VTune analyzer 2.0 beta test archives

VTune analyzer 2.0 beta test archives

jeffrey-gallagher (Intel)'s picture

This readonly thread is an archive of the postings that were posted during the beta test program for this software. The product is now currently available for sale or evaluation at:

http://www.intel.com/software/products/vtune/vlin

cheers

jdg

57 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

Has anyone been successful at getting this to work on Gentoo? What kernel level patches are needed for 2.6.X?

jeffrey-gallagher (Intel)'s picture

We've not tried gentoo around here, goanuj, and probably won't any time soon. However, here's the latest as of this writing on 2.6.

On the VDK website, we have version of VDK sources for kernel 2.6.0-test5 on x86 that works with VTune 7.0 and CLI 1.1.

http://www.intel.com/software/products/opensource/vdk/

However, it requires a patch to kernel 2.6.0-test5 (i.e., users must rebuild that kernel with the patch, and we include the patch along with the VDK sources).

We have not tested/ported VDK driver to any version of kernel 2.6.x (patched or otherwise) on Itanium Linux as yet.

I hope this helps.

cheers

jdg

jeffrey-gallagher (Intel)'s picture

Hello Henric,

You've hit the nail on the head already: the tarball you downloadedvt_cli_2.0.348.tar is not beta, it's final product. Beta licenses don't work with final product.

My understanding is that this licensing issue was a bug, and if you register again you will get an appropriate license (if you have a second email handy to give it a try, that is.) If that doesn't work for you, post here again and we'll investigate what is wrong with your specific registration.

cheers

jdg

Message Edited by jdgallag on 01-28-2004 10:41 AM

jeffrey-gallagher (Intel)'s picture

Henric,

Many thanks for taking the trouble to let us know what happened to you. I'll ask the web site controllers to take a look at the link that caused you the trouble.

cheers

jdg

Hi,

I'm trying to use the sampling feature of vtune to profile my application built with icc but I keep getting "no symbols were found" when I try to view the hotspot function results for a shared object (.so) module.

I've boiled this down to a simple test case which works fine with gcc but not with icc.

$ cat foo.h
#ifndef FOO_H
#define FOO_H
void foo ();
#endif

(please substitute "less-than" for appropriate symbol, the forum things I'm trying to post html when I have an angle bracket)

$ cat foo.c
#include "foo.h"
void foo () {
int i;
for( i = 0; i less-than 100000000; i++) {
}
}

$ cat main.c
#include "foo.h"
int main () {
int i;
for( i = 0; i less-than 5; i++) {
foo();
}
}

$ icc -g -shared -fPIC foo.c -o libfoo.so
$ icc -g -c main.c -o main.o
$ icc main.o -lfoo -L.
$ vtl activity -c sampling -app a.out run
[ vtune runs ]
$ vtl show -hf -mn libfoo.so
[ .. event information .. ]
No symbols were found

If I compile *exactly* the same way using gcc instead of icc I will get the desired results (i.e. I get event %'s about my functions). I've also tried compiling with gcc using the -gdwarf-2 flag in place of the -g flag which also gives the desired results.

Why does this not work with the intel compiler? Is VTune not capable of profiling an application built this way? Does anybody have any ideas what could be wrong?

Thank You,

Nathan

some info:

icc: Version 8.0
gcc: Version 2.96
OS : RedHat 7.3 [kernel 2.4.18-3] (also tried on RHAS2.1)

jeffrey-gallagher (Intel)'s picture

Hey Nathan,

VERY interesting problem description.

But before we dig in too far, which version of vtune are you using?

$ vtl version

VTune Performance Analyzer 2.0 for Linux*
Build: 348

jeffrey-gallagher (Intel)'s picture

Hey team,

If you're going to be in NYC for LinuxWorld JAN 2004, be sure to drop by the Intel booth and say hello to the VTune and compiler team members there!

We'd love to shake your hand, chat about these forums and how you're enjoying using them. Now, here's the formal announcement!

----------------------------------------------------------------------------------------------------

The 2004 LinuxWorld Conference and Expois January 21-23 at New York City'sJavits Center. If you plan on attending,stop by the Intel booth, #645, and check out our showcase ofthe latest Intel software development tools and services. <?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

We are alsoholdinga Linux software development tools Birds-of-a-Feather(BOF)on Wednesday, January 21, 5:30 PM to 7:00 PM in room 1E07. Stop byand receive a special gift!

Linux* Compilers and Profiling Tools: Do they REALLY Make a Difference?

If you've ever done any Linux software optimization or are thinking about it,this session is for you. Industry experts will be on hand to fuel discussion around common issues and use of tools in a variety of development environments.

Learn how to identify your current optimization challenges and hear how others are successfully managing theirs

Get with your fellow travelers to learn which tools work and which don't

Share information about best known practices and related costs

Profiling tools can substantially reduce
your development time. Don't let your busy schedule writing and rewriting code keep you away.For more information and to register for this BOF seminar, visit http://www.linuxworldexpo.com/linuxworldny/V40/conference/session.cvn?eID=388.

If you havent registered for the 2004 LinuxWorld Conference and Expo, do it now: http://www.linuxworldexpo.com/linuxworldny/V40/index.cvn. Join your fellow travelers and industry experts to get the latest on software development tools for Intel-based Linux development.

jeffrey-gallagher (Intel)'s picture

The title says it all.

Everybody here in the forum wants to wish everybody out there a safe, healthy, and prosperous 2004!

(Also, keep sending in those great questions, answers, and comments!)

cheers,

jdg
(for the whole Intel forum team)

Hi,
I'm running Remote Agent for Linux (vtserver),controlling it by a Windows system. Everything seems fine except that I can't get "Thread View" for no process!?
Also,was looking into the VTune Analyzer for Linux documentation and there is no reference to Thread View either!!
Jeff, please don't say "it's not supported for Linux"... :smileywink:

Thanks

jeffrey-gallagher (Intel)'s picture

:smileyvery-happy:

OK you won't hear it from me, rostam.

Fact is, I need a bit more from you to figure out where we're heading. VTServer is not part of vtl or vtl beta (the subject of this forum) but is either part of VTune 7.0 (currently released product) or VTune 7.1 (also in beta right now even as I type).

Let me know which VTune you're working with, and I'll take it from there.

cheers

jdg

So there is a bit of hope still...

I'm usingVTServer from VTune Analyzer7.1 Beta.

I guess you're implying Thread View should work with standalone Analyzer 2.0 for Linux. Is that right?

Thanks

jeffrey-gallagher (Intel)'s picture

Actually, not that I mind being accused of optimism (inadvertent or otherwise) I just wanted to get my bearings before proceeding.

:)

Stay tuned. I'll report back soonest.

jdg

Hi there,

I installed vtune 2.0 for linux on an 8-way SMP Xeon running linux AS2.1. After the installation, I was able to start the vtune driver. But when I invoke the "vtl" for activity collection, I get an error.

This is the error I got:
"Error 0x80040154: Unable to create the VTShell object. Make sure the Shell library is properly installed and registered." I could not find more information about this in the doc pages either.

Can you please help?

Thanks,
JP

jeffrey-gallagher (Intel)'s picture

VERY interesting report, Jayaprakashp!

It might make more sense to open a premier case for this, attaching your install.log file (that was created during the vtl installation).

In the meanwhile, consider:

1)Scan through the install.log file, it might show you some installation failures that can be corrected?

2) Did you install as root?

3) Was a previous version of vtl or the VTServer (the linux remote agent for the Windows version of the VTune analyzer) installed already when you installed?

4) Is NFS in your environment, and if so, can you uninstall and reinstall so that the software is installed fully on local filesystems only?

5) Do you see the same behavior with root and non root users, belonging to the appropriate "vtune" (if you picked that default) user group?

6) No guarantees, but might be worth uninstalling and reinstalling, just to see if the behavior changes.

Please let me know what you find, and, if nothing I suggested helps, be sure to open your premier case.

cheers

jdg

Hi Jdg,

Thanks for responding. I tried all your options and I'll reply to your questions one by one.

1)Scan through the install.log file, it might show you some installation failures that can be corrected?

I've attached it to the end of this email...

2) Did you install as root?

YES. I did install as root.

3) Was a previous version of vtl or the VTServer (the linux remote agent for the Windows version of the VTune analyzer) installed already when you installed?

I removed all installations of Vtune analyzer. I had a remote agent installed. But I found that it is lacking certain features that we expected - call graph fails to work with certain libraries, files were not brought to the windows machine properly... So, we decided to remove it and give the beta 2.0 a shot.I removed the remote agent entirely before installing 2.0.

4) Is NFS in your environment, and if so, can you uninstall and reinstall so that the software is installed fully on local filesystems only?

No. I installed it on the native /opt (default) mount which is on the local disk.

5) Do you see the same behavior with root and non root users, belonging to the appropriate "vtune" (if you picked that default) user group?

Yes. It gives me the same error if I run it as root, or as any other. I added "users" and "root" groups for privileges to run vtl - during installation.

6) No guarantees, but might be worth uninstalling and reinstalling, just to see if the behavior changes.

I tried uninstalling and reinstalling. It is giving me the same error. I checked everything and the required daemons seem to be running fine - ntd, sag, dcom, vtune. There are two failures - which were not there during the first install. For the first install, everything went smooth.

Here's the install log:

Wed 12 13:46:44 Success - User chose the all components installation.

Wed 12 13:46:49 Success - User chose to accept the EULA file for installing the product.

Wed 12 13:46:49 Success - User chose the VTune Performance Analyzer Driver Kit installation.

Wed 12 13:46:56 Success - User chose to accept the EULA file for installing the product.

Wed 12 13:47:27 Success - Test for required hardware and software - done.

Wed 12 13:47:28 Success - The group "sag" was created.

Wed 12 13:47:28 Success - The user "sag" has been added into group "sag".

Wed 12 13:47:28 Success - EntireX DCOM for Linux is installing into /opt/sag directory

Wed 12 13:47:31 Success - Installation of the dcom-license-v6.1.1-36.i386.rpm

Wed 12 13:47:57 Success - Installation of the dcom-v6.1.1-36.i386.rpm

Wed 12 13:47:58 Success - EntireX DCOM for Linux was successfully installed.

Wed 12 13:48:37 Success - Install destination set to /opt/intel/vtune

Wed 12 13:48:44 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 13:48:46 Succe
ss - Installation of the intel-vdk-src-2.0-146.rpm.

Wed 12 13:48:48 Success - Installation of the intel-vdk-2.0-146.x32.rpm.

Wed 12 13:48:49 Success - Scripts have been customized.

Wed 12 13:49:09 Success - User chose to install the VTune analyzer driver boot script.

Wed 12 13:49:11 Success - Configuring autoload of VTune analyzer driver for runlevels 2 through 5

Wed 12 13:49:11 Success - Installation of the VTune Performance Analyzer Driver Kit 2.0 is complete.

Wed 12 13:49:19 Success - The install destination has been set to /opt/intel/vtune

Wed 12 13:49:25 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 13:49:35 Success - Installation of the intel-vtune-cli-shell-2.0b-234.x32.rpm.

Wed 12 13:49:38 Success - Installation of the intel-vtune-cli-ism-2.0b-234.x32.rpm.

Wed 12 13:49:41 Success - Installation of the intel-vtune-bistro-2.0-13001.x32.rpm.

Wed 12 13:49:44 Success - Installation of the intel-vtune-lips-2.0-13001.x32.rpm.

Wed 12 13:50:03 Success - Installation of the intel-vtune-cli-analyzer-2.0b-235.x32.rpm.

Wed 12 13:51:52 Success - Directory for global data was set to /opt/intel/vtune/global_data

Wed 12 13:51:53 Success - Wrappers have been configured.

Wed 12 13:52:18 Success - VTune Performance Analyzer 2.0 Beta for Linux* was successfully installed.

Wed 12 14:15:39 Success - User chose the VTune Performance Analyzer 2.0 for Linux* installation.

Wed 12 14:15:46 Success - User chose to accept the EULA file for installing the product.

Wed 12 14:16:04 Success - Test for required hardware and software - done.

Wed 12 14:16:06 Failed - VTune Performance Analyzer 2.0 Beta for Linux* is already installed.

Wed 12 14:18:08 Success - User chose the VTune Performance Analyzer 2.0 for Linux* installation.

Wed 12 14:18:15 Success - User chose to accept the EULA file for installing the product.

Wed 12 14:18:34 Success - Test for required hardware and software - done.

Wed 12 14:18:34 Failed - Installation of the VTune Performance Analyzer 2.0 Beta for Linux*. dcom-v6.1.1 must be installed.

Wed 12 14:18:47 Success - User chose the VTune Performance Analyzer Driver Kit installation.

Wed 12 14:18:49 Success - User chose to accept the EULA file for installing the product.

Wed 12 14:19:12 Success - Test for required hardware and software - done.

Wed 12 14:19:13 Success - EntireX DCOM for Linux is installing into /opt/sag directory

Wed 12 14:19:16 Success - Installation of the dcom-license-v6.1.1-36.i386.rpm

Wed 12 14:19:42 Success - Installation of the dcom-v6.1.1-36.i386.rpm

Wed 12 14:19:42 Success - EntireX DCOM for Linux was successfully installed.

Wed 12 14:20:07 Success - Install destination set to /opt/intel/vtune

Wed 12 14:20:08 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 14:20:10 Success - Installation of the intel-vdk-src-2.0-146.rpm.

Wed 12 14:20:13 Success - Installation of the intel-vdk-2.0-146.x32.rpm.

Wed 12 14:20:13 Success - Scripts have been customized.

Wed 12 14:20:32 Success - User chose to install the VTune analyzer driver boot script.

Wed 12 14:20:34 Success - Configuring autoload of VTune analyzer driver for runlevels 2 through 5

<
P>Wed 12 14:20:35 Success - Installation of the VTune Performance Analyzer Driver Kit 2.0 is complete.

Wed 12 14:20:38 Success - The install destination has been set to /opt/intel/vtune

Wed 12 14:20:41 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 14:20:45 Success - Installation of the intel-vtune-cli-shell-2.0b-234.x32.rpm.

Wed 12 14:20:48 Success - Installation of the intel-vtune-cli-ism-2.0b-234.x32.rpm.

Wed 12 14:20:52 Success - Installation of the intel-vtune-bistro-2.0-13001.x32.rpm.

Wed 12 14:20:54 Success - Installation of the intel-vtune-lips-2.0-13001.x32.rpm.

Wed 12 14:21:04 Success - Installation of the intel-vtune-cli-analyzer-2.0b-235.x32.rpm.

Wed 12 14:21:59 Success - Directory for global data was set to /opt/intel/vtune/global_data

Wed 12 14:21:59 Success - Wrappers have been configured.

Wed 12 14:22:04 Success - VTune Performance Analyzer 2.0 Beta for Linux* was successfully installed.

Wed 12 14:56:20 Success - User chose the all components installation.

Wed 12 14:56:25 Success - User chose to accept the EULA file for installing the product.

Wed 12 14:56:25 Success - User chose the VTune Performance Analyzer Driver Kit installation.

Wed 12 14:56:28 Success - User chose to accept the EULA file for installing the product.

Wed 12 14:56:55 Success - Test for required hardware and software - done.

Wed 12 14:56:56 Failed - EntireX DCOM for Linux has already been installed.

Wed 12 14:57:20 Failed - VTune Performance Analyzer 2.0 Beta for Linux* is already installed.

Wed 12 15:00:29 Success - User chose the all components installation.

Wed 12 15:00:34 Success - User chose to accept the EULA file for installing the product.

Wed 12 15:00:34 Success - User chose the VTune Performance Analyzer Driver Kit installation.

Wed 12 15:00:37 Success - User chose to accept the EULA file for installing the product.

Wed 12 15:00:59 Success - Test for required hardware and software - done.

Wed 12 15:01:00 Success - EntireX DCOM for Linux is installing into /opt/sag directory

Wed 12 15:01:03 Success - Installation of the dcom-license-v6.1.1-36.i386.rpm

Wed 12 15:01:29 Success - Installation of the dcom-v6.1.1-36.i386.rpm

Wed 12 15:01:29 Success - EntireX DCOM for Linux was successfully installed.

Wed 12 15:01:44 Success - Install destination set to /opt/intel/vtune

Wed 12 15:01:45 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 15:01:47 Success - Installation of the intel-vdk-src-2.0-146.rpm.

Wed 12 15:01:50 Success - Installation of the intel-vdk-2.0-146.x32.rpm.

Wed 12 15:01:50 Success - Scripts have been customized.

Wed 12 15:02:24 Success - User chose to install the VTune analyzer driver boot script.

Wed 12 15:02:26 Success - Configuring autoload of VTune analyzer driver for runlevels 2 through 5

Wed 12 15:02:30 Success - User chose to load the VTune analyzer driver at this time.

Wed 12 15:02:31 Success - Installation of the VTune Performance Analyzer Driver Kit 2.0 is complete.

Wed 12 15:02:36 Success - The install destination has been set to /opt/intel/vtune

Wed 12 15:02:37 Success - User choice of install options "-U --replacefiles" was accepted.

Wed 12 15:02:41 Success - Installation of the intel-vtune-cli-shell-2.0b-234.x32.rpm.

Wed 12 15:02:44 Success - Installation of the intel-vtune-cli-ism-2.0b-234.x32.rpm.

Wed 12 15:02:47 Success - Installation of the intel-vtune-bistro-2.0-13001.x32.rpm.

Wed 12 15:02:49 Success - Installation of the intel-vtune-lips-2.0-13001.x32.rpm.

Wed 12 15:02:58 Success - Installation of the intel-vtune-cli-analyzer-2.0b-235.x32.rpm.

Wed 12 15:03:48 Success - Directory for global data was set to /opt/intel/vtune/global_data

Wed 12 15:03:49 Success - Wrappers have been configured.

Wed 12 15:03:53 Success - VTune Performance Analyzer 2.0 Beta for Linux* was successfully installed.

------------------------------

Can you please investigate and let us know? In the meanwhile, I think I should revert back to the remote agent!

~JP

jeffrey-gallagher (Intel)'s picture

Hey JP,

Do you know if this problem has already been reported to Intel's Support? If yes, what is the case number?

cheers

jdg

I was not able to open an issue with Intel Premier for Beta 2.0. Anyway, I've sent an email with the details I said here to quadsupport@mailbox.inte.com. I hope they open an issue for this.

Thanks,
JP

jeffrey-gallagher (Intel)'s picture

Jay I'm concerned you couldn't open a case. Can you please try again, and note that alphabetically, the product is not under "V" for VTune, but "B" for Beta.

What do you think?

cheers

jdg

Yeah. I did look for beta btw. I tried opening the case for VTUNE beta, but it fails saying "Internal error". May be I'm not officially allowed to post an issue for beta.

Thanks
JP

jeffrey-gallagher (Intel)'s picture

Try opening a report using either another computer, or a different browser on your current computer, or both.

We're not getting reports that there's a problem on our end.

If you registered the beta software, you should be able to open your own premier case.

If you can try one or both of my above suggestions, and you have the same problem, I'll want your email account to check out what's failing for you on this end. You can send your email account to me at vtune.sales@intel.com, subject FOR JDG

cheers
jdg

This was asked previously, but I'd like to re-ask, now that Redhat Enterprise Linux 3.0 is GA.

jeffrey-gallagher (Intel)'s picture

More on this: EL 3.0 won't be supported by vtl 2.0 this year, most likely though at the very beginning of next year.

NOTE though, that when VTune 7.1 releases at the beginning of next year, it will have EL 3.0 support right at its release. If those few weeks will make a difference in your optimization timeschedule, you should surely take a look at 7.1 when it comes out, rather than waiting for vtl 2.0.

Just wanted to make sure you knew all of your options.

cheers

jdg

jeffrey-gallagher (Intel)'s picture

Hey Robert.Zwolinski@compuware.com,

We're still hammering out our schedule for supporting it, RZ, and I'll post the info here as soon as it's solid.

(Yes, we'll be supporting it; the question is --> when.)

Stay tuned.

cheers

jdg

Hi,
I was just browsing the documentation and didn't see "counter monitor" collector mentioned anywhere. Even the man pages are only for callgraph and sampling collectors...
I was wondering if vtune for Linux supports counter monitor collector.
As I haven't installed the package yet, can't checkavailable collectorsthrough "vtl query -lc".

Thanks

jeffrey-gallagher (Intel)'s picture

Anybody else out there missing counter monitor?

Message Edited by MADjdgallag on 12-03-2003 10:42 AM

jeffrey-gallagher (Intel)'s picture

Great question Rostam.

In a word, no. The counter monitor functionality has not yet been added to the Linux product, and won't be added anytime too soon. FYI only.

cheers,

jdg

jeffrey-gallagher (Intel)'s picture

Intel Software NetworkCommunity Forums will be re-launching with a new look and a new platform with increased functionality and features. We will have a downtime on Wednesday, 11/19/03 from 5pm - 8pm PST. During this time, users will have read-only access to the forums.

If you have any questions or concerns, please feel free to contact us here.

Message Edited by intel.software.network.support on 12-01-2005 10:16 AM

Community Admin's picture

I have problems installing the beta on and AS 2.1 (kernel 2.4.18-e37, glibc 2.2.4-32.8).

The install program runs for several minutes (after selectin g 1, accepting licences and seeing "unpacking files ..." on the screen). The install program does not stop completly it uses a percent of cpu or so... (15 seconds in 20 minutes).
First run it got a segv after 20 minutes. Second attempt hasn't terminated yet.
There is no entries after last success in the install.log.
It has extracted a number of rpm-files

What does the install program due?
install the rpms, update some environement shell script and add some files to /etc/init.d?
Why the hell does it have to be done by a buggy install-program?

/Tommy

jeffrey-gallagher (Intel)'s picture

Hey Tommy,

Thanks for taking the time to report the bug you're seeing! Fact is, it's a new one to us around here, I haven't talk to anybody who's seen it before, so we're super glad you checked in.

To answer your first question, the install program installs the vtl software itself; a series of support services that vtl needs called EntireX; an analyzer kernel driver kit; and, if you choose to let it, install will also alter your boot and shutdown script directories to automatically insert the needed kernel sampling drivers on bootup (using insmod) so that vtl sampling will work. Of course, if you tell the install program NOT to edit your rc scripts, it won't, and you can manually insmod the drivers yourself.

Lots of things could be going on, let me suggest you try the following things:

1) If the server runs out of disk space in the installation filesystem (df -v) or runs out of processes during the installation, it could hang up the install itself, no system resources being available to report trouble or write to the log.

2) Is NFS or an otherwise networked filesystem in use on the server you're installing to? If so, try to install locally ONLY, and let's see if the problem changes or stays the same.

A typical install on a slow processor should take about 8 or 9 minutes, no patience required. Is there a second system in your area that you could start an install on just to compare results?

If none of these suggestions shed any light on your installation and the behavior you're seeing, please head on over to our Premier web site and report the failure exactly as you have it here.

Community Admin's picture

It installed on the third attempt in about 60 minutes.
I installed on another machine with almost the same setup (differs on patchlevels for none critical rpms, kernel, libc and so are identical).

About possible reasons:
1) 4.5GB free on / (which includes installation target /opt), so I doubt thats a problem

2) The source was placed on an nfs-mounted partition (from which I frequently install rpms without any problem). The target directory is local though.

If 2) is the reason: what magic does install do? Is it the IA32 code for working over nfs that emulates badly on IA64? (But shurely you use the running operating system functionality to access the nfs filesystem?)

/Tommy

jeffrey-gallagher (Intel)'s picture

Good detective work Tommy.

The issue with NFS is known, and is being worked on. For those folks who can install without it, using only local directories, we encourage it (for now).

FYI only, because you seem interested in the details, we've isolated the issue to a specific call in the code of the support services that vtl requires to operate.

Here's the short story: fcntl() calls don't work reliably on nfs volumes.

We intend to address this issue as quickly as possible, not waiting for the next release of our software. I think we're still talking first quarter of next year, perhaps at the earliest.

jdg

Hi,

I've been trying to get vtune working on a debian itanium2 system. I know it's not a supported system, but maybe someone can tell me what's going on anyway.

I've made all sorts of tricks to get it installed (which I can post on request), and I think I have everything covered in the install process. However, right now the error I get is this:
# /opt/intel/vtune/shared/bin/vtl
Error 0x800706ba: Unable to create the VTShell object. Make sure the Shell library is properly installed and registered.

Can anyone help me to get further, or should I just give up on debian?

jeffrey-gallagher (Intel)'s picture

Rodaz, I admire your spirit of adventure but my guess is that your chances of success are virtually zero. It might not be a worthwhile use of your time to continue. And the reason why follows.

The supporting services that allow vtl to work rely heavily on (and are validated to) very specific Red Hat and SuSE default releases and their subcomponents. This includes not just kernel versions to each and every digit, but also versions of glib, glibc, glibc++, and other libraries as well I think.

I don't know if Debian alters their kernel very much before releasing it with a given distribution, but you could always go to kernel.org and get a supported version, easy peasy. But, it's with the other bits that things get a little complicated.

I'd compare what you're attempting with trying to replace and adjust a new CRT for your computer monitor (not flat panel, obviously). You could spend hours replacing that tube and its flyback transformer, adjusting the color guns, all the other visual components (horizontal, vertical, purity, etc.) but there's still no guarantee you'll like the way the monitor looks when you're done. Clearly for the average user it's just easier to completely swap the monitor out, bam.

My advice: if you need vtl to get some optimizing done, stick to the exact releases mentioned in the release notes.

ALSO note though: if you're married to Debian itself, the Linux remote agents with VTune analyzer 7.0 may work for you. I'd recommend the 7.1 version currently in beta, which you may already know about if you read other forums here at IDS.

If you want to take a look, click here

cheers

jdg

Message Edited by intel.software.network.support on 12-09-2005 02:25 PM

jeffrey-gallagher (Intel)'s picture

Some of you may already be aware that the Borland Conference (BorCon) is being held next week in San Jose, CA, November 1-5. If you're going to be in the area, you should be sure to drop into the Intel booth anytime to say hello.

Or, if your schedule permits be sure to check out one or both(!) of these VTune events, both currently scheduled there for Tuesday, November 4:

TUESDAY NOVEMBER 4, 2003
10:00am -- 11:45am
Using VTune, Technical Presentation
Gary Carleton, Sr. VTune Consulting Engineer
Gary is giving a talk on using the VTune analyzer to tune applications built with C++Builder.

TUESDAY NOVEMBER 4, 2003
1:00pm -- 1:45pm
VTune User SHARE/Birds of a feather
Various VTune team members
Gary and I and other members of the VTune team will be in this 45-minute session where we can talk about anything and everything VTune, whatever's on your mind, kudos, suggestions for improvement, bugs, you name it. This is your chance to SHARE your ideas and feedback with the folks who know the software best, and our managers eat this stuff up for breakfast. Come tell us what's on your mind -- we'll be all ears!

If you are able to drop in, be sure to mention you read about the session in this Forum!

It's all taking place in the San Jose Convention Center, but for more information on BorCon, you should visit their site
here.

Hi,

Just downloaded the beta 2.0. I was going through the VTLRELEASENOTES.htm. It mentioned that "Do NOT install this software on a Linux* system that currently has the VTune Performance Analyzer Linux* Remote Agent on it". Unfortunately, that is what I wanted to do...

I want to install the 2.0, and see if that helps me collect traces better than the remote agent (with 7.0 on a windows m/c). I dont want to mess too much with my current setup as I use the remote agent to get all data.

Can you suggest a way by which I can safely test this on the same machine and see if it will of use? Or atleast, can you tell me what happens if I unload the remote agent driver module and then do the 2.0 installation?

After installation, I want to make sure I can still use remote agent if I want to collect some serious data/or if I think I cannot use 2.0 for a particular case.

Thanks,
JP

jeffrey-gallagher (Intel)'s picture

Hey jayaprakashp,

Do not install VTune analyzer 2.0 for Linux beta on any server with a functioning VTServer, VTune 7.0 on it (ditto for 7.1 beta, as well). There is currently no supported solution for what you want to do.

You make a great case for us changing our current decision on this; however, that certainly won't be happening in time for this beta program.

Note that the installation of VTune analyzer for Linux (vtl, either 1.1 or the 2.0 beta) will definitely corrupt the installation of the VTune analyzer (VTServer, either 7.0 or the 7.1 beta). Neither would then work, you'll have bits and pieces of both and not one happy camper between them.

To correct, from there you'd end up having to uninstall both, and then reinstall the one you wanted to use (in your case, the 7.0 Linux Remote Agent).

Depending on how you chose to uninstall, your project files could still be around, but it's easy to pick the wrong choice in the heat of an uninstall. I'd urge you strongly to avoid this experiment completely.

But, I am hearing what you're saying, that you'd like to be able to do it, and I'm forwarding that information on.

I won't explain our decision here except to say at current prices -- $699 USD a pop -- the VTune team never envisioned that users would buy both versions of the software and install on the same server.

The case you're describing, kicking the beta tires, is certainly one exception. Another might be in the classroom, with limited hardware, and where both software versions might be taught.

Keep these great comments coming! But for now, you'll have to find another server to kick the tires.

cheers

jdg

-hotspot-function doesn't work for listing samples per kernel function:

"
$ vtl view -hotspot-function -mn vmlinux
VTune Performance Analyzer 2.0 for Linux*
Copyright (C) 2000-2003 Intel Corporation. All rights reserved.

Event Summary
Clockticks
120830 = Samples collected due to this event
1779000 = Sample after value used during collection
214956570000 = Total events (samples*SAV)

HotSpot View grouped by function (all values in decimal except where indicated)

No symbols were found
"

Am I doing something wrong, or is this a known problem?

OProfile (http://oprofile.sf.net) manages to do this with the vmlinux file as input, so it should be doable.

Cheers //Johan

I've had luck with
vtl view -hotspot-function -mn vmlinux -sd /usr/src/linux
after recompiling the kernel with "-g".

It seems like all I should have needed was a vmlinux that was not stripped, but I didn't check that before recompiling.

The quick start guide gives info on howto recompile with "-g", but if your vmlinux file has symbols (nm /usr/src/vmlinux), try the above command line. If it works, please let me know. If it doesn,t try recompiling the kernel with "-g".

That worked. Thanks!

jeffrey-gallagher (Intel)'s picture

Excellent.

jeffrey-gallagher (Intel)'s picture

GREAT question, Johan. I'm asking the relevant engineers.

cheers

jdg

I've done a profiling run using the event type "Branches Retired". Here's an assembly view (slightly censored to fit) of the (start of the) result:

"
$ vtl view -code -mn /localhome/johan/clients/johan/ariane/build/linux32-j2se1.4.2/release/libjvm.so -fn mmParScanObject -asm|head -20
VTune Performance Analyzer 2.0 for Linux*
Copyright (C) 2000-2003 Intel Corporation. All rights reserved.

Legend
Ev1 = Branches Retired samples
Ev1 ICACHE Flag Decoder... Source
1 0x72 subl $0x2c, %esp
0 0x72 pushl %ebp
0 0x72 pushl %edi
1 0x72 pushl %esi
0 0x72 0x6a pushl %ebx
1 0x72 0x6e movl 0x40(%esp), %ebp
1 0x72 0x70 movl (%ebp), %eax
45 0x72 movl 0x4c(%eax), %edx
1 0x72 0x6a movl %edx, 0x28(%esp)
1 0x72 0x6e testb $0x1, %dl
0 0x72 0x70 je B_aad17
0 0x72 0x6a addl $-4, %esp
0 0x72 0x70 movl 0x48(%esp), %edx
0 0x71 0x6a pushl %edx
"

Now, how come a movl instruction gets hit by the "Branches Retired" event? I thought that one should trigger on branch instructions only.

Cheers //Johan

jeffrey-gallagher (Intel)'s picture

Walles,

This is a great question. If you find that you are thirsty for more information on this topic in general, don't hesitate to examine the Reference Manual:

/opt/intel/vtune/doc/reference/index.htm

BUT, to answer your question here and now, the thing to remember is, "branches retired" is not a "precise" event. This means there will be some event skid after the event occurs, and before the VTune interrupt stops the processor (which is just a part of the way the VTune sampling methodology works).

In your example, it looks like the event skid is about 8-10 instructions -- the top of the code looks like the start of a procedure (setting up the stack frame?) -- so the top instruction or the one before it is probably the target of a branch that is causing the event.

ALSO, let's take a step back here a second and think about what branching does: it jumps. The skid on branches is accordingly very unpredictable. It could even be MORE than 8-10, it could be a HUGE number, depending how far you are jumping.

In a really bad case, you might not get the vtune interrupt until the processor executes the instruction AFTER the branch. By then the IP will be where the code branched to, which could be very, very far from where you were.

Take a look at the reference guide: I bet you'll enjoy it.

cheers

jdg

jeffrey-gallagher (Intel)'s picture

DBricker asked a great question, and I want everybody to see it.

Jeff,

Does VTune Analyzer 2.0 for Linux* supports data collection for Java applications?

Thank you
Daniel

jeffrey-gallagher (Intel)'s picture

Excellent question, DB!

Absolutely JAVA support is in there now. Read about all the new features:

WHAT'S NEW

Support for several new Linux* distributions, including their default versions of kernel, glibc, libstdc++, and glib, has been added to VTune analyzer 2.0 for Linux*. The new Itanium processor distribution supports Red Hat* Linux* WS 2.1 and ES 2.1. Also, a newly supported IA-32 Linux* distribution has been added, SuSE* Linux* 8.2.

Java* application profiling on supported IA-32 Linux* distributions has been added. Running Java* applications with the VTune analyzer with either sampling or call graph profiling options enabled obtains results similar to native VTune profiling. Java* Virtual Machines (JVMs) from BEA*, IBM*, and Sun* are supported. Note that call graph profiling tracks Java* calls only, and there is currently no support for mixed Java* and native calls in this release. Additionally, there is currently no support for profiling Java* applications on Itanium processor versions of Linux*.

NOTE: See the Release Notes (included with the beta software) for a full list of supported Linux* distributions and Linux* JVMs.

Several differences between this software and the VTune analyzer v. 1.1 are also worth noting:

-- VTune analyzer now provides call graph output on the flow of control of Itanium processor Linux* applications in addition to event-based sampling.

-- The new Getting Started tutorial teaches the basic concepts and functionality of the VTune analyzer for Linux*.

-- Source View is now available.

-- VTune analyzer 2.0 can now import other sampling data files (*.tb5) for viewing.

Using VTL2.0 rc2, build 235:

"
$ vtl view a1::r1 -code -mn libjvm.so -fn mmParScanObject
VTune Performance Analyzer 2.0 for Linux*
Copyright (C) 2000-2003 Intel Corporation. All rights reserved.

Wed Sep 24 15:56:17 2003 ERROR: critical error when initalizing source view data provider:Unexpected problem has occured.
An error occurred during navigation.
"

This error message isn't very helpful. What does it mean, and what can I do about it?

Cheers //Johan

jeffrey-gallagher (Intel)'s picture

Great question Johan. I haven't seen it before myself, I'll pass it over to the engineers and see what they say.

Stay tuned.

cheers

jdg

jeffrey-gallagher (Intel)'s picture

Hey Johan,

Here's the reply from engineering:

the error is probably due to not specifying full path to the le. the error message was replaced with an error message saying the module can not be located.

If that fixes the problem, let us know back here.

IF NOT, you should:

1) pack up the project that shows the problem (will create a file ending in .vxp)

2) go to premier.intel.com, log in, and open a case

3) attach the .vxp file to the case

Our engineering folks eat this stuff for breakfast...

:)

cheers

jdg

Pages

Login to leave a comment.