Pentium D 820 @ 3.4 on air

[smopoim86]

I have a pentium d 820 running at 3.4 (stock2.8) with a zalman cnps7700 AlCu heatsink. What i was wondering is at what temps does this processor generally run, because i am hitting 70 degrees under full load and throttle watch is not finding anything.
I thought that this proc would throttle around 65-68.

any info would be appreciated, i'm feeling a lil too lazy to look for the info myself. Thought some of you may know.

Edit: I reduced the voltage to 1.25 and now my temps are 60-63 under full load

 

[stevty2889]

Originally posted by: smopoim86
I have a pentium d 820 running at 3.4 (stock2.8) with a zalman cnps7700 AlCu heatsink. What i was wondering is at what temps does this processor generally run, because i am hitting 70 degrees under full load and throttle watch is not finding anything.
I thought that this proc would throttle around 65-68.

any info would be appreciated, i'm feeling a lil too lazy to look for the info myself. Thought some of you may know.

They run quite hot, the Zalman 7700 is pretty dated and not really able to keep up with the heat from a smithfield very well. Throttlewatch doesn't seem to work properly with Pentium-D's, as I know for sure my 830 was thorttling but throttlewatch didn't show it. Throttling usualy kicks in at 72c at stock speeds/voltage. If you didn't increase the vcore, you are likely just barely under throttling temps.

 

[Duvie]

Frankly throttlewatch has been unreliable since these dual core chips....

You want to see throttling? Download an app called SNM...run the torture test and more importantly the cpu/cache/fpu test and then switch between cpu1 and cpu2....you will see clock change on at least one of the cores when it begins....

I have seen a chip throttle as much as 1ghz with temps lower then you state...that being said Stevty is right and that point of throttle beginning happened at a different temperature...each time vcore was raised above stock the temperature point dropped at where throttling woudl occur.....It was much as only having a .05v and throttling would begin at 65c.....

We know it was throttling cause as the case was next to an open window in like October in Oregon we could keep taking side case off and the temps would drop and the cpu cores (both of them would stabilize to its correct core speed....



So basically what I am sayings is thottlewatch is not a defintive answer, cause I dont even think it said we were throttling...SNM is SMP aware so it will load (fully) both cores more so then even prime95 and even can detect errors....

 

[stevty2889]

Originally posted by: Duvie
Frankly throttlewatch has been unreliable since these dual core chips....

You want to see throttling? Download an app called SNM...run the torture test and more importantly the cpu/cache/fpu test and then switch between cpu1 and cpu2....you will see clock change on at least one of the cores when it begins....

I have seen a chip throttle as much as 1ghz with temps lower then you state...that being said Stevty is right and that point of throttle beginning happened at a different temperature...each time vcore was raised above stock the temperature point dropped at where throttling woudl occur.....It was much as only having a .05v and throttling would begin at 65c.....

We know it was throttling cause as the case was next to an open window in like October in Oregon we could keep taking side case off and the temps would drop and the cpu cores (both of them would stabilize to its correct core speed....



So basically what I am sayings is thottlewatch is not a defintive answer, cause I dont even think it said we were throttling...SNM is SMP aware so it will load (fully) both cores more so then even prime95 and even can detect errors....

That still doesn't add up to me, since TM1 thorttling doesn't change clock speed..TM2 does, but smithfields don't have TM2..I think there may have been something else funky going on with Marks 820 when you were getting the clock speed changes, because it has neither speedstep nor TM2.

 

[Duvie]

Originally posted by: stevty2889

Originally posted by: Duvie
Frankly throttlewatch has been unreliable since these dual core chips....

You want to see throttling? Download an app called SNM...run the torture test and more importantly the cpu/cache/fpu test and then switch between cpu1 and cpu2....you will see clock change on at least one of the cores when it begins....

I have seen a chip throttle as much as 1ghz with temps lower then you state...that being said Stevty is right and that point of throttle beginning happened at a different temperature...each time vcore was raised above stock the temperature point dropped at where throttling woudl occur.....It was much as only having a .05v and throttling would begin at 65c.....

We know it was throttling cause as the case was next to an open window in like October in Oregon we could keep taking side case off and the temps would drop and the cpu cores (both of them would stabilize to its correct core speed....



So basically what I am sayings is thottlewatch is not a defintive answer, cause I dont even think it said we were throttling...SNM is SMP aware so it will load (fully) both cores more so then even prime95 and even can detect errors....

That still doesn't add up to me, since TM1 thorttling doesn't change clock speed..TM2 does, but smithfields don't have TM2..I think there may have been something else funky going on with Marks 820 when you were getting the clock speed changes, because it has neither speedstep nor TM2.

Nope....Nothing funky could be going on when I was purposely removing and placing side case on and off. Once it hit a certain temp it started to begin...If I could cool it back down past that point it would be fine....There is no way that could be anything then temperature related....I also would raise the vcore in .025v nothces in the bios and even at same cpu speed of aournd 3.4ghz we were testing and it would happen but about 2-3c earlier....

Same SNM on other box on oc'd X2 never reacted funny running ...


I know what you are saying but we saw it....We were able to quantify it with benches at those points...It was obviously slower performance when SNM was showing core slowdown....

 

[stevty2889]

I don't doubt it was temp related, but there is no feature on the 820 that should have been lowering the actual clock speed. When my 830 was thorttling, S&M didn't show any changes in clock speed, nor did throttlewatch show it, but the drastic changes in benchmarks definatly did. When I let him put my 920 on his motherboard, there were other strange things that went on as well, such as unplugging the keyboard allowing it to boot when otherwise it would just sit there, and thats also one of the boards I was getting abnormaly low temps on with my 920.

 

[smopoim86]

Would looking at the cpu speed in CPU-z be a way of seeing if it was throttling. I ran load tests in a loop for about 3 hours and they all finished within 4 seconds of each other.

 

[Markfw]

Yup, Duvie and I played a lot, and you HAVE to run S&M to see it (and watch both cores, one will throttle sometimes and leave the other alone) The more volts you put to it, the lower the temp it will throttle at also. What vcore are you running ? And what motherboard ?

 

[smopoim86]

Asus p5wd2 premium, I would have to check but i think 1.38-1.4 not over 1.4


Edit: 1.375 volts, but when the chip is idle cpu-z is showing 1.4 volts. Why is the voltage going up when the chip is idle? when it's under load the voltage shows up right in the 1.3 range.

 

[Markfw]

more load causes voltage drop, the cpu draws a LOT of power, its a hog ! At that vcore (almost the same as mine) I think it throttles at 65c.

 

[stevty2889]

Originally posted by: Markfw900
Yup, Duvie and I played a lot, and you HAVE to run S&M to see it (and watch both cores, one will throttle sometimes and leave the other alone) The more volts you put to it, the lower the temp it will throttle at also. What vcore are you running ? And what motherboard ?

I have not doubt it was doing it, just trying to point out that what you were seeing, is NOT normal, TM1 thorttling does not change the CPU frequency, and smithfield does not have TM2 thorttling or speed step, so what was happening with S&M and your 820 is not normal, I did not get the same results when my 830 was throttling, nor when my P4's throttled(but at least throttlewatch worked with the P4's).