Unable to control more than 1 processor

[GreyFalcon]

I have a dual processor (2 cpu) system with dual cores. I'm running TThrottle v1.90. It doesn't see both processors, only the 2 cores on the second processor. As a result, when it goes to throttle the CPU back, it only trottles down the second processor and ignores the first one and both its cores. It reads the GPU perfectly however. 

I already know the system is a bit quirky with the temps, as CPUid is already working with me to determine why their monitoring software is misreading the winbond chip and the chipset as different than they are... could this be something along the same lines?

System:

• Motherboard: Asus K8N-DRE
• Graphics card: eVGA 8400GS
• Processors: 1.) AMD Opteron 275 OSP275
                   2.) AMD Opteron 275 OSP275
• Memory: 4GB DDR (divided into 2GB per CPU
• Windows 7 x64

If anything more is needed, I can give it.

Grey

[fred]

I have a dual processor (2 cpu) system with dual cores. I'm running TThrottle v1.90. It doesn't see both processors, only the 2 cores on the second processor. As a result, when it goes to throttle the CPU back, it only trottles down the second processor and ignores the first one and both its cores. It reads the GPU perfectly however.  

I already know the system is a bit quirky with the temps, as CPUid is already working with me to determine why their monitoring software is misreading the winbond chip and the chipset as different than they are... could this be something along the same lines?

System:

• Motherboard: Asus K8N-DRE
• Graphics card: eVGA 8400GS
• Processors: 1.) AMD Opteron 275 OSP275
                  2.) AMD Opteron 275 OSP275
• Memory: 4GB DDR (divided into 2GB per CPU
• Windows 7 x64

If anything more is needed, I can give it.

Grey

What I need is the text from the logging tab.
That looks like this:

Core Temperature: 70 °C, Raw Data: 881E0000
88180000,88180000,881b0000,881b0000,881b0000,881b0000,881e0000,881e0000,
881d0000,881d0000,881d0000,881d0000,881d0000,881d0000,881d0000,881d0000,
0123456777777777
This Processor has [8] 4 Cores with a temperature sensor.

[GreyFalcon]

Here you go.

Code Select Expand

07 August 2010 - 13:39:03 Driver installed properly. Driver Version: 2.0

Program version: 1.90 64Bit
Microsoft  (build 7600)

Language: User: 1033 ENU ,System: 1033 ENU [/L: 1033]

nvidia: found 1 logical devices
nvidia: found 1 physical devices
nvidia: Temperature 47 °C, max Temperature 127 °C

nvidia: GeForce 8400 GS

Vendor ID: AuthenticAMD
Vendor: AMD
HighestIntegerValue: 00000001 - Processor Signature: 00020F12
Misc. info: 02020800
Feature Flags1 00000001
Feature Flags2 178BFBFF

Processor: Dual Core AMD Opteron(tm) Processor 275                       
Processor: Family: Fh, Model: 21, Stepping: 2
Processor: Max Case Temperature: 70.0 °C, Max Power: 0.0 W

Core Temperature: 51 °C, Raw Data: 10640920
10650920,10640920,10640920,10640920,10640920,10630920,10630920,10630920,
10630920,10630920,10630920,10630920,10630920,10640920,10640920,10640920,
0123333333333333
This Processor has (4) 1 Cores with a temperature sensor.

You can help by reading www.efmer.eu/boinc/faq.html How can I help!
Select the send EMail button,or copy everything in this logging window and mail it to me!
boinc@efmer.eu. We use this information to improve this product.

07 August 2010 - 13:39:13 Number of matching Programs (Processes): 0


[fred]

Here you go.

Code Select Expand

07 August 2010 - 13:39:03 Driver installed properly. Driver Version: 2.0

Program version: 1.90 64Bit
Microsoft  (build 7600)

Language: User: 1033 ENU ,System: 1033 ENU [/L: 1033]

nvidia: found 1 logical devices
nvidia: found 1 physical devices
nvidia: Temperature 47 °C, max Temperature 127 °C

nvidia: GeForce 8400 GS

Vendor ID: AuthenticAMD
Vendor: AMD
HighestIntegerValue: 00000001 - Processor Signature: 00020F12
Misc. info: 02020800
Feature Flags1 00000001
Feature Flags2 178BFBFF

Processor: Dual Core AMD Opteron(tm) Processor 275                       
Processor: Family: Fh, Model: 21, Stepping: 2
Processor: Max Case Temperature: 70.0 °C, Max Power: 0.0 W

Core Temperature: 51 °C, Raw Data: 10640920
10650920,10640920,10640920,10640920,10640920,10630920,10630920,10630920,
10630920,10630920,10630920,10630920,10630920,10640920,10640920,10640920,
0123333333333333
This Processor has (4) 1 Cores with a temperature sensor.

You can help by reading www.efmer.eu/boinc/faq.html How can I help!
Select the send EMail button,or copy everything in this logging window and mail it to me!
boinc@efmer.eu. We use this information to improve this product.

07 August 2010 - 13:39:13 Number of matching Programs (Processes): 0


You should see 2 temperatures for the cores.

[GreyFalcon]

There are 2 temperatures. Is it only seeing 1 temperature per processor or only seeing the temperature on the 2 cores in the second as these temps match the ones from CPUid for the second CPU core temperatures. 

The other problem and most concerned one is the throttle is only clamping on the second processor. When it engages and even drops to 5% throttle, the second processor does drop down and hold right near it but the first processor has both cores still running at full power.

[fred]

There are 2 temperatures. Is it only seeing 1 temperature per processor or only seeing the temperature on the 2 cores in the second as these temps match the ones from CPUid for the second CPU core temperatures. 

The other problem and most concerned one is the throttle is only clamping on the second processor. When it engages and even drops to 5% throttle, the second processor does drop down and hold right near it but the first processor has both cores still running at full power.

The AMD processor has a central temperature sensor, so there is "only" one temperature / processor.
The throttling problem is unrelated, a drop to 5% should throttle things down. The throttle is unaware of any CPU, just aware of running programs.
So it may be that the first CPU is doing something different, unrelated to the matching processes. Just one core working on something else may warm it up.
So try finding what is running on the first processor that warms it up.

[GreyFalcon]

Understood about other programs may be the cause for the high use of the processor.

Are you certain about the 1 core reading? From what I've been reading this morning, there is 1 tcase sensor which is the temp on the die and 1 sensor on each core of the chip. There was also a lot of talk of the K8 core readings being as much as 10c off plus or minus from where they should be.

[fred]

Understood about other programs may be the cause for the high use of the processor.

Are you certain about the 1 core reading? From what I've been reading this morning, there is 1 tcase sensor which is the temp on the die and 1 sensor on each core of the chip. There was also a lot of talk of the K8 core readings being as much as 10c off plus or minus from where they should be.

I try to read them all, and over a 1000+ samples the temperatures stay the same.
And AMD is known to have a lot of defect or unreliable die temperature sensors.
The sensors are never meant to read the temperatures accurately, just to regulate the CPU so it stays in the safe zone.
But that's just what TThrottle does, regulate.

The best way to be sure is check the temps with a case sensor, the temperatures read by TThrottle should be about 5C higher.

[GreyFalcon]

Ahhh ok. That makes a lot of sense.

When testing the 'burn in' with Prime95, it did seem to throttle back on all cores (was the only major program running). So it looks like it was some rouge program tucked away some place that has been throwing off the regulation.

Would using the General Rule method over ride a rouge program and throttle everything back? (trying to keep my system alive until new heatsinks arrive next week. Was sold the wrong ones and they do not keep the CPUs cool enough when loaded...)

[fred]

Ahhh ok. That makes a lot of sense.

When testing the 'burn in' with Prime95, it did seem to throttle back on all cores (was the only major program running). So it looks like it was some rouge program tucked away some place that has been throwing off the regulation.

Would using the General Rule method over ride a rouge program and throttle everything back? (trying to keep my system alive until new heatsinks arrive next week. Was sold the wrong ones and they do not keep the CPUs cool enough when loaded...)

There is always some program that fires up unexpectedly  .
Adding the throttle rule should do that.
But don't use this rule as a regulator, it takes up a lot of resources, so set this temperature, to say 10C above the regulator temp.

[GreyFalcon]

Done. Seems to be working well. Base sat at 55c. I sat the trip in the general rule at 60c and saw that it often got to 67c before it was taken back under control so works with safe limits of the chip. Curious about one thing, your program auto configs the Tjunction at 70c when AMD rates the max temp to 69c. (In the last 2 days learned a lot about CPU chips and thermal reading definitions, most this morning! lol) Think it is better to go by what the chip says it is rated for or the company that made it? 

I am simply amazed with how many programs misread this board's sensor chips and thus the CPUs as well... Many only want to look at 1 processor and not both. Of all of the ones out there, your program as well as CoreTemp are the only 2 that accuratly see the system. 

When I initially sat this system up (a week ago) and started my favorite program Everest, it told me all was fine and the cores were all around 35c average and 40c under load... When I loaded CPUid's HWMonitor to work with a gadget I like and made the discovery I was averaging closer to 70c (and the max thermal is 69c) I was petrified! I realized the heatsinks I was sold for the dual opterons with dual cores was not the right ones... (was perfect if they weren't dual core).

I've learned after that HWMeter was misreading the identity of the winbond chip as well as the chipset of the board thus was way off on its figures. (they are going back and forth with me trying to fix the issue with their software) In the mean time, I did a internet search for ANY programs to regulate power and thus lower the temp until the new heatsinks arrive. Nothing... not a single program was out there that saw this board correctly and would attempt to lower voltages or multipliers. When I came across your program I was skeptical at first  but when I saw it hit the magic 55c mark and start slowly throttling back on the 1 processor I was very much relieved that I would still have a usable computer until the new heatsinks arrived.


Even though I would much prefer the lowering of voltage and multipliers (these opterons are the scalable OSP class) this one is a system saver! Great work!