[trents]

No, locking and unlocking the cores and then locking them again will not create instability.

If the system passes two hours of Prime95 blend as a six core at 15x I would say it proves the two locked cores are viable, especially if it does so at stock voltages.

Let me emphasize that I'm asking you to run the CPU at stock speeds and voltages right now as a four core for 20 minutes. That would mean the multiplier would be set to 15x I believe since the stock frequency of the 960T is 3.0 ghz. Correct?

[Chester326]

Originally Posted by trents

No, locking and unlocking the cores and then locking them again will not create instability.

If the system passes two hours of Prime95 blend as a six core at 15x I would say it proves the two locked cores are viable, especially if it does so at stock voltages.

Let me emphasize that I'm asking you to run the CPU at stock speeds and voltages right now as a four core for 20 minutes. That would mean the multiplier would be set to 15x I believe since the stock frequency of the 960T is 3.0 ghz. Correct?

Yeap, it's 3.0Ghz

Oke let me recap also, 20 mins on default stock speed cores locked.
Thats all? Need anything else? Run it again on 6 core for 20 mins with x15?
or anything like that anything i can do to provide you guys the info you need.

[trents]

You got it! I think we're on the same page now.

I know this probably seems like a step backwards to you after you were running at 3.6 but successful overclocking is a systematic, incremental discipline where we don't assume anything. We also try to change as few variables as possible at a time so when we encounter instability we can eliminate as few potential causes as possible.

With a lot of experience you can jump start things to a point but when first learning its much better to take it slow and easy from the start.

If you'll bear with me in this I think you'll be pleased in the end and have much better grasp of the process.

[Chester326]

Originally Posted by trents

You got it! I think we're on the same page now.

I know this probably seems like a step backwards to you after you were running at 3.6 but successful overclocking is a systematic, incremental discipline where we don't assume anything. We also try to change as few variables as possible at a time so when we encounter instability we can eliminate as few potential causes as possible.

With a lot of experience you can jump start things to a point but when first learning its much better to take it slow and easy from the start.

If you'll bear with me in this I think you'll be pleased in the end and have much better grasp of the process.

Nah, even going back Seems exciting OC is a whole new world for me so everything excites me, its a slow procces confirming every step as thorogly as possible, so when we reach a posible thorn to have as little rule outs as possible, its like trying to determine what is wrong with a defunk PC were you try diff thing until you find the defunk part/attribute

See ya laters its benchmark Time


(God i LOVE this)

[Chester326]

Originally Posted by trents

No, locking and unlocking the cores and then locking them again will not create instability.

If the system passes two hours of Prime95 blend as a six core at 15x I would say it proves the two locked cores are viable, especially if it does so at stock voltages.

Let me emphasize that I'm asking you to run the CPU at stock speeds and voltages right now as a four core for 20 minutes. That would mean the multiplier would be set to 15x I believe since the stock frequency of the 960T is 3.0 ghz. Correct?

Oke Results time 4 Cores x15 Multi, 20 mins on Prime is.

Rebooting to try 6 core Benchmark.

[Chester326]

so Far Running the test on 6 cores

[trents]

The attached image in post #25 is not attached. Nothing to see. What was the temp differential as a four core between TMPIN0-1 and core temp?

[Chester326]

Originally Posted by trents

The attached image in post #25 is not attached. Nothing to see. What was the temp differential as a four core between TMPIN0-1 and core temp?

Core temps are 29C Steadily
and TMPin2 is 36 C

[Chester326]

This is the 6 core temps

32 Seems to be the core temp (stuck due to unlock)
42 C TMP01

[keny]

Those core temps are looking dubious! Only an average of 4c increase after 20 mins of prime, I am Interested as to how this pans out with regard to temps as I have the same chip and trying to get even a rough guide for temp reading is impossible, you're core temps are reding ~25c , mine read ~14c , I have also noticed on hwm it reads the chips as 116w when they are supposed to be 95w, and when unlocked they skyrocket to something like 174w wtf ! I know you can use skt temps as a guide but how do you start the math if the initial core temp is optimistic to say the least and skt temp will vary if using air or water

[Chester326]

Originally Posted by keny

Those core temps are looking dubious! Only an average of 4c increase after 20 mins of prime, I am Interested as to how this pans out with regard to temps as I have the same chip and trying to get even a rough guide for temp reading is impossible, you're core temps are reding ~25c , mine read ~14c , I have also noticed on hwm it reads the chips as 116w when they are supposed to be 95w, and when unlocked they skyrocket to something like 167w wtf ! I know you can use skt temps as a guide but how do you start the math if the initial core temp is optimistic to say the least and skt temp will vary if using air or water

oke i will need some translation here plix
(i use Air Cooling CM Haf 932 Case)

[trents]

So, there was 7 C difference between CPU socket temp and core temp as a four core. So subtract that from the TMPIN1 when you're running as a six core to get a core temp estimate.

Keny, they are quite low but his CPU VCORE is only 1.3.

Chester, what is the room temp where your computer is? You're not doing these tests in an unheated room in the winter are you?

[trents]

Okay, go for the two hour Prime test now on stock speeds and voltages. Have HWMonitor open and report back with pics.

[keny]

There's is ultimately something wrong with the way the core temps are being read(I don't mean to hijack this thread I'm just curious) so the math that you are using is 7c cooler than skt temp = core temp.but what if the core temp is reading 10c cooler than the actual temp, sorry to go on about this but I'm just trying to get some constructive answers to make my own assumption on where my own actual core temps are lying. Thanks Mark

[trents]

But its based on the socket temp and it would be odd if both are falsely low unless HWMonitor is not reporting correctly.

Chester, do you have a monitoring utility that came with the motherboard we could check HWMonitor socket temp readings against?

[ssjwizard]

Gigabyte doesnt have one. There temp probes line up the standard way with HWmonitor.
The only software giga offers is cloud OC(moderate junk) and EasyTune(very junk) but AOD and HWmon work great for me on all my gigabyte boards just fine.

[keny]

I here where you are coming from but skt temp is only accurate if the math used to determine the difference between the skt and core is correct "ie" the core temp must be the correct reading on x4 and that's where I think the problem is, I don't think the x4 core temps on these chips are correct, I've tried mine in 3 boards and they all read the same in core temp + hwm, they rarely break 20c which is not possible,

[ssjwizard]

Every now and again you get a chip with a bad temp probe. Regardless of that the socket temp should be right. Its not calculated on the mobo from the core temp its got its own diode inside the socket. Even if you assume the CPU cores are at the socket temp keeping that below 55-60c will ensure a safe and long life for the chip.

[trents]

Originally Posted by ssjwizard

Every now and again you get a chip with a bad temp probe. Regardless of that the socket temp should be right. Its not calculated on the mobo from the core temp its got its own diode inside the socket. Even if you assume the CPU cores are at the socket temp keeping that below 55-60c will ensure a safe and long life for the chip.

Keny, what he said is what my point was. And a 7 degree differential between skt temp and core temp is certainly believable.

[keny]

Originally Posted by ssjwizard

Every now and again you get a chip with a bad temp probe. Regardless of that the socket temp should be right. Its not calculated on the mobo from the core temp its got its own diode inside the socket. Even if you assume the CPU cores are at the socket temp keeping that below 55-60c will ensure a safe and long life for the chip.

This is the way I'm going to read my temps, as I don't trust the core temps on thubans, it's pretty well documented that they can be way of, and no doubt the skt temps are correct as they are motherboard related and as you stated totally seperate from the chip, ok guys thanks for the discussion, I've decided skt temp for me, can't go wrong then.
Mark