Phenom II X2 555 BE unlock strange behavior

[gheorghe]

Hy,

I've searched a way to solve my particularly problem but didn't find anything so far, I hope to find a solution here.

This is my problem:

-I've unlocked my 555 in Phenom II X4 B55 with MOBO Asus M4a79 Deluxe
-all settings are default in BIOS, I only activate the ACC, unleash mode options for unlocking
-the OS is Win 7 Pro 64 bit, is booting and I don't receive any BSOD
-I run the CPI & Memory stability tests immediately after unlocking (Prime 95, OCCT: CPU, S&M, Memtest86+...) no problem found, temps are in idle ~38 C and full load max 57 C

=>the problem appears when I start browsing, I receive pink, green stripes as the content is not well decoded on portion of the page displayed, I tried to increase the voltages of NB CPU VID but there is no change.

I observed that after changing in Task Manager the "Set Affinity" to core 1,2,3 without core 0 the problem disappears, I also saw that in OCCT is failing ONLY the test for power supply (after 2-3 min). In S&M I observed also that tree cores (1,2,3) are at 100% when testing power supply but core 0 oscillate between 78% - 83%, in the core specific tests the load reaches 100% without no problem.

Another problem I received was when I try to instal my OS with CPU unlocked, the error received was that the hardware is not supported.

I read that the reason of locking the cores could be a faulty core but this problem could be from my PSU (MS-Tech VERTIGO series 620 w 80+) or from my ACC option or simply I'm not lucky?

Thank you!

 

[gheorghe]

THX for the quick reply, so from what I understand, first I have to test in X2 and see what are the differences between generic CPU temp and temp/ core?

I test it in dual core but not for 2 hours, in Prime95 I think it was max 1 hour (but with no problem) ..because I first focus on quad core config

I also run the stability test from AMD overdrive tool in dual core and quad core but still no problem detected.

So first steps that I should do (correct me if wrong):

1. Test the stability in x2 for 2-3 hours in Prime95 blend test? (I noticed that S&M raised the generic temp more with 3-4 degrees than Prime95, OCCT...what tool you advice me to use?)

2.Find the correlation between the generic temp and temp per core. (from what I've tested the generic temp is higher the temp per core).

3. If I receive no error then... ?

PS: I couldn't instal the OS in quad config, I was forced to use the original config (x2) to complete the installation. I'm thinking this was because AMD has no Phenom II X4 B55 product on the market (has only the X2 B55) and the Win7 has no drivers for it (but after unlocking when I first started win7 I saw the drivers for X 4 B55 were installed...).

 

[gheorghe]

Ups...srry I don't know what happened but your post disappeared

 

[RGone]

No idea where it went either. Just one of those web things I guess. Not sure what I wrote now but you seem now headed in the correct direction for two core testing and getting the difference between CPU temp that should hold that basic relation after the unlock to give a guesstimation of the CORE TEMP when unlocked.

 

[gheorghe]

The lost reply...

In my mind the question that keeps coming to mind is does everything work as it should if the cores are not "unlocked"?

I notice you say that there were errors when trying to install the O/S with the cpu in "unlocked" mode. I think it would have served better to know for real if the cpu and system works well with the cpu as AMD intended which is as dual core processor.

After the system has been stability tested in dual core mode for at least 2 hours of Prime 95 'blend' mode and temps of the CPU CORES do not exceed 55C, then would be the time to investigate if you have a quad core in hiding.

When the cores are unlocked; be they good or bad cores, you will LOSE the ability to monitor the CORE TEMPS since the cpu no longer outputs CORE TEMPS in unlocked mode so you are only able to see a socket temp called CPU TEMP and is not the most important temp of that cpu.

So you really need to determine the stability of the system before unlocking and get a correlation between the always there cpu temp and the needed CORE TEMP that is not visible when running in unlocked mode.

Ok I will update the post with the result after testing.

 

[trents]

This has gotten quite confusing. Georghe, I think we need to start over.

First, a little clarity here. To determine if the CPU is stable as a four core, it would be necessary to run Prime95 (I prefer the "blend" test as the best all around stress test) for at least two hours and pass the test with all four cores. Not passing the test means either the computer locks up, throws a blue screen of death, Prime95 stops running because of "an unexpected error" or one or more of the Prime95 worker cores "drops out" while the others keeps going. When a worker core drops out you will get an error in that one window telling you what happened.

To determine max temps it is not necessary to run Prime95 for two hours. Twenty minutes will give you a pretty good idea of what max temps will be during a longer run, within about 2c.

The CPU (socket) temp in HWMonitor is often labled "TMPIN0", or "TMPIN1", or "TMPIN2". In other words it may be labeled CPU temp or it may get a generic "TMPINx" name. We could help you pick it out from that section if you will attach a pic of the HWMonitor interface after running a 20 minute stress test.

 

[gheorghe]

And the tests results...

First in dual core config:

temperatures :
voltages:

Quad core config:

the problem:

temp after 30 min:

temps after 7 hours in Prime95:

voltages:

the ONLY failed test OCCT: power supply:

All settings are on default. If you need more info please tell me :d
And the Prime95 result, the only weird things that I observed are the differences between number of tests finished by every core (but I think the explanation is that I watched online tv series after deselecting core0 from task manager/affinity) and the frequency on which I see the core FSB at the faild power supply (it seem that the current clock is 4271 Mhz (16x267) but I didn't modify the setting, maybe it's a bug) :

 

[trents]

If it will run Prime95 for 7 hours with all the core workers hanging in there that's all you need to know. Waht is "Open HWMOnitor"?

 

[RGone]

I like it because is also shows cpu speeds.


The Open Hardware Monitor is a free open source software that monitors temperature sensors, fan speeds, voltages, load and clock speeds of a computer.

 

[Frakk]

I like it because is also shows cpu speeds.


The Open Hardware Monitor is a free open source software that monitors temperature sensors, fan speeds, voltages, load and clock speeds of a computer.

Yeah, i like that, gives more info than CPUID's HWMonitor.. i think i will use that from now on

 

[gheorghe]

So the CPU it's ok, temps, voltages? What about the browsing problem? What's the root cause... how can I solve this?

 

[trents]

What browsing problem? What are we supposed to be seeing from those pics? The res on the pics is so small its hard to tell. Are those video artifacts?

 

[Frakk]

What browsing problem? What are we supposed to be seeing from those pics?

There are Graphical glitches all over the screen, he also has PSU fail warnings in OCCT.

 

[trents]

gheorghe, please put your system info in your "Sig". You can refer to others' Sig if you want to know what kind of stuff should be included there. Go to Quick Links at the top of the page and click. Then click on Edit Signature.

 

[gheorghe]

I cannot upload a better picture...I have stripes all over the screen and in the picture with the youtube video I see horizontal stripes instead of the video. This happens only when core 0 is used.

 

[trents]

There are Graphical glitches all over the screen, he also has PSU fail warnings in OCCT.

That's why it would be helpful to know about his hardware in the Sig.

 

[trents]

I cannot upload a better picture...I have stripes all over the screen and in the picture with the youtube video I see horizontal stripes instead of the video. This happens only when core 0 is used.

Core 0 is not optional is it? Does this happen as a dual core with core 0 used?

 

[gheorghe]

Done.
I don't have this problem when dual core config is used, and the core 0 from quad config I think is different from the one in dual core.

 

[Frakk]

That's why it would be helpful to know about his hardware in the Sig.

Yeah i know, its your show, i'm just pointing you towards the problem

 

[trents]

gheorghe, try increasing the CPU core voltage a little as a quad core and see if the video issues disappear.