Latitude E6400 Owner's Lounge

I understand speedstep, and how the CPU clock frequency changes dynamically with loading. The curious thing is how different programs report the core speed. I can turn off RMclock so that there is no undervolting at all. CPU-Z version 1.50 and Everest CPUID report two different idling clock speeds. When the system is fully loaded, they both report 2.4 GHz.

 

I know that Everest shows you "live" feedback.. Perhaps RMclock is not "live" and grab it before the CPU slows down, OR uses a different mean to calculate the CPU speed, like actually calculate it, and of course when it calculates it, it uses the CPU so it increases. That is my guess.

 

Here's a screen-shot of what I am trying to say. The two programs report different speeds. This has nothing to so with RMClock.

 

In my experience, CPU-z only checks the CPU speed occasionally, so the display is only occasionally correct.

John

 

That is WHQL-ing, which is sometimes referred to as signing. it is actually an entirely different process, i should have been more clear. if a driver does not have a valid signature, it will not install unless you disable driver signing (which can be done, just not by the disable integrity check method.

yes, you can -- it is quite annoying to have to do this on every startup. there is another alternative, which is called ready driver plus (which basically puts a script into your startup which manually f8s for you on each bootup.) available here :: http://citadel.x10hosting.com/readydriverplus/

 

I believe this has to do with SuperFLM. Everest and RMClock should be right. From what I've seen, it seems like CPUZ assumes you're still running at 1066MHz FSB and only checks the multiplier (266MHz x 6 = 1600) when determining the on-the-fly clock speed. In actuality, SuperFLM reduces your FSB and the multiplier; the real clock speed should be ~800MHz.

 

You need to use CPU-z 1.41 or older to see the correct FSB speed.

I'm puzzled why that feature went wrong in the newer versions.

John

 

Unless I'm doing something stupid, 1.41 doesn't read SuperFLM right either; I'm still getting ~1600 on my E6400

 

If RMClock is right, then there is a serious issue with undervolting on P8600. As I was trying to say, when I enable PST transitions, select Performance On Demand, the core speed goes up to 1.6 GHz in idling. Which would consume more power in the idling state. That does not make sense. If I disable PST, the core speed drops down to 800 MHz, but then I cannot undervolt.

Here is another theory. If only one core is being used, as in idling, some programs report the core speed of the core that is running at the slower speed. Maybe something to do with IDA (Intel Dynamic Acceleration) which allows different speeds for the two cores. Perhaps RMClock does not go into IDA when PST is selected.

 

I keep getting "hidden" BSODs with the nVidia drivers -- hidden because they only show up in the Events log and as an obvious video driver crash-and-restart when coming out of sleep. It's pretty obvious that both the sound and video drivers are still a bit, um, "wonky" -- as are the Intel Matrix and Dell ControPoint stuff.

Sorry, I'm not in front of my E6500 at the moment so I can provide my current version numbers, but has anybody found a good fix for these applications' "growing pains"?

Any drivers or versions that seem to provide a stable environment?