Rocket Lake 32M Pi Tweaks for Noobs

[l0ud_sil3nc3]

14 hours ago, GTI-R said:

The OS is booting. But the pi stops after about 3 loops. io mem is 1.65v I've increased SA to about 1.68v.

Try 1.6v IOM and 1.65v SA, should be more than enough for most cpu's.

[Broken APEX49]

9 hours ago, l0ud_sil3nc3 said:

1.6vIOMと1.65vSAを試してみてください。ほとんどのCPUには十分すぎるはずです。

Thanks for the advice.
But now I can't even boot the BIOS.
OS is also questionable (Windows 7)
I'm trying to avoid clearing CMOS if possible.
MAXMEM is 2500, should I lower it?
IO1.6
I tried SA1.65, but it didn't work.
Maybe the timing is too tight.

[l0ud_sil3nc3]

On 8/23/2021 at 6:23 AM, GTI-R said:

Thanks for the advice.
But now I can't even boot the BIOS.
OS is also questionable (Windows 7)
I'm trying to avoid clearing CMOS if possible.
MAXMEM is 2500, should I lower it?
IO1.6
I tried SA1.65, but it didn't work.
Maybe the timing is too tight.

I would remove some more memory 2.5GB is probably making things much harder to pass. Maybe start at 1-1.5GB and see if you can pass with you current timing set.

Time to clear the CMOS

[Broken APEX49]

 

It's a wonder I can't recreate the setting at this time.
Is it the temperature, or, is it the voltage?

Edited August 31, 2021 by GTI-R

[Broken APEX49]

On 2021/8/13 at 午後6時12分, GtiJason said:

になり得る。1Tを使用してギア1で32mパス3866+を取得するには、1.600v IOMを使用する必要がありました
が、通常、1.350v IOM 1.4000vSAで32gbデュアルランク38661Tギア1をパスできます。
また、SAは最大1.7v
2 x 8gbである必要がある場合があります。シングルランクのDIMMは、Gear 1で実際にヒットまたはミスし、機能するか機能しないかのどちらかです。
Gear 2 5000 + c15などが輝い
ているようです。これらの設定はお勧めしませんが、アンビエントで数回使用したことがあり、これまでのところ問題
はありません。多くの場合、失敗することに注意してください。

 

The voltage you're referring to is iom 1.35V SA 1.4V.
When the first timing is loose, right?
Like the 14-215-15-34-1N?
Super pi results are not good at this timing, right?

[Splave]

Drop a crumb for some nubs, per core OC needed for highest clock. Try to clock only core 02 as high as it will go leave the others at 5ghz etc. usually gains 50-100mhz depending on the cpu vs all cores high. Also makes it fail less after the bench completes and or when opening cpuz. Right click CPUz to select the high core. 

Edited August 31, 2021 by Splave

[Broken APEX49]

8 hours ago, Splave said:

Drop a crumb for some nubs, per core OC needed for highest clock. Try to clock only core 02 as high as it will go leave the others at 5ghz etc. usually gains 50-100mhz depending on the cpu vs all cores high. Also makes it fail less after the bench completes and or when opening cpuz. Right click CPUz to select the high core. 

???

 

[Broken APEX49]

I'm sorry.
I don't understand what you're talking about.

[Splave]

Only oc the core you're running pi on

[Broken APEX49]

Are you saying that could be the reason why the pi is not running?

[Broken APEX49]

Let me ask you, professionals.

If the memory timing is set in the BIOS and the OS boots successfully
Is it possible that the DRAM is corrupted if the pi doesn't run at all?
Normally, I think it's various voltage adjustments, but...
I'm confused because pi won't run at all with the same settings.
I didn't keep the voltage data at the time, which was stupid of me.

I'm lost in the maze now.
 

[Splave]

12 hours ago, GTI-R said:

Let me ask you, professionals.

If the memory timing is set in the BIOS and the OS boots successfully
Is it possible that the DRAM is corrupted if the pi doesn't run at all?
Normally, I think it's various voltage adjustments, but...
I'm confused because pi won't run at all with the same settings.
I didn't keep the voltage data at the time, which was stupid of me.

I'm lost in the maze now.
 

It is much easier to just train and boot memory than it is to do actual work with it. 

[Broken APEX49]

Memory training is also very time consuming.
Unusually so.
And when I choose to reboot, it won't start.
I had no choice but to call up the profile again, but I couldn't train it.
It's a vicious cycle of starting all over again.

[Broken APEX49]

My dual rank challenge is coming to an end.
It was a waste of money.
 

[Broken APEX49]

I've made some adjustments.
I can't finish the pi.
First timing only, 12-12-12-28-1T.
DRAM voltage 1.7V
IO MEM auto (1.45V) 1.3 is impossible.
SA Auto (1.6V) 1.4 is not possible at all.
CPU voltage is auto and Windows7 starts without any problem.
Windows7 starts without any problem, but it cannot run completely.
After a while, Windows7 crashes.
What's wrong with me?
I can't help you.
 

[Broken APEX49]

Even with XMP settings, pi does not complete.
I've never had this problem before.

[Broken APEX49]

IMC may be the cause.
Is it possible to identify the problem with IMC?

Edited September 9, 2021 by GTI-R

[Broken APEX49]

忍耐

Edited September 21, 2021 by GTI-R

[Broken APEX49]

First target clear.

[SparkysAdventure]

1 hour ago, GTI-R said:

First target clear.

Congratulations! Awesome results!

[GtiJason]

On 9/16/2021 at 9:19 PM, GTI-R said:

With the same DRAM timing, start at frequency 3733 and run pi, then start at frequency 3866 and run pi.
However, if I start from the beginning with 3866, π does not run at all.
This situation has been confirmed recently.
Is it my imagination?

32m is challenging, but isn't that why we still love it.
Take this picture below, this is TestMem5 Extreme1@anta777.cfg, HCI MemTest 7.0 and LinX 0.9.11 stable
but I can't pass 32m with these settings. It needs higher DRam V, IO Mem V and SA V to pass

[GtiJason]

Sometimes even daily stable mems fail 32m

[Broken APEX49]

I'm not satisfied.
I feel lucky to have made it to the end.
But it didn't run at all on APEX.
SA has 1.69V in it.

Edited September 21, 2021 by GTI-R

[Broken APEX49]

Record Update
Now if I can just get Windows XP to work properly...

[Broken APEX49]

The combination of bclk and dram frequency and cache is a problem.
Can't run DRAM at 3866