What should be the DDR5 settings on the AMD B650E motherboard?

[Klemc]

If it's on QVL it makes sense to use the ASUS Tweaked, but the 7800x3d looks like to not give a * about RAM DOCP ON or OFF anyway.

 

[VivaLuatioN]

EXPO 1 loads EXPO speeds and primary timings. EXPO 2 loads EXPO speeds, primary and secondary timings (this is what I'd choose personally), Tweaked is Asus's own tweaked settings. Take your pick.
Oh, and welcome to the forum.

Thanks for welcoming me,
For my build from June till now i use EXPO 1, if i change to EXPO 2 or EXPO Tweaked does it make the different from EXPO 1?
Sorry if i asking this s****d question.....my first build btw....

If it's on QVL it makes sense to use the ASUS Tweaked, but the 7800x3d looks like to not give a * about RAM DOCP ON or OFF anyway.

Which means EXPO Tweaked is the best way to go, do i need to adjust those ram settings like csm, fast boot, MCR, Power Down?

 

[Klemc]

I always read there is ONE setting important to lower, but wait confirmation, bc i can word it bad, tRFC.

 

[3DVCash]

Thanks for welcoming me,
For my build from June till now i use EXPO 1, if i change to EXPO 2 or EXPO Tweaked does it make the different from EXPO 1?
Sorry if i asking this s****d question.....my first build btw....


Which means EXPO Tweaked is the best way to go, do i need to adjust those ram settings like csm, fast boot, MCR, Power Down?

Expo Tweaked is the most aggressive preset on secondary timings. So yes, it is the most performant of the 3. Ofc, it's not guaranteed to be stable, but that's the case with any Expo setting over JEDEC.

MCR saves your memory training settings so it doesn't have to be done at every boot, potentially at the cost of stability. It should be fine with Expo. If pushing a serious mem OC, you probably want to leave that alone though.

 

[AusWolf]

Thanks for welcoming me,
For my build from June till now i use EXPO 1, if i change to EXPO 2 or EXPO Tweaked does it make the different from EXPO 1?
Sorry if i asking this s****d question.....my first build btw....


Which means EXPO Tweaked is the best way to go, do i need to adjust those ram settings like csm, fast boot, MCR, Power Down?

EXPO 2 shouldn't be much different from 1. Only that it loads the secondary timings from your RAM's SPD table, not just the primaries. That's why I would choose that, because it's the spec your RAM is designed to run at. Tweaked means whatever Asus means by it.

CSM off
Fast boot off
MCR on if it works, off if it gives you issues like hangup during boot,
Power Down is needed for MCR, so on.

 

[ir_cow]

EXPO 2 shouldn't be much different from 1. Only that it loads the secondary timings from your RAM's SPD table, not just the primaries. That's why I would choose that, because it's the spec your RAM is designed to run at. Tweaked means whatever Asus means by it.

CSM off
Fast boot off
MCR on if it works, off if it gives you issues like hangup during boot,
Power Down is needed for MCR, so on.

I'm with you on this.

 

[VivaLuatioN]

Expo Tweaked is the most aggressive preset on secondary timings. So yes, it is the most performant of the 3. Ofc, it's not guaranteed to be stable, but that's the case with any Expo setting over JEDEC.

MCR saves your memory training settings so it doesn't have to be done at every boot, potentially at the cost of stability. It should be fine with Expo. If pushing a serious mem OC, you probably want to leave that alone though.

Thanks for fully explaining What's the deal of Expo Tweaked

EXPO 2 shouldn't be much different from 1. Only that it loads the secondary timings from your RAM's SPD table, not just the primaries. That's why I would choose that, because it's the spec your RAM is designed to run at. Tweaked means whatever Asus means by it.

CSM off
Fast boot off
MCR on if it works, off if it gives you issues like hangup during boot,
Power Down is needed for MCR, so on.

Okay i will try EXPO 2 when i back home

 

[VivaLuatioN]

EXPO 2 shouldn't be much different from 1. Only that it loads the secondary timings from your RAM's SPD table, not just the primaries. That's why I would choose that, because it's the spec your RAM is designed to run at. Tweaked means whatever Asus means by it.

CSM off
Fast boot off
MCR on if it works, off if it gives you issues like hangup during boot,
Power Down is needed for MCR, so on.

Reporting back btw, i have enable EXPO II

CSM OFF
Fast Boot OFF
MCR Disabled
Power Down Disabled

I have read other forum regarding as well, do i need to change UCLK DIV1 Mode to 1:1? UCLK=MEMCLK

 

[AusWolf]

Reporting back btw, i have enable EXPO II

CSM OFF
Fast Boot OFF
MCR Disabled
Power Down Disabled

I have read other forum regarding as well, do i need to change UCLK DIV1 Mode to 1:1? UCLK=MEMCLK

View attachment 367561

It looks like it defaults to 1:1 in your config, so happy days.

 

[Sarajiel]

do i need to change UCLK DIV1 Mode to 1:1? UCLK=MEMCLK

For DDR5-6000, most boards default to 1:1 mode nowadays. Some Asus and a lot of ASRock boards switch to 2:1 mode automagically if you set your memory multiplier higher than to 62x or to 64x. Gigabyte doesn't usually change that setting automatically. Not sure about MSI boards.

 

[VivaLuatioN]

It looks like it defaults to 1:1 in your config, so happy days.

Yup it's so worth it to create an account at this forum and ask about it. Really appreciate what member here and you helping me go thru step by step!

For DDR5-6000, most boards default to 1:1 mode nowadays. Some Asus and a lot of ASRock boards switch to 2:1 mode automagically if you set your memory multiplier higher than to 62x or to 64x. Gigabyte doesn't usually change that setting automatically. Not sure about MSI boards.

Learn something new again, but if i'm not mistaken i didn't went thru for UCLK DIV1 bios for that

 

[Orodruin]

What @Toothless is saying is to adjust FCLK speed to the following ratio
MCLK:UCLK:FCLK = 3:3:2 = 3200:3200:2133

Thats FCLK at 1/3 of 6400MT/s
Allegedly this (3:3:2) is the best ratio for AM5

...but I would also try
3200:3200:2166
3200:3200:2200
...if I could get FCLK to be stable at those speeds and see if gaming FPS is improving.

Note:
When trying increasing FCLK speed above 2000 watch out for WHEA system errors.
Usually they come first before complete system instability, like crashes and restarts.

Hello again.
I changed the motherboard. I switched to Asus ROG Crosshair X670E Hero.
I will try these settings you mentioned on the new motherboard now.
Let's see what the result will be.

Congratulations on a Ryzen that does DDR5-6400. Also, nice to see that the board didn't just sneak in the 2:1 mode.


It just changed the refresh interval (tREFI), which is the value with the single most performance gain.

All the other timings are either EXPO/XMP or JEDEC for this kit. I got the 96GB version of a similar G.Skill kit and these values look very familiar to me. That means you still have some headroom for manual tuning if you want to try it. However, unless you are really want to squeeze out the last frame for 240+ fps gaming or something similar, I wouldn't really bother except for lowering the tRFC, by setting it to half its current value. It's a quirk of JEDEC speed bins above 6400MT/s.

I haven't used an Asus AM5 for quite some while now, so I can't really walk you down to the exact menu where you have to make this and other changes, but the search function should reveal it, otherwise look for the AI tweaker section or whatever Asus calls it in your board's BIOS.


You typically don't get WHEAs with Ryzen 7000 on AM5 if the infinite fabric corrects errors via re-transmissions. You only start noticing those corrected errors once your performance starts tanking, any display connected to the iGPU starts flickering, or your audio cuts out/stutters while gaming or running CPU intensive workloads.

Frankly, I don't want to bother with lower timings. Even the EXPO Tweaked setting was more than enough. The latency dropped from 75Ns to 66. Even that is very good for me.

My screen is already 3440x1440 144hz. So 200fps and above is meaningless to me.

EXPO 2 shouldn't be much different from 1. Only that it loads the secondary timings from your RAM's SPD table, not just the primaries. That's why I would choose that, because it's the spec your RAM is designed to run at. Tweaked means whatever Asus means by it.

CSM off
Fast boot off
MCR on if it works, off if it gives you issues like hangup during boot,
Power Down is needed for MCR, so on.

I made my adjustments exactly as you said.

And it works without any stability issues.

 

[Zach_01]

Might also want to check this too

Ryzen 7000 / AM5 Memory guide (DDR5 6000+ stability, slow boots, memory training etc)

This ones going to be short and sweet, fortunately. I've done a LOT of repetitive testing with a 7900x here to replicate why some users have slow boot issues of what they mistakenly call 'memory training', and i did not want this lost hundreds of pages into the Zen Garden clubhouse thread where...

www.techpowerup.com

 

[Orodruin]

Might also want to check this too

Ryzen 7000 / AM5 Memory guide (DDR5 6000+ stability, slow boots, memory training etc)

This ones going to be short and sweet, fortunately. I've done a LOT of repetitive testing with a 7900x here to replicate why some users have slow boot issues of what they mistakenly call 'memory training', and i did not want this lost hundreds of pages into the Zen Garden clubhouse thread where...

www.techpowerup.com

I will definitely try this.

I did a clean install. I will definitely test it with AIDA64 this weekend.
Thank you very much.

 

[Zach_01]

The “issue” with AIDA64 (bandwidth & latency test) is that the results can be all over the place from run to run with same settings because it’s affected by pure CPU performance also. That can be affected by background tasks and/or even ambient PC conditions.

The most reliable way is to lock CPU on a static frequency depending the CPU. Could be 4.0GHz or 4.5GHz for example.
BUT I don’t recommend it for non experienced users because it involves manual static CPU Vcore also and you have to know what you’re doing for your specific CPU.

So you just leaving the CPU on auto (default) boost like it is and try to replicate operating conditions as much as possible. And you run it multiple times with each MCLK/UCLK/FCLK and timing settings.

 

[ir_cow]

@Zach_01 I'm with you on this one. Only reason to lock CPU frequency / static voltage is for benchmark scaling where you need to have the least amount of variables.

The other benefit is knowing that isn't the cause of your crashes . It's happens to me still when I am so sure of something only to realize, the CPU is still a factor. Days of research wasted because of a BIOS clear and not re-enable a static clock.

Just don't set a CPU overclock / PBO while messing with memory for the first time. Unnecessary variable when your dialing in timings and voltages for memory.

Also be prepared to clear CMOS a lot since AMD is a pain to work with on the memory front. Retry and safe-boot features rarely work. So many hard lockups that require a clear. Otherwise it will boot-loop forever.

 

[Orodruin]

The “issue” with AIDA64 (bandwidth & latency test) is that the results can be all over the place from run to run with same settings because it’s affected by pure CPU performance also. That can be affected by background tasks and/or even ambient PC conditions.

The most reliable way is to lock CPU on a static frequency depending the CPU. Could be 4.0GHz or 4.5GHz for example.
BUT I don’t recommend it for non experienced users because it involves manual static CPU Vcore also and you have to know what you’re doing for your specific CPU.

So you just leaving the CPU on auto (default) boost like it is and try to replicate operating conditions as much as possible. And you run it multiple times with each MCLK/UCLK/FCLK and timing settings.

I have never tested by giving a fixed frequency and voltage.

But I will try it as you said. I don't know what the result will be, to be honest.

Are the Cudimm memories that come with Intel special for Intel or can they be used for AMD? What I am most curious about is if the optimum speeds for AMD are 6000-6400mt/s, if we install 8000mt/s memory, we are disrupting the FCLK ratio and the latencies are increasing, right? I saw latencies like 105NS and 110NS in the new Core Ultra review.

@Zach_01 I'm with you on this one. Only reason to lock CPU frequency / static voltage is for benchmark scaling where you need to have the least amount of variables.

The other benefit is knowing that isn't the cause of your crashes . It's happens to me still when I am so sure of something only to realize, the CPU is still a factor. Days of research wasted because of a BIOS clear and not re-enable a static clock.

Just don't set a CPU overclock / PBO while messing with memory for the first time. Unnecessary variable when your dialing in timings and voltages for memory.

Also be prepared to clear CMOS a lot since AMD is a pain to work with on the memory front. Retry and safe-boot features rarely work. So many hard lockups that require a clear. Otherwise it will boot-loop forever.

So by giving a fixed frequency, we can fix the processor's variability and get a study to see if it is stable, right?

Frankly, I do not apply any other overclock other than opening the EXPO setting on the memories. Stability is much more important to me than high speed.

 

[AusWolf]

I have never tested by giving a fixed frequency and voltage.

But I will try it as you said. I don't know what the result will be, to be honest.

Are the Cudimm memories that come with Intel special for Intel or can they be used for AMD? What I am most curious about is if the optimum speeds for AMD are 6000-6400mt/s, if we install 8000mt/s memory, we are disrupting the FCLK ratio and the latencies are increasing, right? I saw latencies like 105NS and 110NS in the new Core Ultra review.


So by giving a fixed frequency, we can fix the processor's variability and get a study to see if it is stable, right?

Frankly, I do not apply any other overclock other than opening the EXPO setting on the memories. Stability is much more important to me than high speed.

Please do not set a fixed frequency and voltage unless you know what you're doing really well.

The point of the post you replied to was that variations are normal, so don't bother with it. AMD engineers know how your CPU should be running, and it's configured to do that, trust me (or trust them). Your manual settings are most definitely just gonna mess things up and give you more headache.


Benchmarking is bullshit anyway, unless you're testing for REAL problems.

 

[Orodruin]

Please do not set a fixed frequency and voltage unless you know what you're doing really well.

The point of the post you replied to was that variations are normal, so don't bother with it. AMD engineers know how your CPU should be running, and it's configured to do that, trust me (or trust them). Your manual settings are most definitely just gonna mess things up and give you more headache.


Benchmarking is bullshit anyway, unless you're testing for REAL problems.

No, I didn't do anything.

I've known how Ryzen processors have dynamic voltage configurations since the 3000 series.
Frankly, I won't do anything for fixed voltage and frequency adjustments. There's nothing to be afraid of

 

[Zach_01]

Yeah the point is to remove CPU boost variable frequency from the equation.
I'm all against permanent static frequency and voltage since Zen 2 (3000).

But here is the concept behind static CPU clock for memory performance testing.

Lets say that my CPU can do safely* a static 4.6GHz with 1.25V

If I want to tune my memory bandwidth and latency as accurately and consistently as possible I should set my CPU to 4.2GHz with 1.1V (Low speed/voltage for 100% safety).
Of course memory performance will not be the best (because of low CPU freq) but at this point I want to test only the differences between memory speed and timings and not ultimate performance.
Once I find what gives me the best results I then return the CPU to its normal auto boosting condition.

Of course this takes time and effort and 99% of users dont do it, including me.

*(assuming that I would not ever run stress tests that bring CPU to its knees, thermally and current (A) wise like P95 small ffts)

 

[ir_cow]

Also sometimes you only see a difference with a static CPu frequency. This would indicate the turbo clock in it's own creates variable large enough that the settings changed on the memory has no real impact on the application in question. Ie the CPU clock speeds fluctuates enough to give a 1-3 (-/+) % in the results.

 

[VivaLuatioN]

Hi everyone here, I have some issue regarding my computer after a month of expo usage

Recently my computer start acting weird like when I shut down my pc on night and the next day morning I turn on, all looks good until I press Firefox or other application suddenly the computer loading and freeze….then after like 25 second the screen turn black and the pc is restart by itself (no bsod or something error jump up)

then the other day I try to turn on my computer in the morning and no press anything letting steam or discord load up itself and suddenly the monitor screen went black and auto restart as well (no bsod or error code show)

does anyone expo doing this issue?

 

[Toothless]

Hi everyone here, I have some issue regarding my computer after a month of expo usage

Recently my computer start acting weird like when I shut down my pc on night and the next day morning I turn on, all looks good until I press Firefox or other application suddenly the computer loading and freeze….then after like 25 second the screen turn black and the pc is restart by itself (no bsod or something error jump up)

then the other day I try to turn on my computer in the morning and no press anything letting steam or discord load up itself and suddenly the monitor screen went black and auto restart as well (no bsod or error code show)

does anyone expo doing this issue?

No system specs listed at all. Should probably start with that.

 

[VivaLuatioN]

No system specs listed at all. Should probably start with that.

ROG Strix B650E-F
7800X3D
16GBx2 Kingston Fury Beast 6000MT Ram (Asus QVL List)
7900XT GPU

 

[Toothless]

DDU and amddrivercleanup tool in safe mode, install GPU drivers offline.