[Sammelthread] Intel DDR5 RAM OC Thread

Wenn du müde bist, leg dich bitte hin.
Es hat keine Eile.

Ich bin dann auch in etwa ~30min weg 👋
1705519928132.png
mittlerweile 48min
Beitrag automatisch zusammengeführt:

@Veii
90min ist kritisch bzgl. temp :fresse:
1705520805196.png

Freitag kommt der Mo-Ra 420, dann wird das auch 😜
Beitrag automatisch zusammengeführt:


taugt da auch TM5 zum Stabilität testen? läuft ja bedeutend kühler
 
Zuletzt bearbeitet:
Wenn Du diese Anzeige nicht sehen willst, registriere Dich und/oder logge Dich ein.
Anhang anzeigen 960659
Das hier macht dir Panik ?

Es ist eine Feature.

Hitze ist nicht alles.
Tm5 anta's test ist heißer und testet anders
1usmus ist dennoch schlimmer.
Weit weg von Panik 😅
Einfach nur sehr ungewohnt, 7700k war meine letzte Intel CPU, danach 5900X ja und nun zurück bei Intel.
Ich weiß das sie es abkönnen... Muss mich da Mal dran gewöhnen 😁
 
Let's try 1.19v SA
Got error 2 after 40 minutes.
I had a feeling that 1.56V VDD is too much, that can also cause the issues.
So reduced it to 1.50V, no bueno, can't boot. Raised to 1.52V keeping the 1.47V VDDQ. Got 2 error 6 in the first round.
So raised VDDQ and the TX too to keep my fav delta.
The current test is running with these voltages:
VDD:1.52V
VDDQ:1.50V
TX: 1.35V
IMC_VDD: 1.50V
SA: 1.19V
Timings are the same.
1705555089822.png



I remembered that I was able to pass 10 rounds with much lower voltages and found the screenshot. This was with HT off, but for DIMM voltages it shouldn't matter I believe, they could do that with proper settings.

1705555431552.png
 

Anhänge

  • 1705555420150.png
    1705555420150.png
    267,6 KB · Aufrufe: 58
DIMMs wasserkühlen? ;)
Hab ein schönes Setting gefunden (VDDQ-trianing off und nachdem ich schöne Spannungen gefunden hab wieder on). DIMM-Flex zieht die Timings bei überhitzung, er bootet mit tREFI 131071, tRFC 672 und tREFCpb 448, RDRD_sg 15, WRWR_sg 12... bei ca. 50 Grad DIMM-temp wird dann tREFI auf 65535 reduziert und > 60 Grad nimmt er die Timings so zurück wei in dem Screenshot. Läuft auch gerade nach ner stromlosen Nacht sauber in ycruncher seit ner Stunde. Hoffentlich bleibt das... PS: war ein test mit Lüfter aus, normal erreicht das solche Temperaturen nicht. Bei 73 Grad SPD hub crahste es dann aber das ist weit weg von Normalbetrieb.

8400hot.png
 
@Veii

Here is the result.


1705569944720.png



Trtp is on 12, is it okay at 8600? Somewhere I saw maybe you suggested a bit higher for this speed.
Now I reduced the Voltages further (judging by the error 4), test is running, report back later. :)
 
Zuletzt bearbeitet:
Hallo, ich hoffe ihr habt alle einen schönen Tag.

Ich habe mal eine Frage, welche Bios Version ist im Moment die beste für das Encore?

Ich danke im vorraus für die Antwort 🙂

LG Marcus
 
Ich hab die 0080 Testversion drauf. Geht genau wie die 0071 sehr gut.
 
@Veii

Here is the result.


Anhang anzeigen 960783


Trtp is on 12, is it okay at 8600? Somewhere I saw maybe you suggested a bit higher for this speed.
Now I reduced the Voltages further (judging by the error 4), test is running, report back later. :)
Good timezone.
You/we go backwards with this and your last message approach.

Error 6 is bad
Error 0 is worse
Error 2 & 12 is lack of supply and or RTT issues. Generally lack of supply
Error 4 is bad, but not as bad as 6 or 0

Reducing voltage doesnt help against overcurrent crash.
Voltage means nothing alone. Its overcurrent not overvoltage crash.

@X909 equally goes backwards with doing 8+ (7+halt) + 8 + (7+halt) approach for RDRD_SG
You can not expect anything to happen on 15th clock when minimum BurstLength is 16.
I don't understand what you try to do, but its unlikely to work. Hardware behaves like hardware was designed.
If you want to play with fire, reduce tRDWR_SG/DG not tRDRD_ . But even there you will cause (bad) overlapping issues.
Beitrag automatisch zusammengeführt:

Now I reduced the Voltages further (judging by the error 4), test is running, report back later. :)
#4 becomes irrelevant
When #6 (Cpu side or primaries)
and #5 trace dropout cpu side

Exist.
Reports get worse and worse to what i can read.
Keep trying, but i see reports go backwards that way.
Sorry~
Beitrag automatisch zusammengeführt:

Trtp is on 12, is it okay at 8600?
15 is an OK value, but #4 is no timing issue.
It run long before you hit #4.

#4 is a voltage design issue.
Timing issues you shouldn't have, especially with such late reports.

Uncertain on applikable tWR, and tRTP especially goes into this equation.
Dont touch it, unless you have to. Then needs a change on WR too.
Any case, you dont have timing based errors. At least not yesterday.
 
Zuletzt bearbeitet:
Need 1-2 day rest and start again. :)
(y)
Agreed. I will go back to the base 1.56V VDD, maybe try even more.
mm mm
All attempts failed successfully.
Back to old (only #4) issue.
What do you think, should I try to fix ODT?
Can't
Auto calculated for proc side.
Can for Groups, but will mess up all other voltages.
ASUS-HQ knows how to work with groups. There is nothing new i can contribute.

RTTs may be the path to resolve that
But you need more debugging. More data collection, because taken path definitely lead to nowhere.
System would be without issue stable on first report. But DIMMs are not happy.

Rest well :)
Maybe take some time to undust or clean dimm gold fingers.
If you can find it~
 
Maybe take some time to undust or clean dim gold fingers.
If you can find it~
Yes, I will do that. Good idea, I will replace the thermal pads too because I was in a hurry when I installed the covers.
Chiller also here, maybe colder temps will help a little. I don't have time to install it, only in next month, too many projects goind in the work. :)
 
Chiller also here, maybe colder temps will help a little. I don't have time to install it, only in next month, too many projects going in the work. :)
You plan on dailying chiller ? 🤭
Do you need it for a bigger project ?

mm mm
Usually too low voltage to consider charred gold-fingers. Thats 1.7 territory
But i guess it can't harm. :)
 
It will be daily. Only on CPU and RAM, MO-RA will stay on VGA. :)
Its soo overkill , haha
You wouldnt really "need" it
Tho curious to see if you can run 6.4 to 6.5 TVB with such crazy temps 🤭

Side effect could be, that you have to start using PLLs and definitely redo RTTs.
Eh much to do, much to figure out.
It will be fun~

Mind the dew point.
I heard nothing but complains by early research mate who stays on a chiller constantly.
Negative and positive complains. Just much upkeep work.
 
Mind the dew point.
I heard nothing but complaints from early research mate who stays on a chiller constantly.
Negative and positive complaints. Just a lot of upkeep work.
Yeah, due to the dew point I will keep the coolant between 18-20. Actually during the TM5 it's on +2 TVB on the whole time already with the DD block.
It's so overkill, haha
You wouldn't really "need" it
I know, but I buy these as a company. I would never spend my own money on that. :d
 
Hello @Veii

I couldn't resist to try a different approach, so tried SA 1.22V,
15 can be a CPU issue, maybe the last V/F point has too much negative offset. (this is TVB clock)
What do you think?
1705601929202.png


Little step further.
Rising MC_VDD to 1.525V. tRCD WR changed to 50.
1705634735813.png
 

Anhänge

  • 1705600752594.png
    1705600752594.png
    183,7 KB · Aufrufe: 59
Zuletzt bearbeitet:
VDDQ Training off
ODT Training off
RTT_NOM's 80ohm.

Sub 1.4v VDD(2)_CPU.
VDDQ Training: war schon off
ODT Training off: alle drei? (DIMM ODT Training / Read ODT Training / DIMM CA ODT Training)
RTT_NOM's: DQ RTT NOM RD / DQ RTT NOM WR ? falls ja, da hab ich nur clock's zur wahl --> habs in einem älteren post gefunden, ist auch 80clocks nun
Beitrag automatisch zusammengeführt:

IMC VDD dropped ziemlich:
bios setting ist 1,41xV
1705613329196.png
 
Zuletzt bearbeitet:
Hello @Veii

I couldn't resist to try a different approach, so tried SA 1.22V,
15 can be a CPU issue, maybe the last V/F point has too much negative offset. (this is TVB clock)
What do you think?
Anhang anzeigen 960971

Little step further.
Rising MC_VDD to 1.525V. tRCD WR changed to 50.
Anhang anzeigen 961219
Good timezone :)

Higher SA, lower VDDQ delta, higher MC.
Please retest with 1.4v VDDQ_CPU

#15 likely is a CPU issue. Mostly its unstable cores, but it can be everything if alone.
IMC VDD dropped ziemlich:
bios setting ist 1,41xV
1705613329196.png
Ich dachte wir testen TM5 ?
 
Higher SA, lower VDDQ delta, higher MC.
Please retest with 1.4v VDDQ_CPU
I tought the same, so did the test already. :)
Only one single error 4 in 25 cycles.
Higher SA is trouble, lower SA is trouble.
I can do only 1 rounds of VST/VT3 on the second round the VST throw an error always.
No matter what I set. Higher MC doesn't help (tried up to 1.537V), higher TX doesnt't help, tried up to 1.425 (that made it worse actually).

I am thinking on increasing VDD to have some more delta, what do you think?
I know that error 4 is overcurrent / overvoltage, but lower voltages didn't help.
1705664928804.png
 
No matter what I set. Higher MC doesn't help (tried up to 1.537V), higher TX doesnt't help, tried up to 1.425 (that made it worse actually).

I am thinking on increasing VDD to have some more delta, what do you think?
I know that error 4 is overcurrent / overvoltage, but lower voltages didn't help.
Lets try to run CTL0's
VDD2_CPU over 1.5 i dont think it would need.
Likely in the 1.46-1.48 range

Lower SA was ok.
We need to work with RTTs i think.

I see you pushed VDDQ_MEM up
Keep it low.
With more SA delta between CPU & MEM needs to be smaller.
Delta inside mem is an RTT topic.

1.5 VDDQ_MEM
1.4 VDDQ_CPU
1.22 SA

PSU OFF ~ cold boot this,
As delta change needs full retraining.

Y-cruncher will be difficult,
But also variable of board trains strangeness exists
Lets get first TM5 fully error free
Later it may or may not be lack of voltage on CPU Substrates side.
 
Lets try to run CTL0's
VDD2_CPU over 1.5 i dont think it would need.
Likely in the 1.46-1.48 range

Lower SA was ok.
We need to work with RTTs i think.

I see you pushed VDDQ_MEM up
Keep it low.
With more SA delta between CPU & MEM needs to be smaller.
Delta inside mem is an RTT topic.

1.5 VDDQ_MEM
1.4 VDDQ_CPU
1.22 SA

PSU OFF ~ cold boot this,
As delta change needs full retraining.

Y-cruncher will be difficult,
But also variable of board trains strangeness exists
Lets get first TM5 fully error free
Later it may or may not be lack of voltage on CPU Substrates side.
Thank you, let's see. Will report back when It's ready.
 
Ich dachte wir testen TM5 ?
War von mir nur eigentlich als rein interesse halber als frage gemeint mit TM5...
wir testen womit dir es beliebt :giggle:

womit wir am "einfachsten" stabilität erreichen, im endeffekt muss es ja alles abkönnen das system, sei es TM5 / y-Cruncher oder stundenlange zocken...
alles andere ist nicht stabil...

1-2std noch dann wochenende :d
 
War von mir nur eigentlich als rein interesse halber als frage gemeint mit TM5...
wir testen womit dir es beliebt :giggle:

womit wir am "einfachsten" stabilität erreichen, im endeffekt muss es ja alles abkönnen das system, sei es TM5 / y-Cruncher oder stundenlange zocken...
alles andere ist nicht stabil...

1-2std noch dann wochenende :d
Beides errort wenn das andere nicht passt.
Karhu errort ebenso wegen jedem Blödsinn.

TM5 sah bei dir schrecklich aus. Das hat Prio erstmal
Ich bin zwar geleehrt in meinem Gebiet aber dennoch ist das hier über remote.
Leider kein Magier~

Step by step.
Beitrag automatisch zusammengeführt:

Thank you, let's see. Will report back when It's ready.
Sorry for taking so long.
I need help to sepperate those issues.

If we really have to go down the RTT & ODT rabbit hole, because Board can't power dimms correctly.
Especially given how much thermal headroom and support you give it vs what's the more common case.

I dont trust the Board.
But lets see~
At best result is "nothing = no change"

If we have positive change, we have to go down full manual rabbit hole.
 
Zuletzt bearbeitet:
Beides errort wenn das andere nicht passt.
Karhu errort ebenso wegen jedem Blödsinn.

TM5 sah bei dir schrecklich aus. Das hat Prio erstmal
Ich bin zwar geleehrt in meinem Gebiet aber dennoch ist das hier über remote.
Leider kein Magier~

Step by step.
denke du verwechselst mich jetzt nicht mit jemandem anderen, ich hatte nie ein TM5 gepostet... (macht aber nix, bin für jede Hilfe dankbar)
(y) du machst nen super job hier im Forum und das bewundere ich sehr!

Deine Hilfe hat mir schon hierbei zum Erfolg verholfen:
1705678594325.png
Ist zwar "nur" 7600MT's aber selbst daran bin ich bis vor kurzem gescheitert, nun sind es ~90min ohne Fehler 🫶
 
Zuletzt bearbeitet:
Beides errort wenn das andere nicht passt.
Karhu errort ebenso wegen jedem Blödsinn.

TM5 sah bei dir schrecklich aus. Das hat Prio erstmal
Ich bin zwar geleehrt in meinem Gebiet aber dennoch ist das hier über remote.
Leider kein Magier~

Step by step.
Beitrag automatisch zusammengeführt:


Sorry for taking so long.
I need help to sepperate those issues.

If we really have to go down the RTT & ODT rabbit hole, because Board can't power dimms correctly.
Especially given how much thermal headroom and support you give it vs what's the more common case.

I dont trust the Board.
But lets see~
At best result is "nothing = no change"

If we have positive change, we have to go down full manual rabbit hole.
Looks like the low VDDQ is not good for me:
1705678644413.png



Maybe if we want delta I should keep the 1.55V VDDQ and bump the VDD a little from 1.56V. 1.58-1.59V for instance.
But this could be the low VDD2 too.
 
denke du verwechselst mich jetzt nicht mit jemandem anderen, ich hatte nie ein TM5 gepostet... (macht aber nix, bin für jede Hilfe dankbar)
(y) du machst nen super job hier im Forum und das bewundere ich sehr!

Deine Hilfe hat mir schon hierbei zum Erfolg verholfen:
Anhang anzeigen 961344
Ist zwar "nur" 7600MT's aber selbst daran bin ich bis vor kurzem gescheitert, nun sind es ~90min ohne Fehler 🫶
Doch hatte ich :)
Yay !
Looks like the low VDDQ is not good for me:
Anhang anzeigen 961345


Maybe if we want delta I should keep the 1.55V VDDQ and bump the VDD a little from 1.56V. 1.58-1.59V for instance.
But this could be the low VDD2 too.
Those are RTT issues
Yep, powering is f*iretrucked. Indeed.
8600+ is f**iretrucked. Not IMC issue.

Manual mode it is
Let me get home on pc and ill build something for you.

IN-MEM delta scales till 105mV
Without issues.
60mV is guaranteed to work.
100mV should be easy unless powering is f**unctionaly broken.
 
Hardwareluxx setzt keine externen Werbe- und Tracking-Cookies ein. Auf unserer Webseite finden Sie nur noch Cookies nach berechtigtem Interesse (Art. 6 Abs. 1 Satz 1 lit. f DSGVO) oder eigene funktionelle Cookies. Durch die Nutzung unserer Webseite erklären Sie sich damit einverstanden, dass wir diese Cookies setzen. Mehr Informationen und Möglichkeiten zur Einstellung unserer Cookies finden Sie in unserer Datenschutzerklärung.


Zurück
Oben Unten refresh