X6 1100T overclock help

[icecosta]

Hello guys i´m new in the forum and i´m looking for some help overclocking this Cpu

My Hardware is Xfx pro 650w, Asus M5A97 Pro, x6 1100T , 4Gb gskill 1600 cl9, Agility3 120gb, hd7870, Haf 932, Thermalright true spirit 140 in Push/pull

Ok so i´m trying to archive stability at 4.0Ghz.

I had the cpu overclocked at 3.8Ghz with prime 95 stability for 12H with current settings

Vcore offset mode 1.38V
Cpu LLC medium
Spread spectrum disabled
Cpu current capacity 110%
Cpu power phase control standard

Also had the NB/CPU frequency overclocked at 2400Mhz
With an increase in CPU/NB voltage offset mode to 1.18V

Things appeat to be stable
Temps were my biggest problem because My Cpu temp appeard at full load 51ºc in prime 95
And the core temps appeard at 38ºc

So in other forums people suggested that in this case the temperature that matter would be Core temp.

If so i still had a lot off heat room...

Your opinions guys Core temp or Cpu temp?

Well any way either core temp or cpu tem at this speed i was still under operational specs for the chip so i decided to go up to 4.0ghz

That´s when problems started no matter how much vcore it would fail in tests

So i decided to reset Cpu/nb to stock.

And currently trying to stabilize at 4.0Ghz.

Current settings in bios are:

Cpu vcore offset mode 1.45
LLc cpu High
Spread spectrum disabled
Cpu current capacity 110%
Cpu power phase control standard

All the other specs are at stock

Any thing i´m doing wrong?

What about the temperature sensors? Any idea on what to look Cpu or Core temp?

according to Alex Cromwell, Senior Technology Director, Advanced Micro Devices...

As long as your core temperature has not exceeded the high side of the 60 degree mark for extended periods of time you should be ok. 62 degrees holds a generous safety net to begin with

 

[RGone]

Generally speaking to suggest changes it is easier if we know certain things. Of course beginning with a setting that is 'known' to pass P95 Blend mode makes good sense, because a failed P95 Blend test is not going to give a baseline of a configuration that 'does' work.

CPU Tab in CPUz from CPUID com

Memory Tab in CPUz from CPUID com

SPD Tab in CPUz from CPUID com

And this is screen capture of HWMonitor (free version) from CPUID com
HWMonitor has been scrolled enough and large enough to show Min/Max of Voltages and includes the CPU Core Temps which is nOw called "Package" Temps fully visible.

This capture is made of HWMonitor after it has been open on the desktop logging Min/Max temps and voltages while Prime 95 was running Blend Mode test on all cores for at least 20 mins and then the capture of HWMonitor was made and it shows the Min/Max temps and voltages before P95 Blend was started and while running P95 Blend mode and gives much greater insight into how the system is performing without guessing.

In order to attach screenshots of INDIVIDUAL images as suggested, first crop and capture the images with Snipping Tool found in Windows Accessories or equivalent. Then click on Go Advanced, a button at the bottom of every new post window. Then click on the little paperclip tool at the top of the Advanced post window when it opens. Clicking on the paperclip tool brings up the file browser/upload tool and the rest is fairly obvious.

Suggested CPU/Socket temp not to exceed 70c and "Package" temp not to exceed 60c during P95 testing
RGone...

 

[icecosta]

Thanks for the reply

As suggested above

Hwmonitor

image hosting services

Cpuz

image hosting over 10mb

image share

screen capture windows 7

Prime 95

33 minutes



screen shot apple

Now this OC has been stable with the current Vcore for the past week, he was also tested 12H running prime 95 in blend mode.

But right now i´ve altered the Cpu/nb frequency to 2600Mhz and changed voltage to 1.20v

also changed HT link to 2400Mhz - just discovered that it actually impacts on performance, for what i read about it i was convinced that it had no impact but the truth is that it appears to cause more performance at 2400Mhz then at 2000Mhz at least winrar benchmark in here says so.
Ht voltage however was left at stock.

Cpu LLC is beeing kept at medium
Cpu/nb LLC is at regular
Spread spectrum disabled
Cpu current capacity 110%
Cpu/nb current capacity 110%
Cpu power phase control optimized
Pcie frequency is at 100Mhz

all other voltages and options are beeing kept at stock levels

The temps appear within regular operational levels

And i am convinced that it can still go further how ever he seems to need a huge amount of vcore increase from 3.8Ghz to 4.0Ghz

Any sugestions?
Perhaps this is the sweetspot i don´t know?!

 

[storm-chaser]

Thanks for the reply

Cpuz


But right now i´ve altered the Cpu/nb frequency to 2600Mhz and changed voltage to 1.20v

also changed HT link to 2400Mhz - just discovered that it actually impacts on performance, for what i read about it i was convinced that it had no impact but the truth is that it appears to cause more performance at 2400Mhz then at 2000Mhz at least winrar benchmark in here says so.
Ht voltage however was left at stock.

Cpu LLC is beeing kept at medium
Cpu/nb LLC is at regular
Spread spectrum disabled
Cpu current capacity 110%
Cpu/nb current capacity 110%
Cpu power phase control optimized
Pcie frequency is at 100Mhz

all other voltages and options are beeing kept at stock levels

The temps appear within regular operational levels

And i am convinced that it can still go further how ever he seems to need a huge amount of vcore increase from 3.8Ghz to 4.0Ghz

Any sugestions?
Perhaps this is the sweetspot i don´t know?!

Your core voltage is actually a little low which may be why prime is failing. With that high quality air cooler you should have no problem running 1.4 or 1.5v which is what most AMD hexacores need to run stable at 3.9-4.0ghz.

When you applied thermo paste did you just put a bb sized amount on and properly torque the cooler down? Do you have any other fans in the case?

Sometimes the HT link will give you better benchmarks when higher but the general consensus is to leave it as close to stock, if possible because you are just adding more heat to the mix. I've run it at 2.9ghz for a while without any ill effects but the system appears to be stable with it right around 2200 and other benchmarks report higher scores on that setting.

Its good to see you are modifying the CPU-NB because that will really unleash memory performance, which I see you have clocked properly at 1600mhz. You may be able to safely reach 3000 MHz on the NB. You can go to about 1.35v safe on your cpu-nb voltage.

I would say try for higher if you wish, just use HWMonitor to continue tracking your core temps (which was 71* in your last test)

here to help with any questions and good luck.

 

[icecosta]

Hello thanks for your reply.

Prime 95 is not failing at this speed with this vcore, the oc was stable 12H under prime95.

Prime95 fails when i raise the multiplier to 20x and i tried raising vcore up to 1,46v and still continued to fail, and that´s the weird thing he is stable with 3.8Ghz with a low vcore and above that he fails even with 1,46v.

About the temps? they don´t seem to be the problem right?

My case is a Coolermaste Haf932 with the stock fan´s 3x230mm 1 in the lateral intake 1 in the top outtake 1 in the front intake and the 140mm in the back outtake

 

[trents]

Hello thanks for your reply.

Prime 95 is not failing at this speed with this vcore, the oc was stable 12H under prime95.

Prime95 fails when i raise the multiplier to 20x and i tried raising vcore up to 1,46v and still continued to fail, and that´s the weird thing he is stable with 3.8Ghz with a low vcore and above that he fails even with 1,46v.

About the temps? they don´t seem to be the problem right?

My case is a Coolermaste Haf932 with the stock fan´s 3x230mm 1 in the lateral intake 1 in the top outtake 1 in the front intake and the 140mm in the back outtake

No, this is a normal experience and you are hitting "the wall" as we say where the scaling of the stable frequency increase vs. voltage increase begins to change dramatically. That is to say, the first stage of the overclock is characterized by large increases in CPU freqency be kept stable by relatively small increases in core voltage. When you approach the wall you find that to get one more bump of frequency increase requires a much larger vcore increase to make stable.

 

[icecosta]

Hello again, decided to give a quick attempt and raised the multiplier to 19.5x and the vcore to 1,45v with vdroop in full load 1,40V

It was stable enough to pass 10 rounds of Intel burn test but the temps went UP


open source screen capture

screencapture

Did another 10 rounds of IBT with the multiplier at x20 tried to raise vcore up to 1,4750v but it always failed
After that tried with vcore at 1,45 but increased LLC to high and Cpu current power to 20%

Passed the test but the temps specially socket temp or cpu temp went all the way up

image

What do you think? worth trying to get more speed or stick with 3.8Ghz?

 

[trents]

Your core temps are fine but your socket temps are high. How's your wire/cable management inside your case? Are you getting air flow blockage around the socket area by rat's nests of cables and wires? Also, what about fans and ventilation? You have a quality case but how many intake and how many exhaust fans are you employing to move air through the case? I'll attach a pic of what good case ventilation looks like. Also, you might consider positioning a spot fan to blow on the socket/VRM area of the motherboard.

 

[trents]

Would you please create a "Sig" so that your system info appears in every post you make? Go to Quick Links at the top of the page and click on Edit Signature. You may not have enough posts to access this feature yet. Look for a gold star to appear under your name and then you can create a Sig.

Another thing: Please crop those attachments to eliminate all the white space around the relevant image. Snipping Tool in Windows Accessories is great for cropping and capturing images.

 

[ATMINSIDE]

What's your ambient temperature?

 

[icecosta]

Hello

This is the diagram of airflow of the HAF 932 mine it´s exactly the same

Ohhh in the diagram does´n show a lateral 230mm fan also doing intake





The cable managent is in order no cables interfering with the airflow, and i´ve always had this problem with the socket temp being much higher then core temp.

Ambient temperature is around 24ºC

The only thing that might be stoping the airflow near the socket is the Ram modules...

 

[ATMINSIDE]

Your HWMonitor is reading too low for core temps.

 

[trents]

icecosta, the two top fans in the pic should be extracting air not pushing it in. That is creating a dead zone where air will not be moving across motherboard components. If your setup is like that, you need to turn the two top fans around so that they are acting as exhaust.

 

[icecosta]

Yes i agree but all the software i use since Hwmonitor to overdrive coretemp etc etc all show the same readings and it was always like this with this cpu and motherboard ...

Anyone had a problem like this?

Because i see people reporting similiar situations over the net but i´m not quiet certain
If it´s a problem or if it´s just like that

 

[icecosta]

icecosta, the two top fans in the pic should be extracting air not pushing it in. That is creating a dead zone where air will not be moving across motherboard components. If your setup is like that, you need to turn the two top fans around so that they are acting as exhaust.

There´s only one top fan 230mm and she is acting as exhaust - Outtake
The fans that are making intake are the lateral fan and the front fan the top and back are outtake.

Just noticed the diagram in the picture above is not right

Top Fan 230MM outtake
Front Fan 230MM Intake
Lateral Fan 230MM Intake
Back fan 140MM outtake

Any way have been running Boinc with the OC for the past couple of hours and Socket Max temp was 52ªc Core max temp 39ºC Still a huge gap between the Two should i´ve worried?

 

[ATMINSIDE]

I'm simply noting that the monitoring software reads lower than your ambient temperature, which is physically impossible without extreme cooling (LN2, DIce, phase change, etc).

Its possible that your max temps are offset by this also.

 

[trents]

I'm simply noting that the monitoring software reads lower than your ambient temperature, which is physically impossible without extreme cooling (LN2, DIce, phase change, etc).

Its possible that your max temps are offset by this also.

True, but there is also the fact that temp sensors are typically not accurate at the lower end of the scale anyway because they are calibrated to be accurate at the higher end, not to say they will necessarily be accurate at the high end either but that is how the calibration is skewed.

 

[ATMINSIDE]

True, but there is also the fact that temp sensors are typically not accurate at the lower end of the scale anyway because they are calibrated to be accurate at the higher end, not to say they will necessarily be accurate at the high end either but that is how the calibration is skewed.

Makes sense, I just take that into consideration when doing any oc/bench work on my 1100T because I really don't want to replace it. My rule of thumb is usually about 5C for mis-calibration if it is reading low.

 

[icecosta]

Well i Found this in other forum

According to Alex Cromwell
Senior Technology Director
Advanced Micro Devices


""Paul

Sorry I didn't get back to you sooner, I just recently had your email forwarded to my department.

Concerning your question regarding the temperatures with your processor. (1090) the maximum temperature threshold is 62 Celsius which set for the internal die (core) temperature of the chip. The core temperatures have an equational offset to determine temperature which equalizes at about 45 Celsius thus giving you more accurate readings at peak temperatures. The hindrance in this is the sub ambient idle temperature readings you speak of.

The silicon and adhesives used in manufacturing these processors has a peak temperature rating of 97+ Celsius before any form of degradation will take place. The processor also has a thermal shut off safe guard in place that shuts the processor down at 90 Celsius.

The Cpu temperature is read form a sensor embedded within the socket of your motherboard causing about a 7-10 Celsius variance form the actual Cpu temperature, which may be what you are reading about on the net.

You can use an application called AMD overdrive, that will allow you to monitor your temperatures accurately.

As long as your core temperature has not exceeded the high side of the 60 degree mark for extended periods of time you should be ok. 62 degrees holds a generous safety net to begin with.

I hope I was able to answer your questions, If you have any more inquiries don't hesitate to contact us.


Thank You

Alex Cromwell
Senior Technology Director
Advanced Micro Devices
Fort Collins, Colorado
2950 East Harmony Road
Suite 300
Fort Collins, CO
80528-9558 ""

So i take it that is safe to assume that what matters in this case is core temp, what i find weard is my Socket temp beeing that much higher than core temp makes me wonder if it´s a sensor malfunction

 

[RGone]

True, but there is also the fact that temp sensors are typically not accurate at the lower end of the scale anyway because they are calibrated to be accurate at the higher end, not to say they will necessarily be accurate at the high end either but that is how the calibration is skewed.

However as AMD has stated now a year ago when they had reps in the forums by way of emails responding; the lower temp is of absolutely NO concern to them and they only consider a temp for core as it gets to about 45c. So they do not even speculate lower temps to be even close. AMD only has the processor accurately tallying temps as they approach temps that influence the internal shutdown temps.

That is why I never ever pay attention to the minimum temp UNless it is 35c or greater since that on later cpus is indicative of a heatsink issue generally; otherwise the minimum temp is going to show something less than environmental temps.