Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Trevor,


Thanks for looking into this bug. It isn't holding back my research, but 
tracking it down is a fun distraction from applying to post-doc positions.



I am building Trilinos with Trilinos_ENABLE_ALL_PACKAGES=ON and only a 
few parts of SEACAS are automatically disabled. I disabled the TPLs for 
Matio, GLM, and Netcdf because I don't need them. I think it's safe to 
say that I have all the modules.


I've been pulling from the GitHub repository, and I noticed that 
Trilinos 11.10.2 isn't available there, only on the Trilinos site: 
https://trilinos.org/download/previous-releases/download-11-10/ But at 
this point, I doubt that this is related to the issue I am experiencing.



As a side note, If anyone wants help compiling Trilinos, I am getting 
very good at it.



Best,

Mike


On 10/12/2016 10:49 AM, Keller, Trevor (Fed) wrote:


Hi Mike,


Thanks for testing all those combinations! Sorry none worked though. 
For me, the specific combination of Trilinos 11.10.2 with swig 2.0.8 
closed the memory leak back in March. I'm surprised the latest 
versions (6 months later) still create a leak. It makes me suspect 
we're doing something wrong, perhaps neglecting some important module 
in Trilinos? I'll try building the latest release and, if the leak 
persists on my machine as well, try a bit harder to track it down. If 
nothing else, I'll file a bug report and try to get more developers' 
eyes on this.



My software environment is very similar to yours:Boost v1.55, OpenMPI 
1.6.5, and gcc 4.9.2.



With regards,

Trevor



Trevor Keller, Ph.D.
Materials Science and Engineering Division
National Institute of Standards and Technology
100 Bureau Dr. MS 8550; Gaithersburg, MD 20899
Office: 223/A131 or (301) 975-2889



*From:* fipy-boun...@nist.gov  on behalf of 
Michael Waters 

*Sent:* Tuesday, October 11, 2016 4:55:51 PM
*To:* FIPY
*Subject:* Re: Memory Leaks with Trilinos

Hi Trevor,

My mystery deepens. Today, I've compiled and tried the following 
combinations of Trilinos/Swig:


12.8.1/3.0.2

11.14.3/3.0.2

11.12.1/3.0.2

11.10.1/2.0.8

11.2.5/2.0.8

All of these combinations leak memory at about the same rate. I am 
using Boost 1.54, Openmpi 1.6.5, and GCC 4.8.4. Are you using similar?


Thanks,

-Mike


On 03/30/2016 05:06 PM, Keller, Trevor (Fed) wrote:

Mike & Jon,

Running 25 steps on Debian wheezy using PyTrilinos 11.10.2 and swig 2.0.8, 
monitoring with memory-profiler, I do not see a memory leak (see attachment): 
the simulation is fairly steady at 8GB RAM. The same code, but using PyTrilinos 
12.1.0, ramps up to 16GB in the same simulation time.

Later revisions of Trilinos 12 remain defiant (though, compiling swig-3.0.8 has 
improved matters). If one of them compiles and works, I will let you know. If 
not, if possible, Mike, can you test the next incremental version (11.10.2)?

Trevor



From:fipy-boun...@nist.govon behalf of Michael 
Waters
Sent: Wednesday, March 30, 2016 4:36 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos

Hi Jon,

I just compiled an old version of swig (2.0.8) and compiled Trilinos
(11.10.1) against that. Sadly, I am still having the leak.

I am out of ideas for the day... and should be looking for a post doc
anyway.

Thanks,
-mike


On 3/30/16 3:32 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

No worries. If building trilinos doesn't blindside you with something 
unexpected and unpleasant, you're not doing it right.

I have a conda recipe 
athttps://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
 that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
haven't tried to adjust it to 12.x, yet.


On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:


Hi Jon,

I was just reviewing my version of Trilinos 11.10 and discovered that there is 
no way that I compiled it last night after exercising. It has unsatisfied 
dependencies on my machine. So I must apologize, I must have been more tired 
than I thought.

Sorry for the error!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:

It looked to me like steps and accuracy were the way to do it, but my runs 
finish in one step, so I was confused. When I change to accuracy = 10.0**-6, it 
takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED time 
is because I put my laptop to sleep for awhile, but VSIZE is rock-steady).

The fact that things never (or slowly) converge for you and Trevor, in addition 
to the leak, makes me wonder if Trilinos seriously broke something between 11.x 
and 12.x. Trevor's been struggling to build 12.4. I'll try to find time to do 
the same.

In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:


When I did my testing and made those graph

Re: Memory Leaks with Trilinos

[Keller, Trevor (Fed)]

Hi Mike,


Thanks for testing all those combinations! Sorry none worked though. For me, 
the specific combination of Trilinos 11.10.2 with swig 2.0.8 closed the memory 
leak back in March. I'm surprised the latest versions (6 months later) still 
create a leak. It makes me suspect we're doing something wrong, perhaps 
neglecting some important module in Trilinos? I'll try building the latest 
release and, if the leak persists on my machine as well, try a bit harder to 
track it down. If nothing else, I'll file a bug report and try to get more 
developers' eyes on this.


My software environment is very similar to yours: Boost v1.55, OpenMPI 1.6.5, 
and gcc 4.9.2.


With regards,

Trevor



Trevor Keller, Ph.D.
Materials Science and Engineering Division
National Institute of Standards and Technology
100 Bureau Dr. MS 8550; Gaithersburg, MD 20899
Office: 223/A131 or (301) 975-2889



From: fipy-boun...@nist.gov  on behalf of Michael Waters 

Sent: Tuesday, October 11, 2016 4:55:51 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos


Hi Trevor,

My mystery deepens. Today, I've compiled and tried the following combinations 
of Trilinos/Swig:

12.8.1/3.0.2

11.14.3/3.0.2

11.12.1/3.0.2

11.10.1/2.0.8

11.2.5/2.0.8

All of these combinations leak memory at about the same rate. I am using Boost 
1.54, Openmpi 1.6.5, and GCC 4.8.4. Are you using similar?

Thanks,

-Mike

On 03/30/2016 05:06 PM, Keller, Trevor (Fed) wrote:

Mike & Jon,

Running 25 steps on Debian wheezy using PyTrilinos 11.10.2 and swig 2.0.8, 
monitoring with memory-profiler, I do not see a memory leak (see attachment): 
the simulation is fairly steady at 8GB RAM. The same code, but using PyTrilinos 
12.1.0, ramps up to 16GB in the same simulation time.

Later revisions of Trilinos 12 remain defiant (though, compiling swig-3.0.8 has 
improved matters). If one of them compiles and works, I will let you know. If 
not, if possible, Mike, can you test the next incremental version (11.10.2)?

Trevor



From: fipy-boun...@nist.gov<mailto:fipy-boun...@nist.gov> 
<mailto:fipy-boun...@nist.gov> on behalf of Michael 
Waters <mailto:waters.mik...@gmail.com>
Sent: Wednesday, March 30, 2016 4:36 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos

Hi Jon,

I just compiled an old version of swig (2.0.8) and compiled Trilinos
(11.10.1) against that. Sadly, I am still having the leak.

I am out of ideas for the day... and should be looking for a post doc
anyway.

Thanks,
-mike


On 3/30/16 3:32 PM, Guyer, Jonathan E. Dr. (Fed) wrote:


No worries. If building trilinos doesn't blindside you with something 
unexpected and unpleasant, you're not doing it right.

I have a conda recipe at 
https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
haven't tried to adjust it to 12.x, yet.


On Mar 30, 2016, at 2:42 PM, Michael Waters 
<mailto:waters.mik...@gmail.com> wrote:



Hi Jon,

I was just reviewing my version of Trilinos 11.10 and discovered that there is 
no way that I compiled it last night after exercising. It has unsatisfied 
dependencies on my machine. So I must apologize, I must have been more tired 
than I thought.

Sorry for the error!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:


It looked to me like steps and accuracy were the way to do it, but my runs 
finish in one step, so I was confused. When I change to accuracy = 10.0**-6, it 
takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED time 
is because I put my laptop to sleep for awhile, but VSIZE is rock-steady).

The fact that things never (or slowly) converge for you and Trevor, in addition 
to the leak, makes me wonder if Trilinos seriously broke something between 11.x 
and 12.x. Trevor's been struggling to build 12.4. I'll try to find time to do 
the same.

In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters 
<mailto:waters.mik...@gmail.com> wrote:



When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guy

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Trevor,

My mystery deepens. Today, I've compiled and tried the following 
combinations of Trilinos/Swig:


12.8.1/3.0.2

11.14.3/3.0.2

11.12.1/3.0.2

11.10.1/2.0.8

11.2.5/2.0.8

All of these combinations leak memory at about the same rate. I am using 
Boost 1.54, Openmpi 1.6.5, and GCC 4.8.4. Are you using similar?


Thanks,

-Mike


On 03/30/2016 05:06 PM, Keller, Trevor (Fed) wrote:

Mike & Jon,

Running 25 steps on Debian wheezy using PyTrilinos 11.10.2 and swig 2.0.8, 
monitoring with memory-profiler, I do not see a memory leak (see attachment): 
the simulation is fairly steady at 8GB RAM. The same code, but using PyTrilinos 
12.1.0, ramps up to 16GB in the same simulation time.

Later revisions of Trilinos 12 remain defiant (though, compiling swig-3.0.8 has 
improved matters). If one of them compiles and works, I will let you know. If 
not, if possible, Mike, can you test the next incremental version (11.10.2)?

Trevor



From: fipy-boun...@nist.gov  on behalf of Michael Waters 

Sent: Wednesday, March 30, 2016 4:36 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos

Hi Jon,

I just compiled an old version of swig (2.0.8) and compiled Trilinos
(11.10.1) against that. Sadly, I am still having the leak.

I am out of ideas for the day... and should be looking for a post doc
anyway.

Thanks,
-mike


On 3/30/16 3:32 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

No worries. If building trilinos doesn't blindside you with something 
unexpected and unpleasant, you're not doing it right.

I have a conda recipe at 
https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
haven't tried to adjust it to 12.x, yet.


On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:


Hi Jon,

I was just reviewing my version of Trilinos 11.10 and discovered that there is 
no way that I compiled it last night after exercising. It has unsatisfied 
dependencies on my machine. So I must apologize, I must have been more tired 
than I thought.

Sorry for the error!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:

It looked to me like steps and accuracy were the way to do it, but my runs 
finish in one step, so I was confused. When I change to accuracy = 10.0**-6, it 
takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED time 
is because I put my laptop to sleep for awhile, but VSIZE is rock-steady).

The fact that things never (or slowly) converge for you and Trevor, in addition 
to the leak, makes me wonder if Trilinos seriously broke something between 11.x 
and 12.x. Trevor's been struggling to build 12.4. I'll try to find time to do 
the same.

In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:


When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
 wrote:


I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 MiB 
and saturates and RSS saturates at around 5 MiB. VSIZE is more relevant for 
tracking leaks, as RSS is deeply tied to your system's swapping architecture 
and what else is running; either way, neither seems to be leaking, but this 
problem does use a lot of memory.

What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:


Hello,

I still have a large memory leak when using Trilinos. I am not sure where to 
start looking so I made an example code that produces my problem in hopes that 
someone can help me.

But! my example is cool. I implemented Density Functional Theory in FiPy!

My code is slow, but runs in parallel and is simple (relative to most DFT 
codes). The example I have attached is just a lithium and hydrogen atom. The 
electrostatic boundary conditions are goofy but work well enough for 
demonstration purposes. If you set use_trilinos to True, the code will slowly 
use more memory. If not, it will try to use Pysparse.

Thanks,
-Michael Waters
__

Re: Memory Leaks with Trilinos

[Keller, Trevor (Fed)]

Thanks Jon, building swig v. 2.0.8 and using your cmake did the trick for 
11.10.2. Running the DFT program now.



Trevor



From: fipy-boun...@nist.gov  on behalf of Guyer, 
Jonathan E. Dr. (Fed) 
Sent: Wednesday, March 30, 2016 3:32 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos

No worries. If building trilinos doesn't blindside you with something 
unexpected and unpleasant, you're not doing it right.

I have a conda recipe at 
https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
haven't tried to adjust it to 12.x, yet.


On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:

> Hi Jon,
>
> I was just reviewing my version of Trilinos 11.10 and discovered that there 
> is no way that I compiled it last night after exercising. It has unsatisfied 
> dependencies on my machine. So I must apologize, I must have been more tired 
> than I thought.
>
> Sorry for the error!
> -Mike Waters
>
> On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
>> It looked to me like steps and accuracy were the way to do it, but my runs 
>> finish in one step, so I was confused. When I change to accuracy = 10.0**-6, 
>> it takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED 
>> time is because I put my laptop to sleep for awhile, but VSIZE is 
>> rock-steady).
>>
>> The fact that things never (or slowly) converge for you and Trevor, in 
>> addition to the leak, makes me wonder if Trilinos seriously broke something 
>> between 11.x and 12.x. Trevor's been struggling to build 12.4. I'll try to 
>> find time to do the same.
>>
>> In case it matters, I'm running on OS X. What's your system?
>>
>> - Jon
>>
>> On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:
>>
>>> When I did my testing and made those graphs, I ran Trilinos in serial.
>>> Syrupy didn't seem to track the other processes memory. I watched in
>>> real time as the parallel version ate all my ram though.
>>>
>>> To make the program run longer while not changing the memory:
>>>
>>> steps = 100  # increase this, (limits the number of self-consistent
>>> iterations)
>>> accuracy = 10.0**-5 # make this number smaller, (relative energy
>>> eigenvalue change for being considered converged )
>>> initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
>>> solver iterations per self-consistent iteration, to small and it's slow,
>>> to high and the solutions are not stable)
>>>
>>> I did those tests on a machine with 128 GB of ram so I wasn't expecting
>>> any swapping.
>>>
>>> Thanks,
>>> -mike
>>>
>>>
>>> On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
>>>> I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
>>>>
>>>>
>>>> On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
>>>>  wrote:
>>>>
>>>>> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 
>>>>> MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more 
>>>>> relevant for tracking leaks, as RSS is deeply tied to your system's 
>>>>> swapping architecture and what else is running; either way, neither seems 
>>>>> to be leaking, but this problem does use a lot of memory.
>>>>>
>>>>> What do I need to do to get it to run longer?
>>>>>
>>>>>
>>>>>
>>>>> On Mar 25, 2016, at 7:16 PM, Michael Waters  
>>>>> wrote:
>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> I still have a large memory leak when using Trilinos. I am not sure 
>>>>>> where to start looking so I made an example code that produces my 
>>>>>> problem in hopes that someone can help me.
>>>>>>
>>>>>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>>>>>>
>>>>>> My code is slow, but runs in parallel and is simple (relative to most 
>>>>>> DFT codes). The example I have attached is just a lithium and hydrogen 
>>>>>> atom. The electrostatic boundary conditions are goofy but work well 
>>>>>> enough for demonstration purposes. If you set use_trilinos to True, the 
>>>>>> code will slowly use more memory. If not, it will

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Jon,

I just compiled an old version of swig (2.0.8) and compiled Trilinos 
(11.10.1) against that. Sadly, I am still having the leak.

I am out of ideas for the day... and should be looking for a post doc 
anyway.

Thanks,
-mike


On 3/30/16 3:32 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> No worries. If building trilinos doesn't blindside you with something 
> unexpected and unpleasant, you're not doing it right.
>
> I have a conda recipe at 
> https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
> that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
> haven't tried to adjust it to 12.x, yet.
>
>
> On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:
>
>> Hi Jon,
>>
>> I was just reviewing my version of Trilinos 11.10 and discovered that there 
>> is no way that I compiled it last night after exercising. It has unsatisfied 
>> dependencies on my machine. So I must apologize, I must have been more tired 
>> than I thought.
>>
>> Sorry for the error!
>> -Mike Waters
>>
>> On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
>>> It looked to me like steps and accuracy were the way to do it, but my runs 
>>> finish in one step, so I was confused. When I change to accuracy = 
>>> 10.0**-6, it takes 15 steps, but still no leak (note, the hiccup in RSS and 
>>> in ELAPSED time is because I put my laptop to sleep for awhile, but VSIZE 
>>> is rock-steady).
>>>
>>> The fact that things never (or slowly) converge for you and Trevor, in 
>>> addition to the leak, makes me wonder if Trilinos seriously broke something 
>>> between 11.x and 12.x. Trevor's been struggling to build 12.4. I'll try to 
>>> find time to do the same.
>>>
>>> In case it matters, I'm running on OS X. What's your system?
>>>
>>> - Jon
>>>
>>> On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:
>>>
 When I did my testing and made those graphs, I ran Trilinos in serial.
 Syrupy didn't seem to track the other processes memory. I watched in
 real time as the parallel version ate all my ram though.

 To make the program run longer while not changing the memory:

 steps = 100  # increase this, (limits the number of self-consistent
 iterations)
 accuracy = 10.0**-5 # make this number smaller, (relative energy
 eigenvalue change for being considered converged )
 initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
 solver iterations per self-consistent iteration, to small and it's slow,
 to high and the solutions are not stable)

 I did those tests on a machine with 128 GB of ram so I wasn't expecting
 any swapping.

 Thanks,
 -mike


 On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
>
>
> On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
>  wrote:
>
>> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 
>> 11 MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more 
>> relevant for tracking leaks, as RSS is deeply tied to your system's 
>> swapping architecture and what else is running; either way, neither 
>> seems to be leaking, but this problem does use a lot of memory.
>>
>> What do I need to do to get it to run longer?
>>
>>
>>
>> On Mar 25, 2016, at 7:16 PM, Michael Waters  
>> wrote:
>>
>>> Hello,
>>>
>>> I still have a large memory leak when using Trilinos. I am not sure 
>>> where to start looking so I made an example code that produces my 
>>> problem in hopes that someone can help me.
>>>
>>> But! my example is cool. I implemented Density Functional Theory in 
>>> FiPy!
>>>
>>> My code is slow, but runs in parallel and is simple (relative to most 
>>> DFT codes). The example I have attached is just a lithium and hydrogen 
>>> atom. The electrostatic boundary conditions are goofy but work well 
>>> enough for demonstration purposes. If you set use_trilinos to True, the 
>>> code will slowly use more memory. If not, it will try to use Pysparse.
>>>
>>> Thanks,
>>> -Michael Waters
>>> ___
>>> fipy mailing list
>>> fipy@nist.gov
>>> http://www.ctcms.nist.gov/fipy
>>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>> 
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
 ___
 fipy mailing list
 fipy@nist.gov
 

Re: Memory Leaks with Trilinos

[Daniel Wheeler]

FYI, the Docker installation for FiPy: https://github.com/wd15/fipy-dockerize

On Wed, Mar 30, 2016 at 3:32 PM, Guyer, Jonathan E. Dr. (Fed)
 wrote:
> No worries. If building trilinos doesn't blindside you with something 
> unexpected and unpleasant, you're not doing it right.
>
> I have a conda recipe at 
> https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
> that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
> haven't tried to adjust it to 12.x, yet.
>
>
> On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:
>
>> Hi Jon,
>>
>> I was just reviewing my version of Trilinos 11.10 and discovered that there 
>> is no way that I compiled it last night after exercising. It has unsatisfied 
>> dependencies on my machine. So I must apologize, I must have been more tired 
>> than I thought.
>>
>> Sorry for the error!
>> -Mike Waters
>>
>> On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
>>> It looked to me like steps and accuracy were the way to do it, but my runs 
>>> finish in one step, so I was confused. When I change to accuracy = 
>>> 10.0**-6, it takes 15 steps, but still no leak (note, the hiccup in RSS and 
>>> in ELAPSED time is because I put my laptop to sleep for awhile, but VSIZE 
>>> is rock-steady).
>>>
>>> The fact that things never (or slowly) converge for you and Trevor, in 
>>> addition to the leak, makes me wonder if Trilinos seriously broke something 
>>> between 11.x and 12.x. Trevor's been struggling to build 12.4. I'll try to 
>>> find time to do the same.
>>>
>>> In case it matters, I'm running on OS X. What's your system?
>>>
>>> - Jon
>>>
>>> On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:
>>>
 When I did my testing and made those graphs, I ran Trilinos in serial.
 Syrupy didn't seem to track the other processes memory. I watched in
 real time as the parallel version ate all my ram though.

 To make the program run longer while not changing the memory:

 steps = 100  # increase this, (limits the number of self-consistent
 iterations)
 accuracy = 10.0**-5 # make this number smaller, (relative energy
 eigenvalue change for being considered converged )
 initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
 solver iterations per self-consistent iteration, to small and it's slow,
 to high and the solutions are not stable)

 I did those tests on a machine with 128 GB of ram so I wasn't expecting
 any swapping.

 Thanks,
 -mike


 On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
>
>
> On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
>  wrote:
>
>> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 
>> 11 MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more 
>> relevant for tracking leaks, as RSS is deeply tied to your system's 
>> swapping architecture and what else is running; either way, neither 
>> seems to be leaking, but this problem does use a lot of memory.
>>
>> What do I need to do to get it to run longer?
>>
>>
>>
>> On Mar 25, 2016, at 7:16 PM, Michael Waters  
>> wrote:
>>
>>> Hello,
>>>
>>> I still have a large memory leak when using Trilinos. I am not sure 
>>> where to start looking so I made an example code that produces my 
>>> problem in hopes that someone can help me.
>>>
>>> But! my example is cool. I implemented Density Functional Theory in 
>>> FiPy!
>>>
>>> My code is slow, but runs in parallel and is simple (relative to most 
>>> DFT codes). The example I have attached is just a lithium and hydrogen 
>>> atom. The electrostatic boundary conditions are goofy but work well 
>>> enough for demonstration purposes. If you set use_trilinos to True, the 
>>> code will slowly use more memory. If not, it will try to use Pysparse.
>>>
>>> Thanks,
>>> -Michael Waters
>>> ___
>>> fipy mailing list
>>> fipy@nist.gov
>>> http://www.ctcms.nist.gov/fipy
>>>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>>
>> 
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

 ___
 fipy mailing list
 fipy@nist.gov
 http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>>> 
>>>
>>>
>>> 

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

No worries. If building trilinos doesn't blindside you with something 
unexpected and unpleasant, you're not doing it right.

I have a conda recipe at 
https://github.com/guyer/conda-recipes/tree/trilinos_upgrade_11_10_2/trilinos 
that has worked for me to build 11.10.2 on both OS X and Docker (Debian?). I 
haven't tried to adjust it to 12.x, yet.


On Mar 30, 2016, at 2:42 PM, Michael Waters  wrote:

> Hi Jon,
> 
> I was just reviewing my version of Trilinos 11.10 and discovered that there 
> is no way that I compiled it last night after exercising. It has unsatisfied 
> dependencies on my machine. So I must apologize, I must have been more tired 
> than I thought. 
> 
> Sorry for the error!
> -Mike Waters
> 
> On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
>> It looked to me like steps and accuracy were the way to do it, but my runs 
>> finish in one step, so I was confused. When I change to accuracy = 10.0**-6, 
>> it takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED 
>> time is because I put my laptop to sleep for awhile, but VSIZE is 
>> rock-steady).
>> 
>> The fact that things never (or slowly) converge for you and Trevor, in 
>> addition to the leak, makes me wonder if Trilinos seriously broke something 
>> between 11.x and 12.x. Trevor's been struggling to build 12.4. I'll try to 
>> find time to do the same.
>> 
>> In case it matters, I'm running on OS X. What's your system?
>> 
>> - Jon
>> 
>> On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:
>> 
>>> When I did my testing and made those graphs, I ran Trilinos in serial. 
>>> Syrupy didn't seem to track the other processes memory. I watched in 
>>> real time as the parallel version ate all my ram though.
>>> 
>>> To make the program run longer while not changing the memory:
>>> 
>>> steps = 100  # increase this, (limits the number of self-consistent 
>>> iterations)
>>> accuracy = 10.0**-5 # make this number smaller, (relative energy 
>>> eigenvalue change for being considered converged )
>>> initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of 
>>> solver iterations per self-consistent iteration, to small and it's slow, 
>>> to high and the solutions are not stable)
>>> 
>>> I did those tests on a machine with 128 GB of ram so I wasn't expecting 
>>> any swapping.
>>> 
>>> Thanks,
>>> -mike
>>> 
>>> 
>>> On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
 I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
 
 
 On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
  wrote:
 
> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 
> MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more 
> relevant for tracking leaks, as RSS is deeply tied to your system's 
> swapping architecture and what else is running; either way, neither seems 
> to be leaking, but this problem does use a lot of memory.
> 
> What do I need to do to get it to run longer?
> 
> 
> 
> On Mar 25, 2016, at 7:16 PM, Michael Waters  
> wrote:
> 
>> Hello,
>> 
>> I still have a large memory leak when using Trilinos. I am not sure 
>> where to start looking so I made an example code that produces my 
>> problem in hopes that someone can help me.
>> 
>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>> 
>> My code is slow, but runs in parallel and is simple (relative to most 
>> DFT codes). The example I have attached is just a lithium and hydrogen 
>> atom. The electrostatic boundary conditions are goofy but work well 
>> enough for demonstration purposes. If you set use_trilinos to True, the 
>> code will slowly use more memory. If not, it will try to use Pysparse.
>> 
>> Thanks,
>> -Michael Waters
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
> 
> 
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
 
 ___
 fipy mailing list
 fipy@nist.gov
 http://www.ctcms.nist.gov/fipy
   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>>> 
>>> ___
>>> fipy mailing list
>>> fipy@nist.gov
>>> http://www.ctcms.nist.gov/fipy
>>>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>> 
>> 
>> 
>> ___
>> fipy mailing list
>> 
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>> 
>>   [ NIST internal ONLY: 
>> https://email.nist.gov/mailman/listinfo/fipy
>>  ]
>> 
> 
> ___

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Jon,

I was just reviewing my version of Trilinos 11.10 and discovered that 
there is /no way/ that I compiled it last night after exercising. It has 
unsatisfied dependencies on my machine. So I must apologize, I must have 
been more tired than I thought.


Sorry for the error!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
It looked to me like steps and accuracy were the way to do it, but my 
runs finish in one step, so I was confused. When I change to accuracy 
= 10.0**-6, it takes 15 steps, but still no leak (note, the hiccup in 
RSS and in ELAPSED time is because I put my laptop to sleep for 
awhile, but VSIZE is rock-steady).


The fact that things never (or slowly) converge for you and Trevor, in 
addition to the leak, makes me wonder if Trilinos seriously broke 
something between 11.x and 12.x. Trevor's been struggling to build 
12.4. I'll try to find time to do the same.


In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters > wrote:



When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
mailto:jonathan.gu...@nist.gov>> wrote:


I'm not seeing a leak. The below is for trilinos. VSIZE grows to 
about 11 MiB and saturates and RSS saturates at around 5 MiB. VSIZE 
is more relevant for tracking leaks, as RSS is deeply tied to your 
system's swapping architecture and what else is running; either 
way, neither seems to be leaking, but this problem does use a lot 
of memory.


What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:



Hello,

I still have a large memory leak when using Trilinos. I am not 
sure where to start looking so I made an example code that 
produces my problem in hopes that someone can help me.


But! my example is cool. I implemented Density Functional Theory 
in FiPy!


My code is slow, but runs in parallel and is simple (relative to 
most DFT codes). The example I have attached is just a lithium and 
hydrogen atom. The electrostatic boundary conditions are goofy but 
work well enough for demonstration purposes. If you 
set use_trilinos to True, the code will slowly use more memory. If 
not, it will try to use Pysparse.


Thanks,
-Michael Waters
___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Jon,

I was just reviewing my version of Trilinos 11.10 and discovered that 
there is /no way/ that I compiled it last night after exercising. It has 
unsatisfied dependencies on my machine. So I must apologize, I must have 
been more tired than I thought.


Sorry for the error!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
It looked to me like steps and accuracy were the way to do it, but my 
runs finish in one step, so I was confused. When I change to accuracy 
= 10.0**-6, it takes 15 steps, but still no leak (note, the hiccup in 
RSS and in ELAPSED time is because I put my laptop to sleep for 
awhile, but VSIZE is rock-steady).


The fact that things never (or slowly) converge for you and Trevor, in 
addition to the leak, makes me wonder if Trilinos seriously broke 
something between 11.x and 12.x. Trevor's been struggling to build 
12.4. I'll try to find time to do the same.


In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters > wrote:



When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
mailto:jonathan.gu...@nist.gov>> wrote:


I'm not seeing a leak. The below is for trilinos. VSIZE grows to 
about 11 MiB and saturates and RSS saturates at around 5 MiB. VSIZE 
is more relevant for tracking leaks, as RSS is deeply tied to your 
system's swapping architecture and what else is running; either 
way, neither seems to be leaking, but this problem does use a lot 
of memory.


What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:



Hello,

I still have a large memory leak when using Trilinos. I am not 
sure where to start looking so I made an example code that 
produces my problem in hopes that someone can help me.


But! my example is cool. I implemented Density Functional Theory 
in FiPy!


My code is slow, but runs in parallel and is simple (relative to 
most DFT codes). The example I have attached is just a lithium and 
hydrogen atom. The electrostatic boundary conditions are goofy but 
work well enough for demonstration purposes. If you 
set use_trilinos to True, the code will slowly use more memory. If 
not, it will try to use Pysparse.


Thanks,
-Michael Waters
___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

My plots are in KiBs, not MiBs (I didn't do anything to post-process the output 
of syrupy). >10 GiB for 1.3 M cells is consistent with our typical (excessive) 
memory usage.

We've been using Trilinos for a long time (~10 years?) so if you can manage to 
build them, you can try Trilinos versions going quite a way back.

As far as the previous calculations, you're right, of course. When I clear out 
previous results, it still doesn't leak (for me), but it runs forever.

On Mar 30, 2016, at 1:10 PM, Michael Waters  wrote:

> Hi Jon, 
> 
> Last night, I compiled and tried Trilinos 11.10 from the Trilinos git repo, I 
> am still getting the memory leak. I see it in both RSS and VSIZE. Looking 
> back at your memory numbers from your previous email, they seem too small for 
> even storing the float data for all the cells. 
> 
> The slow convergence that Trevor is seeing is normal. I did some more testing 
> and found that using initial_solver_iterations_per_step = 20 will improve 
> convergence relative to the number of self-consistent iterations in this 
> example. The total energy should go to about -316.xx eV when it is converged. 
> In the terminal output, it looks like this:
> 
> 
> 
> Step 24 Total Energy -315.957897784 dE 1.86055284542 dE/|toten| 
> 0.00588861002833 max_rel_error 0.125229938669
> 
> Also, the script will read previous calculations as starting conditions, to 
> prevent this, delete the psi_e_*.txt files. 
> 
> I am running these calculations on Ubuntu 14.04 64bit. Trevor, what are you 
> running on? I might be able to help with compilation. 
> 
> What is the oldest version of Trilinos that FiPy has used? I could try that 
> and see if my problem persists.
> 
> Thanks again!
> -Mike Waters
> 
> On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
>> It looked to me like steps and accuracy were the way to do it, but my runs 
>> finish in one step, so I was confused. When I change to accuracy = 10.0**-6, 
>> it takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED 
>> time is because I put my laptop to sleep for awhile, but VSIZE is 
>> rock-steady).
>> 
>> The fact that things never (or slowly) converge for you and Trevor, in 
>> addition to the leak, makes me wonder if Trilinos seriously broke something 
>> between 11.x and 12.x. Trevor's been struggling to build 12.4. I'll try to 
>> find time to do the same.
>> 
>> In case it matters, I'm running on OS X. What's your system?
>> 
>> - Jon
>> 
>> On Mar 29, 2016, at 3:59 PM, Michael Waters  wrote:
>> 
>>> When I did my testing and made those graphs, I ran Trilinos in serial. 
>>> Syrupy didn't seem to track the other processes memory. I watched in 
>>> real time as the parallel version ate all my ram though.
>>> 
>>> To make the program run longer while not changing the memory:
>>> 
>>> steps = 100  # increase this, (limits the number of self-consistent 
>>> iterations)
>>> accuracy = 10.0**-5 # make this number smaller, (relative energy 
>>> eigenvalue change for being considered converged )
>>> initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of 
>>> solver iterations per self-consistent iteration, to small and it's slow, 
>>> to high and the solutions are not stable)
>>> 
>>> I did those tests on a machine with 128 GB of ram so I wasn't expecting 
>>> any swapping.
>>> 
>>> Thanks,
>>> -mike
>>> 
>>> 
>>> On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
 I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
 
 
 On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
  wrote:
 
> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 
> MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more 
> relevant for tracking leaks, as RSS is deeply tied to your system's 
> swapping architecture and what else is running; either way, neither seems 
> to be leaking, but this problem does use a lot of memory.
> 
> What do I need to do to get it to run longer?
> 
> 
> 
> On Mar 25, 2016, at 7:16 PM, Michael Waters  
> wrote:
> 
>> Hello,
>> 
>> I still have a large memory leak when using Trilinos. I am not sure 
>> where to start looking so I made an example code that produces my 
>> problem in hopes that someone can help me.
>> 
>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>> 
>> My code is slow, but runs in parallel and is simple (relative to most 
>> DFT codes). The example I have attached is just a lithium and hydrogen 
>> atom. The electrostatic boundary conditions are goofy but work well 
>> enough for demonstration purposes. If you set use_trilinos to True, the 
>> code will slowly use more memory. If not, it will try to use Pysparse.
>> 
>> Thanks,
>> -Michael Waters
>> ___
>> fipy ma

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Jon,

Last night, I compiled and tried Trilinos 11.10 from the Trilinos git 
repo, I am still getting the memory leak. I see it in both RSS and 
VSIZE. Looking back at your memory numbers from your previous email, 
they seem too small for even storing the float data for all the cells.


The slow convergence that Trevor is seeing is normal. I did some more 
testing and found that using initial_solver_iterations_per_step = 20 
will improve convergence relative to the number of self-consistent 
iterations in this example. The total energy should go to about -316.xx 
eV when it is converged. In the terminal output, it looks like this:




Step 24 Total Energy -315.957897784 dE 1.86055284542 dE/|toten| 
0.00588861002833 max_rel_error 0.125229938669


Also, the script will read previous calculations as starting conditions, 
to prevent this, delete the psi_e_*.txt files.


I am running these calculations on Ubuntu 14.04 64bit. /Trevor,/ what 
are you running on? I might be able to help with compilation.


What is the oldest version of Trilinos that FiPy has used? I could try 
that and see if my problem persists.


Thanks again!
-Mike Waters

On 3/30/16 11:52 AM, Guyer, Jonathan E. Dr. (Fed) wrote:
It looked to me like steps and accuracy were the way to do it, but my 
runs finish in one step, so I was confused. When I change to accuracy 
= 10.0**-6, it takes 15 steps, but still no leak (note, the hiccup in 
RSS and in ELAPSED time is because I put my laptop to sleep for 
awhile, but VSIZE is rock-steady).


The fact that things never (or slowly) converge for you and Trevor, in 
addition to the leak, makes me wonder if Trilinos seriously broke 
something between 11.x and 12.x. Trevor's been struggling to build 
12.4. I'll try to find time to do the same.


In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters > wrote:



When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:

I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
mailto:jonathan.gu...@nist.gov>> wrote:


I'm not seeing a leak. The below is for trilinos. VSIZE grows to 
about 11 MiB and saturates and RSS saturates at around 5 MiB. VSIZE 
is more relevant for tracking leaks, as RSS is deeply tied to your 
system's swapping architecture and what else is running; either 
way, neither seems to be leaking, but this problem does use a lot 
of memory.


What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:



Hello,

I still have a large memory leak when using Trilinos. I am not 
sure where to start looking so I made an example code that 
produces my problem in hopes that someone can help me.


But! my example is cool. I implemented Density Functional Theory 
in FiPy!


My code is slow, but runs in parallel and is simple (relative to 
most DFT codes). The example I have attached is just a lithium and 
hydrogen atom. The electrostatic boundary conditions are goofy but 
work well enough for demonstration purposes. If you 
set use_trilinos to True, the code will slowly use more memory. If 
not, it will try to use Pysparse.


Thanks,
-Michael Waters
___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov 
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]



___
fipy mailing list
fipy@nist.gov
h

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

It looked to me like steps and accuracy were the way to do it, but my runs 
finish in one step, so I was confused. When I change to accuracy = 10.0**-6, it 
takes 15 steps, but still no leak (note, the hiccup in RSS and in ELAPSED time 
is because I put my laptop to sleep for awhile, but VSIZE is rock-steady).

The fact that things never (or slowly) converge for you and Trevor, in addition 
to the leak, makes me wonder if Trilinos seriously broke something between 11.x 
and 12.x. Trevor's been struggling to build 12.4. I'll try to find time to do 
the same.

In case it matters, I'm running on OS X. What's your system?

- Jon

On Mar 29, 2016, at 3:59 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:

When I did my testing and made those graphs, I ran Trilinos in serial.
Syrupy didn't seem to track the other processes memory. I watched in
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of
solver iterations per self-consistent iteration, to small and it's slow,
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
mailto:jonathan.gu...@nist.gov>> wrote:

I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 MiB 
and saturates and RSS saturates at around 5 MiB. VSIZE is more relevant for 
tracking leaks, as RSS is deeply tied to your system's swapping architecture 
and what else is running; either way, neither seems to be leaking, but this 
problem does use a lot of memory.

What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:

Hello,

I still have a large memory leak when using Trilinos. I am not sure where to 
start looking so I made an example code that produces my problem in hopes that 
someone can help me.

But! my example is cool. I implemented Density Functional Theory in FiPy!

My code is slow, but runs in parallel and is simple (relative to most DFT 
codes). The example I have attached is just a lithium and hydrogen atom. The 
electrostatic boundary conditions are goofy but work well enough for 
demonstration purposes. If you set use_trilinos to True, the code will slowly 
use more memory. If not, it will try to use Pysparse.

Thanks,
-Michael Waters
___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
[cid:707090B0-3796-43B2-A403-760CE215C6DF]
___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Michael Waters]

When I did my testing and made those graphs, I ran Trilinos in serial. 
Syrupy didn't seem to track the other processes memory. I watched in 
real time as the parallel version ate all my ram though.

To make the program run longer while not changing the memory:

steps = 100  # increase this, (limits the number of self-consistent 
iterations)
accuracy = 10.0**-5 # make this number smaller, (relative energy 
eigenvalue change for being considered converged )
initial_solver_iterations_per_step = 7 # reduce this to 1,  (number of 
solver iterations per self-consistent iteration, to small and it's slow, 
to high and the solutions are not stable)

I did those tests on a machine with 128 GB of ram so I wasn't expecting 
any swapping.

Thanks,
-mike


On 3/29/16 3:38 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.
>
>
> On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
>  wrote:
>
>> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 
>> MiB and saturates and RSS saturates at around 5 MiB. VSIZE is more relevant 
>> for tracking leaks, as RSS is deeply tied to your system's swapping 
>> architecture and what else is running; either way, neither seems to be 
>> leaking, but this problem does use a lot of memory.
>>
>> What do I need to do to get it to run longer?
>>
>>
>>
>> On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:
>>
>>> Hello,
>>>
>>> I still have a large memory leak when using Trilinos. I am not sure where 
>>> to start looking so I made an example code that produces my problem in 
>>> hopes that someone can help me.
>>>
>>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>>>
>>> My code is slow, but runs in parallel and is simple (relative to most DFT 
>>> codes). The example I have attached is just a lithium and hydrogen atom. 
>>> The electrostatic boundary conditions are goofy but work well enough for 
>>> demonstration purposes. If you set use_trilinos to True, the code will 
>>> slowly use more memory. If not, it will try to use Pysparse.
>>>
>>> Thanks,
>>> -Michael Waters
>>> ___
>>> fipy mailing list
>>> fipy@nist.gov
>>> http://www.ctcms.nist.gov/fipy
>>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>>
>> 
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

I guess I spoke too soon. FWIW, I'm running Trilinos version: 11.10.2.


On Mar 29, 2016, at 3:34 PM, Guyer, Jonathan E. Dr. (Fed) 
 wrote:

> I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 MiB 
> and saturates and RSS saturates at around 5 MiB. VSIZE is more relevant for 
> tracking leaks, as RSS is deeply tied to your system's swapping architecture 
> and what else is running; either way, neither seems to be leaking, but this 
> problem does use a lot of memory.
> 
> What do I need to do to get it to run longer?
> 
> 
> 
> On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:
> 
>> Hello,
>> 
>> I still have a large memory leak when using Trilinos. I am not sure where to 
>> start looking so I made an example code that produces my problem in hopes 
>> that someone can help me.
>> 
>> But! my example is cool. I implemented Density Functional Theory in FiPy! 
>> 
>> My code is slow, but runs in parallel and is simple (relative to most DFT 
>> codes). The example I have attached is just a lithium and hydrogen atom. The 
>> electrostatic boundary conditions are goofy but work well enough for 
>> demonstration purposes. If you set use_trilinos to True, the code will 
>> slowly use more memory. If not, it will try to use Pysparse. 
>> 
>> Thanks, 
>> -Michael Waters
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
> 
> 
> 
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

I'm not seeing a leak. The below is for trilinos. VSIZE grows to about 11 MiB 
and saturates and RSS saturates at around 5 MiB. VSIZE is more relevant for 
tracking leaks, as RSS is deeply tied to your system's swapping architecture 
and what else is running; either way, neither seems to be leaking, but this 
problem does use a lot of memory.

What do I need to do to get it to run longer?



On Mar 25, 2016, at 7:16 PM, Michael Waters 
mailto:waters.mik...@gmail.com>> wrote:

Hello,

I still have a large memory leak when using Trilinos. I am not sure where to 
start looking so I made an example code that produces my problem in hopes that 
someone can help me.

But! my example is cool. I implemented Density Functional Theory in FiPy!

My code is slow, but runs in parallel and is simple (relative to most DFT 
codes). The example I have attached is just a lithium and hydrogen atom. The 
electrostatic boundary conditions are goofy but work well enough for 
demonstration purposes. If you set use_trilinos to True, the code will slowly 
use more memory. If not, it will try to use Pysparse.

Thanks,
-Michael Waters
___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
 [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


[cid:8B4A0446-BEA1-43AB-8791-88D0917469DF]
___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Michael Waters]

Hi Trevor,

I am seeing the same behavior as you regarding Pysparse vs Trilinos.

I compiled Trilinos version 12.4.2 on our machine running Ubuntu 14.04.  
But! when I get the PyTrilinos version from PyTrilinos, it says: 
12.3;(;Dev;)

I see there is a new version of Trilinos out (12.6), I'll compile that 
and report back.

Thanks for investigating,
-Mike

On 3/29/16 2:41 PM, Keller, Trevor (Fed) wrote:
> Hi Mike,
>
> I can confirm the leak in this code/solver system using FiPy on a Linux box. 
> Some details:
> FiPy: 3.1-dev131-g77851ef
> PyTrilinos: 12.1;(;Dev;)
> PySparse: 1.1.1
>
> I commented out the solver and preconditioner specifications, using the 
> default behavior when launching the script with --pysparse or --trilinos 
> flags instead. I also disabled garbage collection. Memory use was monitored 
> using memory-profiler, launching the script using
>  OMP_NUM_THREADS=6 mprof run mod-fipy-dft.py --trilinos
> or
>  OMP_NUM_THREADS=6 mprof run mod-fipy-dft.py --pysparse
>
> For 100 steps, the pysparse solver plateaued around 8GB, with regular spikes 
> in RAM consumption while solving the equations.
> For 100 steps, the pytrilinos solver just keeps gobbling up  RAM -- it had 
> about 37GB at the last step.
> Using the solver (LinearBicgstabSolver) and preconditioner 
> (JacobiPreconditioner) you specified for Trilinos, it's a little better, 
> ramping up to just 24GB after 100 steps.
>
>
> Digging around a little bit, there was a patch to PyTrilinos  (for v. 12.4) 
> in October meant to address some memory leaks -- I'm cloning and re-building 
> the project now. Can you report your PyTrilinos version, please, using
>  >>> from PyTrilinos import __version__ as PyTriVer
>  >>> print PyTriVer
>
> Thank you,
> Trevor
>
>
>
> Trevor Keller, Ph.D.
> Materials Science and Engineering Division
> National Institute of Standards and Technology
> 100 Bureau Dr. MS 8550; Gaithersburg, MD 20899
> Office: 223/A131 or (301) 975-2889
>
>
> ____________
> From: fipy-boun...@nist.gov  on behalf of Michael 
> Waters 
> Sent: Monday, March 28, 2016 2:10 PM
> To: FIPY
> Subject: Re: Memory Leaks with Trilinos
>
> Thanks, If you think DFT in FiPy is strange, you should see how I make
> isosurfaces in POV-Ray. :)
>
> That said, If anyone is tinkering with my example and has questions, I
> am glad to answer them!
>
> Cheers,
> -Mike Waters
>
>
>
> On 03/28/2016 01:46 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
>> Mike, thanks for the example, and for the rather perverse application of 
>> FiPy!
>>
>> I'll fiddle with this and see what I get.
>>
>> - Jon
>>
>> On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:
>>
>>> Hello,
>>>
>>> I still have a large memory leak when using Trilinos. I am not sure where 
>>> to start looking so I made an example code that produces my problem in 
>>> hopes that someone can help me.
>>>
>>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>>>
>>> My code is slow, but runs in parallel and is simple (relative to most DFT 
>>> codes). The example I have attached is just a lithium and hydrogen atom. 
>>> The electrostatic boundary conditions are goofy but work well enough for 
>>> demonstration purposes. If you set use_trilinos to True, the code will 
>>> slowly use more memory. If not, it will try to use Pysparse.
>>>
>>> Thanks,
>>> -Michael Waters
>>> ___
>>> fipy mailing list
>>> fipy@nist.gov
>>> http://www.ctcms.nist.gov/fipy
>>>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>> [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Keller, Trevor (Fed)]

Hi Mike,

I can confirm the leak in this code/solver system using FiPy on a Linux box. 
Some details:
FiPy: 3.1-dev131-g77851ef
PyTrilinos: 12.1;(;Dev;)
PySparse: 1.1.1

I commented out the solver and preconditioner specifications, using the default 
behavior when launching the script with --pysparse or --trilinos flags instead. 
I also disabled garbage collection. Memory use was monitored using 
memory-profiler, launching the script using
OMP_NUM_THREADS=6 mprof run mod-fipy-dft.py --trilinos
or
OMP_NUM_THREADS=6 mprof run mod-fipy-dft.py --pysparse

For 100 steps, the pysparse solver plateaued around 8GB, with regular spikes in 
RAM consumption while solving the equations.
For 100 steps, the pytrilinos solver just keeps gobbling up  RAM -- it had 
about 37GB at the last step.
Using the solver (LinearBicgstabSolver) and preconditioner 
(JacobiPreconditioner) you specified for Trilinos, it's a little better, 
ramping up to just 24GB after 100 steps.


Digging around a little bit, there was a patch to PyTrilinos  (for v. 12.4) in 
October meant to address some memory leaks -- I'm cloning and re-building the 
project now. Can you report your PyTrilinos version, please, using
>>> from PyTrilinos import __version__ as PyTriVer
>>> print PyTriVer

Thank you,
Trevor



Trevor Keller, Ph.D.
Materials Science and Engineering Division
National Institute of Standards and Technology
100 Bureau Dr. MS 8550; Gaithersburg, MD 20899
Office: 223/A131 or (301) 975-2889



From: fipy-boun...@nist.gov  on behalf of Michael Waters 

Sent: Monday, March 28, 2016 2:10 PM
To: FIPY
Subject: Re: Memory Leaks with Trilinos

Thanks, If you think DFT in FiPy is strange, you should see how I make
isosurfaces in POV-Ray. :)

That said, If anyone is tinkering with my example and has questions, I
am glad to answer them!

Cheers,
-Mike Waters



On 03/28/2016 01:46 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> Mike, thanks for the example, and for the rather perverse application of FiPy!
>
> I'll fiddle with this and see what I get.
>
> - Jon
>
> On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:
>
>> Hello,
>>
>> I still have a large memory leak when using Trilinos. I am not sure where to 
>> start looking so I made an example code that produces my problem in hopes 
>> that someone can help me.
>>
>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>>
>> My code is slow, but runs in parallel and is simple (relative to most DFT 
>> codes). The example I have attached is just a lithium and hydrogen atom. The 
>> electrostatic boundary conditions are goofy but work well enough for 
>> demonstration purposes. If you set use_trilinos to True, the code will 
>> slowly use more memory. If not, it will try to use Pysparse.
>>
>> Thanks,
>> -Michael Waters
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Michael Waters]

Thanks, If you think DFT in FiPy is strange, you should see how I make 
isosurfaces in POV-Ray. :)

That said, If anyone is tinkering with my example and has questions, I 
am glad to answer them!

Cheers,
-Mike Waters



On 03/28/2016 01:46 PM, Guyer, Jonathan E. Dr. (Fed) wrote:
> Mike, thanks for the example, and for the rather perverse application of FiPy!
>
> I'll fiddle with this and see what I get.
>
> - Jon
>
> On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:
>
>> Hello,
>>
>> I still have a large memory leak when using Trilinos. I am not sure where to 
>> start looking so I made an example code that produces my problem in hopes 
>> that someone can help me.
>>
>> But! my example is cool. I implemented Density Functional Theory in FiPy!
>>
>> My code is slow, but runs in parallel and is simple (relative to most DFT 
>> codes). The example I have attached is just a lithium and hydrogen atom. The 
>> electrostatic boundary conditions are goofy but work well enough for 
>> demonstration purposes. If you set use_trilinos to True, the code will 
>> slowly use more memory. If not, it will try to use Pysparse.
>>
>> Thanks,
>> -Michael Waters
>> ___
>> fipy mailing list
>> fipy@nist.gov
>> http://www.ctcms.nist.gov/fipy
>>   [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
>
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>[ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]

Re: Memory Leaks with Trilinos

[Guyer, Jonathan E. Dr. (Fed)]

Mike, thanks for the example, and for the rather perverse application of FiPy!

I'll fiddle with this and see what I get.

- Jon

On Mar 25, 2016, at 7:16 PM, Michael Waters  wrote:

> Hello,
> 
> I still have a large memory leak when using Trilinos. I am not sure where to 
> start looking so I made an example code that produces my problem in hopes 
> that someone can help me.
> 
> But! my example is cool. I implemented Density Functional Theory in FiPy! 
> 
> My code is slow, but runs in parallel and is simple (relative to most DFT 
> codes). The example I have attached is just a lithium and hydrogen atom. The 
> electrostatic boundary conditions are goofy but work well enough for 
> demonstration purposes. If you set use_trilinos to True, the code will slowly 
> use more memory. If not, it will try to use Pysparse. 
> 
> Thanks, 
> -Michael Waters
> ___
> fipy mailing list
> fipy@nist.gov
> http://www.ctcms.nist.gov/fipy
>  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]


___
fipy mailing list
fipy@nist.gov
http://www.ctcms.nist.gov/fipy
  [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]