Cache again....

[Raistlin]

@jj2007 = thanks again, you rock (re: Intel, OpenLibSys.org source)

@Nidud = you nailed the OS scheduler bit - thanks ! - I'am unsure however what they mean by page fault...
            - seems Microsoft uses the term loosely to describe issues with virtual memory, re: hard/soft/cache faults combined?
            - But I could be wrong, will investigate.

@Everyone_else = ....or we could just write a kernel driver and get it over with ? I found the kernel driver tutorials on the old masmforum site,
                            hope they are still relevant. Supposedly there's a trick to include the driver inside your exe, thereby not having to install it.
                            CrystalCpuID and others use this trick. Just for interest sake -> licensing ($$$) is required for all the other suggestions. Besides
                            it goes against my moral fiber to pay for something I can do myself. Then there's also religion:  we can write it better, faster, smoother
                            and more flexible than anyone else using assembly

[AW]

Supposedly there's a trick to include the driver inside your exe

That is not the most difficult part of the project. 
It is just placed inside the file as a resource, or as an included file, then extracted and placed where needed. Many applications do it, namely well know Sysinternal applications like DbgView.exe.

[Raistlin]

Code: [Select]

That is not the most difficult part of the project.  :biggrin:
per AW27 = I'am getting the feeling, I don't know what I'am letting myself in for.

But then again it wouldn't be the first time............. you have to be a little crazy to undertake large
projects that have indistinct outcomes at making a difference. The belief is - it's worth it to find out.

[jj2007]

or we could just write a kernel driver

I vaguely remember that recent Windows versions install only signed drivers; where "signed" translates to a lot of money

Kernel-Mode Code Signing Requirements

For developers:

A kernel-mode driver that is not a boot-start driver must have either a test-signed catalog file or the driver file must include an embedded test signature. This applies to any type of PnP or non-PnP kernel-mode driver.

[AW]

or we could just write a kernel driver

I vaguely remember that recent Windows versions install only signed drivers; where "signed" translates to a lot of money

Kernel-Mode Code Signing Requirements

Not only signed drivers, class 3 signed drivers from only a restrict number of authorities accredited by Microsoft.
I actually have one of those certificates, but have not paid a lot of money because there are opportunities for big discounts.

[Raistlin]

After doing some serious reading - I'am abandoning the kernel mode driver idea 
It's just too convoluted and takes the fun right out of it. What a bunch of B/S just
to get to one OS performance privileged instruction. What a shame....
For those interested in the amount of hoops you need to jump through, here's a taste:
http://www.davidegrayson.com/signing/

Any idea in regards to an API that exposes the performance metrics - documented or not ?
The following looks promising : https://msdn.microsoft.com/en-us/library/windows/desktop/ms724509%28v=vs.85%29.aspx?f=255&MSPPError=-2147217396

[AW]

http://www.davidegrayson.com/signing/

Well, the article about driver signing contains lots of bullshit and actually Windows 7 supports SHA-2 signing.

Any idea in regards to an API that exposes the performance metrics - documented or not ?

The NtQuerySystemInformation contains many undocumented classes, but you can grab some information in the Net, namely here.

[Raistlin]

Thanks AW27 - you've been a real   help and diplomatic in your responses