AMD Dragon Platform exhibition

[accord99]

proof?

Based on Shanghai power consumption, typical problems of early new CPUs to have problems with temperature sensors, etc. But in the end, the idle temperature doesn't matter very much as compared to the load temperature and both are subject to error and therefore less important that actual power consumption.

[initialised]

Based on Shanghai power consumption, typical problems of early new CPUs to have problems with temperature sensors, etc. But in the end, the idle temperature doesn't matter very much as compared to the load temperature.

At 4GHz+ i7 can have delta-T of 45-50C which usually kills the OC by overshooting the 95C limit and either crashing or throttling. 28C is promising assuming that delta-T + idle-T < Tmax

[T_Flight]

At 4GHz+ i7 can have delta-T of 45-50C which usually kills the OC by overshooting the 95C limit and either crashing or throttling. 28C is promising assuming that delta-T + idle-T < Tmax

There are something like 20 CPU's here at XS that have ran the Chess benchmark at between 4-4.35GHz, none of them were crashing or throttling. That benchmark fully loads all cores 100% and does 8 threads. I have one at 3.8GHz and have ran that bench and am over 30 degrees from the TjMax with a box stock cooler, when the cores are fully loaded for 10 minutes HT enabled 8 threads. It reaches max temp and planes off at around 2 minutes of calcultions. Once it reaches that temp it stays there. Not even near the TM limit.

[Tiridum]

There are something like 20 CPU's here at XS that have ran the Chess benchmark at between 4-4.35GHz, none of them were crashing or throttling. That benchmark fully loads all cores 100% and does 8 threads. I have one at 3.8GHz and have ran that bench and am over 30 degrees from the TjMax with a box stock cooler, when the cores are fully loaded for 10 minutes HT enabled 8 threads. It reaches max temp and planes off at around 2 minutes of calcultions. Once it reaches that temp it stays there. Not even near the TM limit.

you can disable throtling...

[Movieman]

I'll be at the Boston little get together this Friday to see for myself.
yup, already got my invitation..

[Brother Esau]

I'll be at the Boston little get together this Friday to see for myself.
yup, already got my invitation..

Hey.......N00B Admin

[T_Flight]

Yes, I know, but there is no reason to. I've never gotten anywhere near the TjMax. 4 will get closer to it, but I haven't heard anyone that has said their system was crashing and throttling.

[Rammsteiner]

Based on Shanghai power consumption, typical problems of early new CPUs to have problems with temperature sensors, etc. But in the end, the idle temperature doesn't matter very much as compared to the load temperature and both are subject to error and therefore less important that actual power consumption.

First ES/steppings which dont hit retail are early CPU's as well

Since ambient should be around 20, I think 28 on idle does say a lot. Especially if you consider that both 9850BE and 6400+ idle with a TRUE where like ~30 on stock with ~20 ambient.

[accord99]

First ES/steppings which dont hit retail are early CPU's as well

Or Brisbane whose DTS were broken on release and has apparently yet to be fixed, or the stuck early Intel 45nm DTS, or their wildly changing Tjmax values.

Since ambient should be around 20, I think 28 on idle does say a lot. Especially if you consider that both 9850BE and 6400+ idle with a TRUE where like ~30 on stock with ~20 ambient.

It means very little without knowing the accuracy of temperature sensor; if there was a power consumption figure then that could corroborate or reject the idle value.

[informal]

Or Brisbane whose DTS were broken on release and has apparently yet to be fixed, or the stuck early Intel 45nm DTS, or their wildly changing Tjmax values.


It means very little without knowing the accuracy of temperature sensor; if there was a power consumption figure then that could corroborate or reject the idle value.

You will have all the numbers very soon.Patience.