can lower CLDO_VDDP until 820mv, but i can't lower SOC and VDDG_IOD voltages lower than that because my Matisse just randomly restarts.
Stay with 900mV then, even DRAM calculator mentioned this
More is not needed. Early on people could drop it to 700mV.
VDDG IOD and SOC you'll figure out together with y-cruncher
OC'ing T-Force 4133 cl18 bottom part of the post, but i think you've read it already
50mV stepping or 75mV , but 75mV doesn't always work

2T will need more ClkDrvStr, and 1T will also
If you want to run low CAD_BUS, you need a strong procODT
I think the opposite way is better & generates less heat
 

·
Registered
Joined
·
121 Posts
Probably even gigabyte has an SPD-Z tool (named differently) that deciphers Intel XMP ~ inside the bios
It likely is different each frequency, but you can use it - yes
Either 4* or tCCD_L *
Is 7 clocks, i used the ROG program to see it.
So will be 7*RDDS


Stay with 900mV then, even DRAM calculator mentioned this
More is not needed. Early on people could drop it to 700mV.
VDDG IOD and SOC you'll figure out together with y-cruncher
OC'ing T-Force 4133 cl18 bottom part of the post, but i think you've read it already
50mV stepping or 75mV , but 75mV doesn't always work
Apparently increase tRFC had decreased the amount of errors, just with error 4 for now..
 

·
Registered
Joined
·
81 Posts
And the result, 60 cycles, one single error 4 on the 58th cycle :/ but at least no freeze !
View attachment 2518242
Again gonna need your help guys.

So starting at the above settings ( one single error 4 at 58 cycles ), I tried to correct that by testing one by one:
  • ClkDrvStr at 30 instead of 40 => almost immediate error 8 ( in the first 10mn ).
  • Vdimm at 1.445 instead of 1.45 => error 10 after around 30 cycles.
  • RttNom /7 instead of /6 => error 1 after around 10 cycles.
  • RttPark /4 instead of /3 => error 12 after around 30 cycles, then error 14 on cycle 42.

Lowest latency observed on those tests was 54.6ns, most runs were between 54.7 and 55.2ns

I am currently trying with ProcODT 36.9 instead of 34.3, but kinda at a loss on what to do next.

It's like a never ending story, it seems to be almost stable though, so I am sure I am just missing that little thing that will push it stable ! Any idea what could help ?
 

·
Overclock the World
Joined
·
3,142 Posts
It's like a never ending story, it seems to be almost stable though, so I am sure I am just missing that little thing that will push it stable ! Any idea what could help ?
Probably an air conditioner, if the room keeps getting warmer and warmer on such long tests
or an UPS if house current is the issue
 

·
Registered
Joined
·
81 Posts
Probably an air conditioner, if the room keeps getting warmer and warmer on such long tests
or an UPS if house current is the issue
Temperature is pretty good, stable around 24-25C during the day and during the night tests it's actually going down.
Ram goes around 48-50C during the tests pretty fast ( around 10mn ) but not higher. Should I think of adding a waterblock to put the ram in my loop ?

Computer is already on a UPS ;)

It seems like there's something I am missing, like I am changing one thing at a time but maybe need to change a couple at the same time to reach stable ?

At that point, would you consider the single error 4 to not be an issue, knowing it appeared at 58 cycles ?

Or go with "Vdimm at 1.445 instead of 1.45 => error 10 after around 30 cycles" => change tWR to 10 instead of 12 after that since error 10 with just vdimm ?
 

·
Overclock the World
Joined
·
3,142 Posts
At that point, would you consider the single error 4 to not be an issue, knowing it appeared at 58 cycles ?

Or go with "Vdimm at 1.445 instead of 1.45
Dropping voltage doesn't seem to help you
it's too far into the test. It can be anything external at this point too
Like the CPU loosing stability at an hours long allcore test
 
  • Rep+
Reactions: gled_fr

·
Registered
Joined
·
224 Posts
Usually no, tCKE in combination with RTT_WR makes issues
Same as CAD_BUS TIMINGs do ~ but they coexist together with a lot of work

RTT_WR changes NOM and PARK behaviour once you enable or change it's strengthness
I do have a feeling /2 requires high VDIMM-IN to even function, but /3 has to run close to always
Using RTT_WR generally has a RTT_PARK range. Barely any influence to RTT_NOM
If WR trows errors ~ it's because WR keeps autocorrecting and something bothers it. Be it too low ClkDrvStr , tCKE usage or SETUP Times usage
If it fully refuses to post - either PARK is too strong , or non existent
I'd need an example of what doesn't work for you
What's the difference between tCKE 0 and tCKE 1? Do both qualify as "tCKE not used"? For pretty much anything above XMP my board sets tCKE to 1 if on Auto (meaning my whole progress was based on this setting).
 

·
Overclock the World
Joined
·
3,142 Posts
What's the difference between tCKE 0 and tCKE 1? Do both qualify as "tCKE not used"? For pretty much anything above XMP my board sets tCKE to 1 if on Auto (meaning my whole progress was based on this setting).
I'm sorry, I don't know.
Zero should not be used at all - because it can not be defined as such
I don't think it can even in HEX
tCKE on 1 is always on, while i phrased it wrongly ~ thank you @jomama22
tCKE max should be the opposite, yet i'm not sure how "always sleeping" can work while memory has to be always active.
But maybe it means "all cells except X are sleeping" ~ i am honestly not sure

It was a mystery that tCKE started to get used without running PD-Mode on Vermeer
And then also seeing/learning about Aggressive Powerdown and Normal Powerdown , which this tCKE value changes
Still a too foreign topic with lack of information. Surely the Chiphell community knows far more
At least the range is kind of figured out
 

·
Registered
5900x
Joined
·
5 Posts
You won't like this reponse :D
cLDO_VDDP is far to high
Match it as absolute minimum with VDDG CCD, which if really stable on stock ~ looks fine at 950mV
It is influencing at what MCLK you can post but lower is always better ~ as it allows for lower procODT to function (if remain powering is correct)
I'll take any info I can get so I appreciate the response! :)
In the time since I initially posted I have probably bookmarked about 5 of your other posts on this thread. Lots of great information, but sometimes a bit jarring to put it all together. With what I was able to find, I was able to get these settings to run through y-cruncher, karhu, tm5, and hci memtest without errors other than WHEA 19 popping up about twice a minute since boot. I wasn't able to find a clear answer on how to alleviate the WHEA errors other than a few discussions on this thread a few months back that didn't really have a resolution other than to drop down to 3733/1866 (which worked for me). Docp is off and most timings except the main 4, tcwl, and procODT are on auto.
2518384

Is this 4x8GB Micron Rev.E ?
Or maybe Hynix CJR ?
Should be 4x8GB Samsung B-die
2518385

You want to push ClkDrvStr to something along the lines of 30-20-30-20 for the start
Maybe 40-20-30-24 if the board struggles with 4 dimms

You also want to work GDM away and start your foundation as GMD off 2T ~ yes it's faster , and saves time for you
Give SD,DD's a try as 1-4-4-1-6-6 (tRDRD = 4-4)
In the ASUS Bios go to Tools - ASUS SPD-Z , disable first both armory crates and other asus spyware (if something more is left) & read out the XMP profiles
The Bios will generate tCCD_L value for you. Remember it , it is important
Check all 4 dimms if they have the same tCCD_L value
If it's 6, then use tRRD_S * tCCDL (6) = 42 for you.
If it's 7 , then you likely are looking at 49 as a value

Let us know which ICs these are , then we can talk about a baseline :)
For tRAS as a baseline you want tRCD*2 + tCCD_L . This will always work, doesn't matter how bad the kit is.
Then tRC = tRP + tRAS (which will also always work if tRAS is correct to begin with)
I'll try to mess around with some of these tonight. Per my above zentimings, I did manage to get GDM off 2T.
Getting rid of the extra ASUS bloatware was one of the first things I did on my new build! :LOL:
tCCD_L was blank under XMP, but was 6 under JEDEC. Sorry for the confusion on my part, but what timing is tRRD_S * tRRD_L plugged in for?

Thanks for taking the time to look at this! It is much appreciated! I try to learn what I can from searching this forum, but it's a lot info to take in and comprehend.

Edit: Read another post and saw tRRDS * tRRDL should be the value for tFAW. Messed around with some setting and am currently running some stability tests for these timings:
2518396
 

·
Overclock the World
Joined
·
3,142 Posts

·
Registered
5900x
Joined
·
5 Posts

·
Registered
B550 gaming edge, 5800x, 5600x, 3600xt, 1600AF, 2070S, 4x8 3200c14, 2tb adata 8200sx PRO, 500gb 970e
Joined
·
575 Posts
They are all labeled as ver 4.31(Corsair). It was actually one of the first things I checked when I received them.
for those to be B-die they act just like the C-die set i had. which
were HORRIBLE to overclock or even use XMP on AMD systems (built for AMD)
turned out they were C-die which for a long time i had thought they were B-die
for the timings your trying to run on those id have to go with what Veii is stating.
awful loose to be B-die. if those are indeed B-die you should have no issues running c16-16
or even with a little work something like this
(3200c14-14 tforce dark pro) see if you can do this set here tho (not c14-14) look at the second picture...
if those cant do c16-16 (4x8 set so you shouldn have issues) might need your own proc/RTT/Drv strength's etc

(EDIT) @Veii looks like those are indeed B-die says B-die finder...
2518407

2518405

2518404
 

·
Registered
Joined
·
81 Posts
Font Screenshot Software Technology Computer

Finally a stable run !

So either 36.9 ProcODT did the trick, or like @Veii said previous settings were already OK and something external triggered error 4

Now to work my way to CL14 and try to get lower latency !

it's good to see some progress :)
 

·
Registered
B550 gaming edge, 5800x, 5600x, 3600xt, 1600AF, 2070S, 4x8 3200c14, 2tb adata 8200sx PRO, 500gb 970e
Joined
·
575 Posts
i think i have a problem....(missing two {2400G-another 3600XT} which were sold...
didnt realize i had these in the closet....no wonder ive been broke since i started PC gaming/tinkering....
(clarify-knew i had the 5600x but thought i sold the 3600xt and gave the 1600af to my uncle)
(dont mind the shoes ETC got most of everything packed and ready to go for the morning.)
2518409
 

·
Registered
B550 gaming edge, 5800x, 5600x, 3600xt, 1600AF, 2070S, 4x8 3200c14, 2tb adata 8200sx PRO, 500gb 970e
Joined
·
575 Posts
Ok, so here lately ive been "gaming" more than what i was before,
and well me hitting 70c in DaysGone (cracked but regardless)
ive been wondering *** is causing this. as when i close the game im still hitting 65-higher
while with just chrome open. i thought "randomly" to open zenstates and what i found was x10 scaler..
which i have MANUALLY set inside bios to x2 (this happened on verified bios not just the modded one too)
so i set it back to x2 with zenstates (reason its strange is bc i set EDC/TDC/PPT DOWN tdc as low as 48, EDC as low as 80
and STILL (before seeing the x10 scaler) i was STILL pinging out 70c in game and 65+ at idle....
has ANYONE else had this happen? or have they had a "high ass temp" that they know should NOT be the case?
i can run dam near any game without issue but this x10 scaler "bug/issue" im getting is starting to make me wonder how this
is happening....
 

·
Registered
Joined
·
620 Posts
lol, my bad luck of the past week goes from annoying (having to return memory to Amazon for refund) to worse. Decided to give my loop early spring cleaning on the basis of reseating my CPU in case it was causing any issues with memory. All seemed to go well draining, cleaning, repasting and then putting back together. Usual leak tests done, no leaks.

While I booted up and posted I was having terrible issues getting my commander pro to be recognised in ICUE and Windows kept making the sound as if USB was being plugged/unplugged. Whilst trying to troubleshoot this the PC just turned off.

Then it seemed absolutely no power was going to it. Great, thought the power supply had blown. But I tested it with the jig that allows it, fans and anything directly connected to the power supply to power up (commander pro clearly working as fans and everything else run through this and it was powering).

So I took memory out, CPU out and graphics one by one trying to see if power would come back through the motherboard. Nothing.

Took the motherboard out of the case to test it suspended safely on its own, nothing. Still dead. What I have noticed is on the back down the bottom centrally if my power supply is connected and turned on it heats up ridiculously within seconds. To the point it'll get too hot to touch.

No idea what has happened, seems like a short-circuit of some sorts or something else. No evidence of water or water damage.

ASUS customer support so far have been absolutely dreadful. You spend all your time telling them what you've done to begin with, they respond with a generic have you done this, when you have. Now begins my 5th different agent who keep coming back asking stupid things like have you updated the BIOS when it appears your mobo is dead. If I finally get this to RMA (it's under warranty) goodness knows what to expect.

So on that note if you've got to the end of this post I'm considering buying an Aorus X570 master. If Asus take this back and repair it, great, I'll just sell it. Seems like it can be like 4-6 weeks or longer if Asus RMA something for repair.

Is the Gigabyte a good shout?

That's what auto set me to - 43.6, 24/24/24/24.
HCI Memtest is 2000% stable though.
Couldn't stabilize tRCDRD at 16, no matter what voltage I threw on it
Does it mean TM5 is more "robust"?

P.S. Made my assumptions back from my intel tests

That is vDIMM 1.425.
Yeah TM5 and y-cruncher seem to be the best ways to "punish" memory to make sure it is truly stable. Even then some TM5 profiles seem to push memory a bit more and/or can find out instability better than others.

1usmusv3 and anta777 seem to be the most recommended.

If you're not passing them then IMO you cannot call yourself stable even if you can bench and play games for weeks with no crashes.
 

·
Registered
Joined
·
717 Posts
inspect more of our brolken dual ccd units
Curios what you meant by "broken" ))
BTW talked to two "dual-CCD" 5600Х owners on a local forum. Both have no issues running 2000+ FCLK, no #19's, but supposedly have lower SIDD cores. One specimen can easily handle < -10 CO magnitudes on two best cores, that however reduces its SC performance noticeably.
 
10701 - 10720 of 12824 Posts
Top