Quantcast
Channel: Intel® VTune™ Profiler (Intel® VTune™ Amplifier)
Viewing all 1347 articles
Browse latest View live

2015 Update 1 crash log


2015 Update 1 MSVC GUI "mark timeline" says "not implemented"

$
0
0

Greetings,

as I said in the topic, when I click "mark timeline", in the GUI log reading "not implemented.

PS: I am using native x64 (not CLR app, nor Win Store app).

specs: Win 8.1 Pro x64, x64 target app, MSVC 2013 with the latest updates to both MSVC and Windows.

Regards

How to collect the detailed analysis of advanced-hotspots results

$
0
0

Hello,

I am running the application for the advance hotspot analysis using intel vtune. On GUI, this analysis gives me information like Function/Call Stack, CPU Time by Utilization and CPI Rate. I want to get the exact information but using command-line. If I use command

/opt/intel/vtune_amplifier_xe/bin64/amplxe-cl -report hotspots -r r006ah/

the information I get are the columns of Function, Module and CPU Time:Self but CPI Rate column is missing. Can you please give me any idea how to extract CPI Rate using commandline.

Appreciate your help.

Regards,

Can we get more elaborate information of module/function in advanced-hotspots analysis

$
0
0

Hello,

I am trying to find the hotspots of the application using intel vtune advance hotspots analysis. When I look at the Bottom-Up window. It provides the column "Function/Call Stack" along the Module column. I was expecting to see the functions of the running application in the column Function/Call Stack. But Functions/call stack column gives information from the modules like libjvm.so ,  ld-2.17.so, libc-2.17 etc. I have observed that one function is named as the "Outside any known module". I suspect that this function contains the information of my application (for example if I am running the helloworld program then it includes the main function of source code and print out statement).  Is it right?

Regards,

Resolving Remote Collection Symbols for a complex linux project using Amplifier XE 15 update1

$
0
0

Hi,

      I have root access on a remote SuSE 12.1 Xeon host running a DPDK application which I'm trying to profile using Vtune Amplifier XE 15 update1. I have installed remote collection on the server and am collecting data from the server successfully by attaching to the running DPDK application. I have access to all the kernel source and project sources but during finalization I'm getting a lot of warnings for kernel modules and system libraries which state that debug info cannot be found for these kernel modules and system libraries which the application is using. The documentation in VTune for setting up binary search paths is geared towards a windows user in the linux version and does not give any useful examples for complex linux envirionments. Does anyone on this list have more detailed documentation on setting up object file and symbol resolution for a remotely collected linux project ? 

      I'm getting useful information for the DPDK application by setting up the source paths to it but would like to resolve the warning messages during finalization so I know I'm getting the full analysis of all the system interaction on the UUT I'm testing,

 

     Thanks in advance for any helpful info,

             Regards,

                 Dave Sturtevant

Need more specific information for function elapsed time

$
0
0

Hi, now I profile some java code with VTune

I want to check whether some function are CPU-intensive or Data-intensive
for example, suppose one function that loads some data from array, and calculate arithmetic operation like this

result = (array[index+offset] ^ variable>>>=8) &0xff;

in this case, I think their elapsed time can be divided into two types; CPU-operation time (plus, exclusive-or, shift, etc) and Data-operation time (array[index+offset])

Can I get this kinds of information from VTune without modifying original source code?

 

 

Thanks

VTune Amplifier XE: 2015 fails to obtain license

$
0
0

Hello all,

we have a floating license for VTune since some time now. Recently, I got an email that a new version is available - the 2015 one. But I cannot get it to work with our existing license that we use with the 2013 version. We updated the license server to the latest version (l_isl_server_p_2.3.0.002_intel64.tar.gz), but I keep getting this error with the 2015 version of VTune:

Error: A license for PerfAnl could not be obtained
Your license is not current enough to allow you to use this
newer version of our software.  Usually this occurs because
your support services license expired before we created this
version.  You will need to purchase a new license.

Does this really mean we have to purchase a new license to get access to the 2015 version of VTune? Why did I get an email to the download section of 2015 then? I assumed that purchasing a license includes access to new versions of the software? This is quite sad :(

pintool compiling on windows

$
0
0

Hi all,

I currently encounter a problem compiling pintool on windows. Below is the environment:

OS: windows 7 sp1 x64

Visual Studio: 2012

pin: pin-2.14-67254-msvc11-windows.zip

To compile using the makefile, (inside makefile, linux command "cut" is used) I install cygwin (make, cut).

According to the release note: http://software.intel.com/sites/landingpage/pintool/docs/67254/README

Running commands:

cd source\tools\SimpleExamples
make opcodemix.dll

It should work. However it shows there is no target "opcodemix.dll". After deleting the ".dll", it starts compiling. But still, it shows this error:

E:\pin-2.14-vc11\source\tools\SimpleExamples>make opcodemix.dll
make: *** No rule to make target 'opcodemix.dll'.  Stop.

E:\pin-2.14-vc11\source\tools\SimpleExamples>make opcodemix
cl     opcodemix.cpp   -o opcodemix
Microsoft (R) C/C++ Optimizing Compiler Version 17.00.61030 for x64
Copyright (C) Microsoft Corporation.  All rights reserved.

cl : Command line warning D9035 : option 'o' has been deprecated and will be rem
oved in a future release
opcodemix.cpp
opcodemix.cpp(41) : fatal error C1083: Cannot open include file: 'pin.H': No suc
h file or directory
<builtin>: recipe for target 'opcodemix' failed
make: *** [opcodemix] Error 2

Could anybody help me to solve this by either going further in this way or point out another way I can try?

Thanks very much!


Visualization of Linux perf data files

$
0
0

Hello all,

I just tried out the new VTune 2015 feature of visualizing Linux perf data files. I have some issues with it. Since I could not find any list of known limitations, I raise them here in the hope that this feature will improve over time:

  • one cannot open perf.data or perf.data.old files. The files apparently must end with the  .perf suffix. This is cumbersome, as perf defaults to write perf.data files and rotates the file to perf.data.old. So it would be nice if you could make sure one can also open these files directly without having to rename them first.
  • I cannot see any userspace data. I used this example file:
    #include <cmath>
    
    using namespace std;
    
    int bar()
    {
      return cos(M_PI) * sin(0);
    }
    
    int foo()
    {
      return bar();
    }
    
    int main()
    {
      for (int i = 0; i < 100000; ++i) {
        foo();
      }
      return 0;
    }

    and compiled it without aggressive optimizations to get some callgraphs without inlining and run perf on it:
     

    g++ -g3 -O0 test.cpp -o test
    perf record --call-graph dwarf test
    perf report -g graph --no-children --stdio
    # To display the perf.data header info, please use --header/--header-only options.
    #
    # Samples: 25  of event 'cycles'
    # Event count (approx.): 8529342
    #
    # Overhead  Command  Shared Object      Symbol
    # ........  .......  .................  ...................................................................................
    #
        25.82%     test  libm-2.20.so       [.] 0x000000000001c923
                   |
                   --- 0x7f77b8ce5923
                       __gnu_cxx::__enable_if<std::__is_integer<int>::__value, double>::__type std::sin<int>(int)
                       bar()
                       foo()
                       main
                       __libc_start_main
                       _start

    Opening the same data file in vtune though only shows [vmlinux] and [Outside any known module] without any call graph information.

I hope that someone will look into this and improve this feature for future releases.

BSOD on Windows 7 using Amplifier 2015 Update 1

$
0
0

I am getting a 100% repeatable BSOD on Windows 7 running Amplifier 2015 Update 1 (build 380310), native code.

Advanced Hotspots ->Hotspots -> NO PROBLEM
Advanced Hotspots ->Hotspots , stacks and context switches ->BSOD after a few seconds.

Suggestions?

 

AttachmentSize
DownloadBSOD.zip38.29 KB

General Exploration--LLC

$
0
0

Hi,

I would please like some help on the following. I am doing a general exploration experiment on a small test case run on 24 cores on liniux. While in the "hardware issues view I see CPI being  over 1" and quite high on some particular threads(2.57etc)    LLC, Contested Acceess, Branch prediction and data sharing  shows to be  0. (As each thread is running on an array (size number of threads) I would expect to see data sharing actually.   Instead I see nothing, so no explanation why CPI is high. My run is quite small. Is it possible I am not hitting a hardware count limit and I see nothing? Is it possible for me to adjust this?

Thanks

 

 

 

 

Metric for execution count per line

$
0
0

Hi All, 

I'm working on Optimizing a Fortran Application, and observing the effects on Vtune Amplifier XE 2013 along side. I'm experimenting with the precision of operations in one of lines in my code. 

For instance ,
abc(1:BATCH_SIZE) = exp (-r_arr(1:BATCH_SIZE) * (1.d0/(4.d0 * ri * r1_arr(outer_j:jend))))

contains all 64-bit operands. Now, if I downgrade the operands to 32-bit and re-write my expression as : 

abc(1:BATCH_SIZE) = exp (real(-r_arr(1:BATCH_SIZE)) * (1.0/(4.0 * real(ri) * real(r1_arr(outer_j:jend)))))   

I get some reduction in the CPU Time from the per line counter through VTune (both cases profiled for same elapsed time) . Also, another metric that should change (increase) is the execution count of this line, because the 32-bit code will execute quicker than 64-bit code.

Which metric should I be looking for through VTune for comparison? 

Regards,
Apaar

 

 

phtread, vtunes amplifier crash

$
0
0

Hi,

I try to use vtune amplifier with our software package, but it crashes with following message:

Assertion 'g_instr_global.is_libpthread_so_loaded == 0' failed.[ASSERTION CONTEXT]g_instr_global.is_libpthread_so_loaded == 1
[CONTEXT END]. Please contact the technical support.amplxe: Error: Assertion 'g_instr_global.is_libpthread_so_loaded == 0' failed.[ASSERTION CONTEXT]g_instr_global.is_libpthread_so_loaded == 1
[CONTEXT END]. Please contact the technical support.
Segmentation fault (core dumped)

I've already tried to preload the pthread library, but it's not working. Has anybody an idea ?

 

regards 

flo

 

Running VTune amplifier on a HyperV virtual machine

$
0
0

Hi,

Is it possible to run VTune amplifier on a HyperV virtual machine?  Currently I get the following error when I try to run a hotspot analysis

    11/17/14 11:15:49  Collection failed. The data cannot be displayed. 
    PMU resources are not available. Hardware Event-based Sampling is not supported on this operating system. PMU resources may be made unavailable either by the BIOS or Hypervisor.

The only thread I found on these forums with a similar issue was this one.   From my understanding the software was not running on a VM in this case - so disabling Hypervisor makes sense.

Thanks

Cannot locate debugging symbol

$
0
0

Hi,

I'm new in using profiling tools. I have few questions:

1) Does Intel VTune Amplifier XE 2015 currently have the specific feature of analyzing the performance of a source code on implementing OpenMP API?

2) When I tried to run Advanced Hotspots Analysis, it was completed with the warning:

cannot locate debugging symbols for file '/lib/x86_64-linux/libc-2.19.so'.

cannot locate debugging symbols for file '/lib/x86_64-linux/ld-2.19.so'.

3) What is the effect of missing debugging symbols?

4) When I was about to use VTune to perform analysis on OpenMP region, it displayed nothing. It is because of my OS version? Below is the info of my OS:

My OS: Linux Mint 17 Cinnamon 64-bit

Cinnamon Version: 2.2.13

Linux Kernel: 3.13.0-24-generic

 

 


Blue Screen during advanced hotspots analysis in attach-to-process mode

$
0
0

Hi,

I saw a Windows blue screen during a VTune analysis for the second time now, so I decided to create a new post about this.

I did an advanced hotspots analysis in attach-to-process mode while 3 instances of my application were running. They were running already for nearly a day (I did a long-term test) before I attached to one of these (independent) instances (advising VTune to attach to a certain process ID). I set the evaluation interval to 10 ms (default), and configured to catch only user events instead of all.

About a minute after attaching, a blue screen came up which said: "A clock interrupt was not received on a secondary processor within the required time interval." Yet I have seen this blue screen on that machine only when using VTune.

Used hardware: Dell T5610.L, 2x Xeon E5-2687W v2 @ 3.40 GHz, 64 GB RAM.

If Intel is interested in any logs or further information, don't hesitate to ask.

Regards, Hagen

Collection failed - [Instrumentation Engine]: SYSCALL_INSPECTOR:

$
0
0

Hello,
for any kind of profiling, after my application finishes, I get:

[Instrumentation Engine]: SYSCALL_INSPECTOR: The
NTDLL!NtSetContextThread function jumps out of NTDLL, at 0x077960400.
It may be hooked by a PIN-incompatible software installed on the system
Incompatible operating system or incompatible software installed on the
system Pin is exiting due to fatal error

What can cause that?

Thanks, Marek.

Vtune Amplifier XE 2015 with "Error: Amplifier cannot detect Intel Xeon Phi coprocessor configuration"

$
0
0

Hi, 

I am using VTune? Amplifier XE 2015 to profile a large program running natively on Xeon Phi,unfortunately it fails in the first place.

Then I tried to build the sample application: Matrix included in Vtune Amplifer and copied the native binary matrix.mic to the mic0:/tmp.

Basically I followed the steps as shown in the link below:

https://software.intel.com/en-us/articles/how-to-analyze-xeon-phi-coprocessor-applications-using-intel-vtune-amplifier-xe-2015

The application runs well on Xeon Phi, but I got errors when I was running the application using Vtune with amplxe-cl

$ amplxe-cl -r /home/local/xxx  -target-system=mic-native:0 -collect advanced-hotspots  /tmp/matrix.mic
amplxe: Using target: mic-native:0
amplxe: Error: Cannot find product on the target system at mic0:/amplxe/
Make sure VTune Amplifier installation directory on the remote system option in the Project Properties dialog box is set to the correct path. Alternatively, you may use the --target-install-dir option to specify the correct path from command line.
amplxe: Error: Amplifier cannot detect Intel Xeon Phi coprocessor configuration.
amplxe: Fatal error: Cannot create analysis type. Check input parameters or reinstall the product.

There is same issue with amplxe-gui, any suggestions? Thanks in advance. 

  System Info
                HOST OS                 : Linux
                OS Version              : 2.6.32-504.1.3.el6.x86_64
                Driver Version          : 3.4.1-1
                MPSS Version            : 3.4.1
                Host Physical Memory    : 16251 MB
Device No: 0, Device Name: mic0
        Version
                Flash Version            : 2.1.02.0381
                SMC Firmware Version     : 1.11.4404
                SMC Boot Loader Version  : 1.8.4326
                uOS Version              : 2.6.38.8+mpss3.4.1
                Device Serial Number     : ADKC31600630

Regards,

Qi

collection failed: Internal Error

$
0
0

Hi,

vtune was used to profile a process owned by root and failed to collect. Please see below for the screen capture.

OS is oracle linux server 6.5(2.6.32-431.el6.x86_64)

Thanks,

Jianbing

 

[root@oxford015 jianbingc]# amplxe-cl -collect -target-pid=12261

Copyright (C) 2009-2014 Intel Corporation. All rights reserved.

Intel(R) VTune(TM) Amplifier XE 2015 (build 367959)

amplxe: Collection started. To stop the collection, either press CTRL-C or enter from another console window: amplxe-cl -r /home/jianbingc/r001hs -command stop.

^Camplxe: Collection failed.

amplxe: Internal Error

VTune Data Collection automatically triggering prematurely

$
0
0

Hello, 

I am trying to perform Basic Hotspots analysis using VTune Amplifier XE 2015 on my win64 machine, and experiencing problems with VTune Data Collection log process triggering prematurely. 

This is my workflow. I click on the "Start" or "Start As Paused" (for Start/Resume workflow) button and am in the process of running a script in my application for profiling. However, after a few seconds VTune Data Collection Log is automatically getting triggered without my performing any other action in VTune (no "Pause" or "Stop"), and before my script is done running. 

Thus, the Data Collection Log errors out as follows. In "Finalization complete with Warnings" section it states that "Cannot load data file <data_trace_file_path> (Data is corrupted). 

FWIW, I was able to get VTune to successfully perform analysis the first time I used it with a different version of my application (using a different project) using the Start As Paused, Resume and Stop controls. 

I don't think I am doing anything different in the second scenario, but it is not working out. Can you help figure out what is wrong? Please let me know if you need any further information for debugging. 

This is probably irrelevant, but for reference, I have an install version of VTune. Initially, I had some licensing problems which was resolved by your tech support. 

thanks!

 

Viewing all 1347 articles
Browse latest View live


<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>