INFO: Generating MadAnalysis5 default cards tailored to this process
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MadAnalysis 5 |
Fix Released
|
High
|
Benjamin Fuks |
Bug Description
Hi guys,
Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
Is it possible to fix this? 6 seconds to create some default cards is much too long.
(To be honest, I want to remove MA5 from my system just due to that)
Thanks,
Olivier
Benjamin Fuks (fuks) wrote : | #1 |
Changed in madanalysis5: | |
importance: | Undecided → High |
status: | New → In Progress |
assignee: | nobody → Benjamin Fuks (fuks) |
milestone: | none → v1.6 |
Olivier Mattelaer (olivier-mattelaer) wrote : Re: [Bug 1692480] Re: INFO: Generating MadAnalysis5 default cards tailored to this process | #2 |
Hi Benj,
I still have this:
INFO: Generating MadAnalysis5 default cards tailored to this process
INFO: Cards created in 5.43s
Which sounds quite slow.
Cheers,
Olivier
> On 22 May 2017, at 23:56, Benjamin Fuks <email address hidden> wrote:
>
> Hi Olivier,
>
> You were right and I have found the single guilty line triggering that
> compilation. This should now be fixed. Do you mind confirming it (please
> reinstall a fresh ma5; the tarball has been updated)? Thanks in advance!
>
> Cheers,
>
> Benj
>
> ** Changed in: madanalysis5
> Importance: Undecided => High
>
> ** Changed in: madanalysis5
> Status: New => In Progress
>
> ** Changed in: madanalysis5
> Assignee: (unassigned) => Benjamin Fuks (fuks)
>
> ** Changed in: madanalysis5
> Milestone: None => v1.6
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https:/
>
> Title:
> INFO: Generating MadAnalysis5 default cards tailored to this process
>
> Status in MadAnalysis 5:
> In Progress
>
> Bug description:
> Hi guys,
>
> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
> Is it possible to fix this? 6 seconds to create some default cards is much too long.
> (To be honest, I want to remove MA5 from my system just due to that)
>
> Thanks,
>
> Olivier
>
> To manage notifications about this bug go to:
> https:/
Benjamin Fuks (fuks) wrote : | #3 |
Hi Olivier,
This is weird and seems computer dependent. It is much faster on my laptop.
Do you mind doing the following:
- setting the log level to the debug mode when calling the interpreter (in common_interface and export_v4);
- dumping the screen output into a file or on the screen (removing the std_channel redirection) so that I could have a look.
The compilation is triggered by the comparison of architecture information of the current call and previous call of the interpreter. There must be something different in your case, and I need to find out what so that I could understand and fix it. Thanks in advance!
Cheers,
Benj
PS: normally, you should see the O(10) compilation steps on the screen output.
> On 23 May 2017, at 01:33 , Olivier Mattelaer <email address hidden> wrote:
>
> Hi Benj,
>
> I still have this:
> INFO: Generating MadAnalysis5 default cards tailored to this process
> INFO: Cards created in 5.43s
>
> Which sounds quite slow.
>
> Cheers,
>
> Olivier
>
>
>> On 22 May 2017, at 23:56, Benjamin Fuks <email address hidden> wrote:
>>
>> Hi Olivier,
>>
>> You were right and I have found the single guilty line triggering that
>> compilation. This should now be fixed. Do you mind confirming it (please
>> reinstall a fresh ma5; the tarball has been updated)? Thanks in advance!
>>
>> Cheers,
>>
>> Benj
>>
>> ** Changed in: madanalysis5
>> Importance: Undecided => High
>>
>> ** Changed in: madanalysis5
>> Status: New => In Progress
>>
>> ** Changed in: madanalysis5
>> Assignee: (unassigned) => Benjamin Fuks (fuks)
>>
>> ** Changed in: madanalysis5
>> Milestone: None => v1.6
>>
>> --
>> You received this bug notification because you are subscribed to the bug
>> report.
>> https:/
>>
>> Title:
>> INFO: Generating MadAnalysis5 default cards tailored to this process
>>
>> Status in MadAnalysis 5:
>> In Progress
>>
>> Bug description:
>> Hi guys,
>>
>> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
>> Is it possible to fix this? 6 seconds to create some default cards is much too long.
>> (To be honest, I want to remove MA5 from my system just due to that)
>>
>> Thanks,
>>
>> Olivier
>>
>> To manage notifications about this bug go to:
>> https:/
>
> --
> You received this bug notification because you are a bug assignee.
> https:/
>
> Title:
> INFO: Generating MadAnalysis5 default cards tailored to this process
>
> Status in MadAnalysis 5:
> In Progress
>
> Bug description:
> Hi guys,
>
> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
> Is it possible to fix this? 6 seconds to create some default cards is much too long.
> (To be honest, I want to remove MA5 from my system just due to that)
>
> Thanks,
>
> Olivier
>
> To manage notifications about this bug go to:
> https:/
Olivier Mattelaer (olivier-mattelaer) wrote : | #4 |
Hi,
Sure, here is the detailed information:
Cheers,
Olivier
INFO: Generating MadAnalysis5 default cards tailored to this process
MA5: Platform: Darwin 16.5.0 [MAC/OSX mode]
MA5-DEBUG:
MA5-DEBUG: Machine - Cross platform information
MA5-DEBUG: Machine type: x86_64
MA5-DEBUG: Processor name: i386
MA5-DEBUG: Platform: Darwin-
MA5-DEBUG: Platform release: 16.5.0
MA5-DEBUG: System: Darwin
MA5-DEBUG: Node: mac-1q0-072.local
MA5-DEBUG: Number of cores: 8
MA5-DEBUG:
MA5-DEBUG: Machine - OS-specific information
MA5-DEBUG: Java version: ('', '', ('', '', ''), ('', '', ''))
MA5-DEBUG: Windows version: ('', '', '', '')
MA5-DEBUG: Mac Os version: ('10.12.4', ('', '', ''), 'x86_64')
MA5-DEBUG: Unix distribution: Darwin-
MA5-DEBUG:
MA5: Reading user settings ...
MA5-DEBUG: Opening the file: /Users/
MA5-DEBUG: Lines to interpret:
MA5-DEBUG: - zlib_includes = /Users/
MA5-DEBUG: - zlib_libs = /Users/
MA5-DEBUG: - pdflatex_veto = 1
MA5-DEBUG: - latex_veto = 1
MA5-DEBUG: Closing the file: /Users/
MA5-DEBUG:
MA5-DEBUG: User
MA5-DEBUG: User name: omattelaer
MA5-DEBUG: User ID: 503
MA5-DEBUG: Expanding folder ~/: /Users/omattelaer/
MA5-DEBUG: Variable $USER: omattelaer
MA5-DEBUG: Variable $USERNAME:
MA5-DEBUG: Variable $LNAME:
MA5-DEBUG: Variable $LOGNAME: omattelaer
MA5-DEBUG: Variable $HOME: /Users/omattelaer
MA5-DEBUG: Variable $HOMEDRIVE:
MA5-DEBUG: Variable $posix:
MA5-DEBUG: Variable $HOMEPATH:
MA5-DEBUG:
MA5-DEBUG: Web access
MA5-DEBUG: enable
MA5-DEBUG:
MA5-DEBUG: Temporary folder
MA5-DEBUG: Variable $TMPDIR: /var/folders/
MA5-DEBUG: Check if the folder /var/folders/
MA5-DEBUG: -> found
MA5-DEBUG: Variable $TMP:
MA5-DEBUG: Variable $TEMP:
MA5-DEBUG: temporary folder will be used for MA5: /var/folders/
MA5-DEBUG:
MA5-DEBUG: Download dir
MA5-DEBUG: Check if the folder /var/folders/
MA5-DEBUG: -> found
MA5-DEBUG: download folder will be used for MA5: /var/folders/
MA5-DEBUG:
MA5-DEBUG: Text editor
MA5-DEBUG: Look for the global variable $EDITOR ...
MA5-DEBUG: -> variable $EDITOR found : emacs
MA5-DEBUG:
MA5: Checking mandatory packages:
MA5-DEBUG: -------
MA5-DEBUG: Detect package python
MA5-DEBUG: Try to detect automatically the package ...
MA5-DEBUG: Extract more informations related to the package ...
MA5-DEBUG:
MA5-DEBUG: Python release: 2.7.10
MA5-DEBUG: Python build: ('default', 'Feb ...
Benjamin Fuks (fuks) wrote : Re: [Bug 1692480] INFO: Generating MadAnalysis5 default cards tailored to this process | #5 |
Hi Olivier,
The time is basically being spent on checking whether the ma5 interpreter can run properly run here. There is no compilation. This is of course not necessary at this stage. I suggest to implement a way to call the interpreter with all the checks being bypassed that you will then be able to use within aMC. What do yo tihnk?
Cheers,
B>
> On 23 May 2017, at 12:41 , Olivier Mattelaer <email address hidden> wrote:
>
> Hi,
>
> Sure, here is the detailed information:
>
> Cheers,
>
> Olivier
>
> INFO: Generating MadAnalysis5 default cards tailored to this process
> MA5: Platform: Darwin 16.5.0 [MAC/OSX mode]
> MA5-DEBUG:
> MA5-DEBUG: Machine - Cross platform information
> MA5-DEBUG: Machine type: x86_64
> MA5-DEBUG: Processor name: i386
> MA5-DEBUG: Platform: Darwin-
> MA5-DEBUG: Platform release: 16.5.0
> MA5-DEBUG: System: Darwin
> MA5-DEBUG: Node: mac-1q0-072.local
> MA5-DEBUG: Number of cores: 8
> MA5-DEBUG:
> MA5-DEBUG: Machine - OS-specific information
> MA5-DEBUG: Java version: ('', '', ('', '', ''), ('', '', ''))
> MA5-DEBUG: Windows version: ('', '', '', '')
> MA5-DEBUG: Mac Os version: ('10.12.4', ('', '', ''), 'x86_64')
> MA5-DEBUG: Unix distribution: Darwin-
> MA5-DEBUG:
> MA5: Reading user settings ...
> MA5-DEBUG: Opening the file: /Users/
> MA5-DEBUG: Lines to interpret:
> MA5-DEBUG: - zlib_includes = /Users/
> MA5-DEBUG: - zlib_libs = /Users/
> MA5-DEBUG: - pdflatex_veto = 1
> MA5-DEBUG: - latex_veto = 1
> MA5-DEBUG: Closing the file: /Users/
> MA5-DEBUG:
> MA5-DEBUG: User
> MA5-DEBUG: User name: omattelaer
> MA5-DEBUG: User ID: 503
> MA5-DEBUG: Expanding folder ~/: /Users/omattelaer/
> MA5-DEBUG: Variable $USER: omattelaer
> MA5-DEBUG: Variable $USERNAME:
> MA5-DEBUG: Variable $LNAME:
> MA5-DEBUG: Variable $LOGNAME: omattelaer
> MA5-DEBUG: Variable $HOME: /Users/omattelaer
> MA5-DEBUG: Variable $HOMEDRIVE:
> MA5-DEBUG: Variable $posix:
> MA5-DEBUG: Variable $HOMEPATH:
> MA5-DEBUG:
> MA5-DEBUG: Web access
> MA5-DEBUG: enable
> MA5-DEBUG:
> MA5-DEBUG: Temporary folder
> MA5-DEBUG: Variable $TMPDIR: /var/folders/
> MA5-DEBUG: Check if the folder /var/folders/
> MA5-DEBUG: -> found
> MA5-DEBUG: Variable $TMP:
> MA5-DEBUG: Variable $TEMP:
> MA5-DEBUG: temporary folder will be used for MA5: /var/folders/
> MA5-DEBUG:
> MA5-DEBUG: Download dir
> MA5-DEBUG: Check if the folder /var/folders/
> MA5-DEBUG: -> found
> MA5-DEBUG: download folder will be used for MA5...
Olivier Mattelaer (olivier-mattelaer) wrote : | #6 |
Hi,
Whatever makes the code not spending 5s at initialisation sounds a good idea.
I have add a bunch of print statement inside you code and found the responsible line for the slow down of the init:
precompiler.
That line takes a full 5s.
So I have just commented it and now this makes the code fast enough (0.4s). I have no idea what that lines does, but that's the bottleneck.
Cheers,
Olivier
> On 23 May 2017, at 15:19, Benjamin Fuks <email address hidden> wrote:
>
> Hi Olivier,
>
> The time is basically being spent on checking whether the ma5
> interpreter can run properly run here. There is no compilation. This is
> of course not necessary at this stage. I suggest to implement a way to
> call the interpreter with all the checks being bypassed that you will
> then be able to use within aMC. What do yo tihnk?
>
> Cheers,
>
> B>
>
>
>> On 23 May 2017, at 12:41 , Olivier Mattelaer <email address hidden> wrote:
>>
>> Hi,
>>
>> Sure, here is the detailed information:
>>
>> Cheers,
>>
>> Olivier
>>
>> INFO: Generating MadAnalysis5 default cards tailored to this process
>> MA5: Platform: Darwin 16.5.0 [MAC/OSX mode]
>> MA5-DEBUG:
>> MA5-DEBUG: Machine - Cross platform information
>> MA5-DEBUG: Machine type: x86_64
>> MA5-DEBUG: Processor name: i386
>> MA5-DEBUG: Platform: Darwin-
>> MA5-DEBUG: Platform release: 16.5.0
>> MA5-DEBUG: System: Darwin
>> MA5-DEBUG: Node: mac-1q0-072.local
>> MA5-DEBUG: Number of cores: 8
>> MA5-DEBUG:
>> MA5-DEBUG: Machine - OS-specific information
>> MA5-DEBUG: Java version: ('', '', ('', '', ''), ('', '', ''))
>> MA5-DEBUG: Windows version: ('', '', '', '')
>> MA5-DEBUG: Mac Os version: ('10.12.4', ('', '', ''), 'x86_64')
>> MA5-DEBUG: Unix distribution: Darwin-
>> MA5-DEBUG:
>> MA5: Reading user settings ...
>> MA5-DEBUG: Opening the file: /Users/
>> MA5-DEBUG: Lines to interpret:
>> MA5-DEBUG: - zlib_includes = /Users/
>> MA5-DEBUG: - zlib_libs = /Users/
>> MA5-DEBUG: - pdflatex_veto = 1
>> MA5-DEBUG: - latex_veto = 1
>> MA5-DEBUG: Closing the file: /Users/
>> MA5-DEBUG:
>> MA5-DEBUG: User
>> MA5-DEBUG: User name: omattelaer
>> MA5-DEBUG: User ID: 503
>> MA5-DEBUG: Expanding folder ~/: /Users/omattelaer/
>> MA5-DEBUG: Variable $USER: omattelaer
>> MA5-DEBUG: Variable $USERNAME:
>> MA5-DEBUG: Variable $LNAME:
>> MA5-DEBUG: Variable $LOGNAME: omattelaer
>> MA5-DEBUG: Variable $HOME: /Users/omattelaer
>> MA5-DEBUG: Variable $HOMEDRIVE:
>> MA5-DEBUG: Variable $posix:...
Benjamin Fuks (fuks) wrote : | #7 |
Hi Olivier,
There was a feature that I had included some time ago and that would be perfect here. The ma5 interpreter allows for the option "no_compilation
Cheers,
Benj
> On 23 May 2017, at 14:58 , Olivier Mattelaer <email address hidden> wrote:
>
> Hi,
>
> Whatever makes the code not spending 5s at initialisation sounds a good
> idea.
>
> I have add a bunch of print statement inside you code and found the responsible line for the slow down of the init:
> precompiler.
>
> That line takes a full 5s.
>
> So I have just commented it and now this makes the code fast enough
> (0.4s). I have no idea what that lines does, but that's the bottleneck.
>
> Cheers,
>
> Olivier
>
>> On 23 May 2017, at 15:19, Benjamin Fuks <<email address hidden> <mailto:<email address hidden>>> wrote:
>>
>> Hi Olivier,
>>
>> The time is basically being spent on checking whether the ma5
>> interpreter can run properly run here. There is no compilation. This is
>> of course not necessary at this stage. I suggest to implement a way to
>> call the interpreter with all the checks being bypassed that you will
>> then be able to use within aMC. What do yo tihnk?
>>
>> Cheers,
>>
>> B>
>>
>>
>>> On 23 May 2017, at 12:41 , Olivier Mattelaer <email address hidden> wrote:
>>>
>>> Hi,
>>>
>>> Sure, here is the detailed information:
>>>
>>> Cheers,
>>>
>>> Olivier
>>>
>>> INFO: Generating MadAnalysis5 default cards tailored to this process
>>> MA5: Platform: Darwin 16.5.0 [MAC/OSX mode]
>>> MA5-DEBUG:
>>> MA5-DEBUG: Machine - Cross platform information
>>> MA5-DEBUG: Machine type: x86_64
>>> MA5-DEBUG: Processor name: i386
>>> MA5-DEBUG: Platform: Darwin-
>>> MA5-DEBUG: Platform release: 16.5.0
>>> MA5-DEBUG: System: Darwin
>>> MA5-DEBUG: Node: mac-1q0-072.local
>>> MA5-DEBUG: Number of cores: 8
>>> MA5-DEBUG:
>>> MA5-DEBUG: Machine - OS-specific information
>>> MA5-DEBUG: Java version: ('', '', ('', '', ''), ('', '', ''))
>>> MA5-DEBUG: Windows version: ('', '', '', '')
>>> MA5-DEBUG: Mac Os version: ('10.12.4', ('', '', ''), 'x86_64')
>>> MA5-DEBUG: Unix distribution: Darwin-
>>> MA5-DEBUG:
>>> MA5: Reading user settings ...
>>> MA5-DEBUG: Opening the file: /Users/
>>> MA5-DEBUG: Lines to interpret:
>>> MA5-DEBUG: - zlib_includes = /Users/
>>> MA5-DEBUG: - zlib_libs = /Users/
>>> MA5-DEBUG: - pdflatex_veto = 1
>>> MA5-DEBUG: - latex_veto = 1
>>> MA5-DEBUG: Closing the f...
Olivier Mattelaer (olivier-mattelaer) wrote : | #8 |
Seems to work.
Can I always put it on True? Or should it be put on True only for the card creation?
Cheers,
Olivier
On 24 May 2017, at 16:39, Benjamin Fuks <<email address hidden>
"no_compilation
Benjamin Fuks (fuks) wrote : | #9 |
Hi Olivier,
Only for the card creation, as for the analysis part a compilation is necessary (a c++ code is generated and must be compiled).
Cheers,
Benj
> On 24 May 2017, at 17:33 , Olivier Mattelaer <email address hidden> wrote:
>
> Seems to work.
>
> Can I always put it on True? Or should it be put on True only for the
> card creation?
>
> Cheers,
>
> Olivier
>
> On 24 May 2017, at 16:39, Benjamin Fuks
> <<email address hidden>
>
> "no_compilation
>
> --
> You received this bug notification because you are a bug assignee.
> https:/
>
> Title:
> INFO: Generating MadAnalysis5 default cards tailored to this process
>
> Status in MadAnalysis 5:
> In Progress
>
> Bug description:
> Hi guys,
>
> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
> Is it possible to fix this? 6 seconds to create some default cards is much too long.
> (To be honest, I want to remove MA5 from my system just due to that)
>
> Thanks,
>
> Olivier
>
> To manage notifications about this bug go to:
> https:/
Olivier Mattelaer (olivier-mattelaer) wrote : | #10 |
Ok thanks,
You can consider this as fixed then. (2.5.5 is released with the associated modification)
Thanks,
Olivier
> On 25 May 2017, at 02:13, Benjamin Fuks <email address hidden> wrote:
>
> Hi Olivier,
>
> Only for the card creation, as for the analysis part a compilation is
> necessary (a c++ code is generated and must be compiled).
>
> Cheers,
>
> Benj
>
>
>> On 24 May 2017, at 17:33 , Olivier Mattelaer <email address hidden> wrote:
>>
>> Seems to work.
>>
>> Can I always put it on True? Or should it be put on True only for the
>> card creation?
>>
>> Cheers,
>>
>> Olivier
>>
>> On 24 May 2017, at 16:39, Benjamin Fuks
>> <<email address hidden>
>>
>> "no_compilation
>>
>> --
>> You received this bug notification because you are a bug assignee.
>> https:/
>>
>> Title:
>> INFO: Generating MadAnalysis5 default cards tailored to this process
>>
>> Status in MadAnalysis 5:
>> In Progress
>>
>> Bug description:
>> Hi guys,
>>
>> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
>> Is it possible to fix this? 6 seconds to create some default cards is much too long.
>> (To be honest, I want to remove MA5 from my system just due to that)
>>
>> Thanks,
>>
>> Olivier
>>
>> To manage notifications about this bug go to:
>> https:/
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https:/
>
> Title:
> INFO: Generating MadAnalysis5 default cards tailored to this process
>
> Status in MadAnalysis 5:
> In Progress
>
> Bug description:
> Hi guys,
>
> Looks like MA5 is again systematically recompiling, which slows down significantly MG5...
> Is it possible to fix this? 6 seconds to create some default cards is much too long.
> (To be honest, I want to remove MA5 from my system just due to that)
>
> Thanks,
>
> Olivier
>
> To manage notifications about this bug go to:
> https:/
Changed in madanalysis5: | |
status: | In Progress → Fix Released |
Hi Olivier,
You were right and I have found the single guilty line triggering that compilation. This should now be fixed. Do you mind confirming it (please reinstall a fresh ma5; the tarball has been updated)? Thanks in advance!
Cheers,
Benj