topic I also posted this on in Software Tuning, Performance Optimization & Platform Monitoring

Can't run the Intel Power Gadget

John_D_8 — Thu, 05 Mar 2015 03:16:27 GMT

I'm trying to run the Intel Power Gadget on a Win 8.1 x64 system with a Core i7-720QM. The program and the MS Visual C++ 2010 redistributable library appear to install OK, but the tool will not start. I try launching it, and nothing happens. I attached windbg to the executable, but did not learn anything. I also investigated it with Process Monitor and Dependency Walker, but couldn't get to the bottom of it.

The same utility runs fine on a different computer. Curiously, I might have not installed the MSVC 2012 redistributable on that PC. I don't see it listed in Programs and Features. The power gadget runs just fine though.

Any ideas?

Can you post screenshot of

Bernard — Thu, 05 Mar 2015 08:54:01 GMT

Can you post screenshot of Dependency Walker and screenshot of windbg output? In your case Power Gadget can exhibit so called Silent Process Termination.

Attached are 4 files. Two

John_D_8 — Thu, 05 Mar 2015 17:11:06 GMT

Attached are 4 files. Two are screenshots from Dependency Walker and windbg, and two are text outputs from those utilities.

Thanks for thinking about it.

I am the author of Intel

Seung-Woo_K_Intel — Thu, 05 Mar 2015 17:34:47 GMT

I am the author of Intel Power Gadget. Can you tell me your machine's spec? Please run CPUZ utility and send me the report.

I also posted this on

John_D_8 — Thu, 05 Mar 2015 17:35:59 GMT

I also posted this on StackOverflow. A user kindly analyzed the executable and came up with this conclusion:

i took a look at the offending exe it seems there is an integer division by zero exception in EnergyLib64.dll->Initialization Routine called by initterm (LdrpRunInitializeRoutine) when it checks for some processor specific functionality using cpuid the result of cpuid after some calculations is shifted right by 20 shr eax,20 which makes eax 0 and the divisor ebp is also 0 so div eax, ebp results in a divide by zero exception which leads to immediate termination. via msvcrt!exit at 0x......9e78

So, it seems that there's a bug which only affects systems with certain CPUs.

Can someone at Intel report this to the utility authors? The authors appear to be: Seung-Woo Kim, Karthik Krishnan, Vardhan Dugar, Joseph Jin-Sung, and Jun De Vega.

Thanks.

We are still using VS 2010

Seung-Woo_K_Intel — Thu, 05 Mar 2015 17:38:52 GMT

We are still using VS 2010 for backward compatibility, so the run time distributable that comes with the package should be enough. There must be something else going on.

Quote:Seung-Woo Kim (Intel)

John_D_8 — Thu, 05 Mar 2015 17:43:51 GMT

Seung-Woo Kim (Intel) wrote:

I am the author of Intel Power Gadget. Can you tell me your machine's spec? Please run CPUZ utility and send me the report.

Attached is a screenshot from CPUZ. Thanks for investigating.

That's why I asked for CPU-Z

Seung-Woo_K_Intel — Thu, 05 Mar 2015 17:47:40 GMT

That's why I asked for CPU-Z report to get the exact CPU specification, which comes from cpuid :)

Quote:Seung-Woo Kim (Intel)

John_D_8 — Thu, 05 Mar 2015 17:53:29 GMT

Seung-Woo Kim (Intel) wrote:

That's why I asked for CPU-Z report to get the exact CPU specification, which comes from cpuid :)

Attached is the full CPU-Z report as a text file in case you need more info than the screenshot I posted earlier.

Yep, it's Clarksfield. Intel

Seung-Woo_K_Intel — Thu, 05 Mar 2015 17:55:23 GMT

Yep, it's Clarksfield. Intel Power Gadget supports SandyBridge and later.

I should put up a dialog box

Seung-Woo_K_Intel — Thu, 05 Mar 2015 17:58:13 GMT

I should put up a dialog box when it's not supported rather than quietly quitting. Another idea for the next release... :)

Quote:Seung-Woo Kim (Intel)

John_D_8 — Thu, 05 Mar 2015 17:59:21 GMT

Seung-Woo Kim (Intel) wrote:

Yep, it's Clarksfield. Intel Power Gadget supports SandyBridge and later.

OK, thanks. I read the System Requirements too quickly and thought that all Core CPUs were supported.