Tools and other software RMClock -> Undervolting X2-5050e issue when changing P-State (1 Viewer)

[joz]

Hi guys,

my first post ever here and not a noob but no pro either with RightMarks RMClock.
For all my system specs please see my profile

I wanted to undervolt this beauty to get even less power usage from the 5050. I've undervolted before on a AMD 3800+, I learned how to work RMClock while doing that little project.
I can easily undervolt my CPU but the problem is that I can't get it stable at all. Whenever I have more then 1 P-State defined I have random freezes in windows.
If I just enable 1 P-state (with say MP 5x, or MP 13x) I can get it to work stable, it crashes while changing P-State.

I have searched on google and came across people with similair issues.
After posting on a different forum someone told me to play with the transition delay (have not touched it yet), might that be worth trying?
You guys have any other idea?

I know this system is stable as a rock on stock vcore. I already figured out the highs and lows that are stable; MP 5x = 0,75vcore, MP 13x = 1,05vcore. If I combine these two I have the random freezing going on, if I just set one of these and check stability all is fine.
I did install AMD CPU driver (I'm on XP) and enabled C&Q in the bios.

Thanks in advance for any tips you guys have for me

 

[pinok]

Hi joz,
this is my first post too
I've a 5050e and it seems to be rock solid, even if I use crystlcpuid to undervolt it.
I followed a guide, but I forgot the url , where it explains how to detect the min multiplier and voltage.
It is a rmclock function, anyway.
Then you can use that settings both in rmclock or cpucrystalid, as I did.
For me, the min voltage suggested for the 5050e cpu is 0.9V and the min multplier is 5x.

With abit m2hd, 1 hd WD green (640gb), 1 dvdrw, 1 keyboard/mouse wireless transmitter and 1 2gb memory stick I have 36w power consumption in idle

 

[joz]

Hey pinok,

I have not tried undervolting with crystlcpuid but although you mean things well, this is no answer.
I know how to work RMClock, or at least I know how to undervolt but the problem is that switching P-states (power states) makes the platform unstable. If I just have 1 p-state (say MP 5x with voltage 0,75, which is orthos stable for 12 hours) selected there's no problem.
If I have more then one the CPU can switch between these states and then it crashes.

 

[pinok]

Hi,
what I was saying is just that RMClock suggested me to use 0.9V as lower value to undervolt, and with this value I have no problem (until now ).
You use 0,75V that is far enough from 0.9V and it is difficult for me to understand that rmclock suggested you such a different value for the same CPU.
And, with that value, you have problems when switching p-states.

Could it be that your undervolt value 0.75 is good for using only a p-state, whilst 0.9 is good both for a p-state and switching more p-states?

Even if you know how to undervolt with rmclock, you could look at this guide A Guide to Undervolting your Processor - Clunk's Forums.
At step 3 talks about "Performance on demand" and "p-states".

Bye !

 

[joz]

Thanks pinok for the guide, it was helpful!

However it did not really solve my problem but I think I'm getting towards a stable setup now! AWESOME!
For anyone else experiencing this, setting the bios setting "spread spectrum" to disabled will increase stability (as well as when overclocking). Besides that my particalur CPU is actually based upon a mobile AMD X-2 variant. So in RMClock I had to set the "change p-state mehtod" to "single transition" instead of the default "multiple transition".
It's a little premature to say but it's way more stable then it was with my previous attempts.

And although Pinok has a point, 0,75 volt is stable on multiplier of 5x when stressing that for 8 + hours. However it is not when having more p-states enabled. I got it running stable now with 3 p-states enabled (5x, 9x, 13x) and got the lowest @ 0,85 volt (the rest respectively 0,975 and 1,1 volt), which is a pretty good result I would say, hope it's stable though! Still some testing needed to be done to be certain.

 

[etheesdad]

joz - just wondering how you got on with this... did you manage a stable underclock? What cpu speed did you end up with?

 

[joz]

Hi etheesdad,

I have a stable undervolt but it took multiple non-default settings to get there;
- Disabling spread spectrum
- Changing the "transition state" to "single" instead of "multi step"

I run it now with 3 p-states because that was already a pain in the ass to figure out getting stable. I noticed that half multipliers are kinda doomed to be instable on my setup.
Imo you undervolt not underclock with an HTPC (or you bought the wrong cpu in the first place). I just got the standard speeds and went down in voltage, not in clockspeed. I want to game a little on the machine too.

The total power savings is not that much though like ~5w in idle and 10 in load. However RMClock makes the switching less jumpy when for example inside MP idling screen. RMClock keeps it in low p-state as where default windows c&q will jump all over.