Intel® Parallel Amplifier

How to provide C source code to Hotspot Analysis

Can you please tell me how to provide the source code of any program to intel vtune. My goal is to observe which function in the application is taking most of the execution time. I provide the executable file to intel vtune and I am running the application in the linux environment. I have observed that in the Botton-up tab, you can select the grouping mode "source file /class/fucntion/call stack" and the "function/call stack". I was expecting to see the functions' detail implemented in the source code of the program. But that is not the case.

Cannot load data file (data file is corrupted)

I just installed VTune Amplifier XE 2015. Trying to run the tachyon tutorial example I get the error: "Cannot load data file `D:\tmp\r000hs\data.0\7576-2976.0.trace' (Data file is corrupted)." I also get the warning: "Cannot find data to precompute. Skipping the precomputation step."

project property----Duration Estimation

    I'm using Intel Vtune Amplifier XE to to profiling jobs and I'm confused about the selection of Duration Estimation in Project Property. If I choose Between 1min and 15min and enable multiple runs, does it mean that the duration time of multiple runs is between 1min and 15min or the single run time of target application is between 1min and 15min?

Mapping assembly / source.

From the documents, I read that the attribution of cycles to instructions or source code lines can sometimes be one off. A line too early, or too late.

In my run, I see very inaccurate mappings. Both between source code and assembly code. And also between cycles spent and source code. See the attached screen shot. My source code is mostly intrinsics, so I have a pretty good understanding to what it should map. Yet, a simple intrinsic sometimes maps to a whole page of assembly? How could this be?

How can I profile the java application for hotspot analysis?

Hello,

I am running the wordcount example of hadoop on intel vtune for hotspot analysis. Command is bin/hadoop jar /usr/local/hadoop/hadoop-examples-1.2.1.jar wordcount inputfile outputfile. Hotspot analysis provides the modules information libzip.so, [compiled java code], [Dynamic code], libjvm.so, libc-2.17.so, libz.so.1.2.8 and [unknown]. My understanding is that [unknown] module is the wordcount application. However, to be sure I want to see what function inside the wordcount example are the hotspot functions. 

"Cannot locate file vtsspp.ko"

Today I started an evaluation of VTune Amplifier for Linux. (2015 edition, update 1.) I installed it on 'Ubuntu 14.04' with sudo root. The installation process reported success. Yet, when I do an advanced hotspot analysis, I get this warning: Cannot locate file 'vtsspp.ko' When installed with sudo root, and the installation process succeeds, should the kernel module not be built properly and be able to be found? When I search for the module with find, it yields nothing: $ find /lib/modules -name vtsspp.ko -print $ Thanks, Bram Stolk

“memory bound” metric in Vtune

Hello,

I am analyzing “memory bound” metric in my code with Vtune. According to "Intel® 64 and IA-32 Architectures Optimization Reference Manual-B.3.2.3":

%L2 Bound =(CYCLE_ACTIVITY.STALLS_L1D_PENDING - CYCLE_ACTIVITY.STALLS_L2_PENDING)/CLOCKS

But in my Vtune results, CYCLE_ACTIVITY.STALLS_L1D_PENDING is smaller than CYCLE_ACTIVITY.STALLS_L2_PENDING, why?

ittnotify in xeon phi offload regions

I'm trying to call the functions __itt_pause() and __itt_resume() from within offload regions so that I can only sample specific sections of my code in a knc-bandwidth collection. However I'm having problems with linking that I can't figure out. I am using Intel(R) VTune(TM) Amplifier XE 2015 Update 1.

My code looks like:

#pragma offload_attribute(push,target(mic))
#include <ittnotify.h>
#pragma offload_attribute(pop)

...

#pragma offload target(mic:offload_target)
{
    __itt_resume();
}

My make options are:

S’abonner à Intel® Parallel Amplifier