search

openworm / OpenWorm

Repository for the main Dockerfile with the OpenWorm software stack and project-wide issues
http://openworm.org
MIT License
2.68k stars 211 forks source link

Worm leaking #163

Closed vellamike closed 10 years ago

vellamike commented 10 years ago

When I run using -l_to it looks like the membrane behaves differently and becomes somewhat porous. Looking at it closely - I think there might be no membrnae!

Do you know why this might be @skhayrulin @a-palyanov ?

Here is a screenshot of what I mean:

leaky_worm

Neurophile commented 10 years ago

Latest master branch produces similar behavior for me. I get a cloud of fluid forming around the upper surface of the worm.

vellamike commented 10 years ago

@Neurophile is it consistent behaviour for you?

Neurophile commented 10 years ago

Only 2 runs so far, but yes 2:2. Also, latest seems to have fixed some other bad behavior I saw before, still gathering data.

vellamike commented 10 years ago

This is either the changes I made to sphFluid.cl or something Sergey introduced in the last couple of days.

On 18 December 2013 16:22, Neurophile notifications@github.com wrote:

Only 2 runs so far, but yes 2:2. Also, latest seems to have fixed some other bad behavior I saw before, still gathering data.

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30855713 .

vellamike commented 10 years ago

The membranes, which are usually represented as light blue triangles, are missing when simulating using -l_to and visualising using -l_from. This screenshot shows the surface which should be covered in little turquoise triangles but isn't:

screenshot from 2013-12-18 17 18 09

Neurophile commented 10 years ago

Run with no command line args. Particles escape the worm, but the cyan mesh appears to be in place: screenshot 2013-12-18 13 18 37

vellamike commented 10 years ago

Are you sure the particles are escaping the worm there? When I run with no CLI args its fine.

On 18 December 2013 22:25, Neurophile notifications@github.com wrote:

Run with no command line args. Particles escape the worm, but the cyan mesh appears to be in place: [image: screenshot 2013-12-18 13 18 37]https://f.cloud.github.com/assets/5497006/1778333/f512deee-6832-11e3-9c30-15a02918042d.png

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30887392 .

skhayrulin commented 10 years ago

Hi Guys! It' looks very strange I couldn't reproduce such behavior. about work of -l_to detectives it doesn't affect on work of opencl code and on work of full simulation it just copies position buffers to file on each 10 iteration. -l_from copy buffers from file to simulation back and there are no any calculation. About display of triangle of membranes it's a bug but it does't affect on participles when you run simulation with -l_from. With what branch are your working?

Neurophile commented 10 years ago

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

@vellamike particles are definitely escaping in my previous image. You can't tell from that angle, but if I look down the worm axis it is obvious.

vellamike commented 10 years ago

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579 .

skhayrulin commented 10 years ago

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579> .

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233 .

Best regards, Khayrulin Sergey

vellamike commented 10 years ago

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579> .

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297 .

skhayrulin commented 10 years ago

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297> .

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403 .

Best regards, Khayrulin Sergey

vellamike commented 10 years ago

electrophysiology branch shouldn't even be working for you though - in src it's currently missing the owOpenCLConstant.h file so you shouldn't be able to compile it.

On 19 December 2013 07:59, Sergey notifications@github.com wrote:

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297> .

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30911167 .

skhayrulin commented 10 years ago

Hm strange it works fine for me

2013/12/19 Mike Vella notifications@github.com

electrophysiology branch shouldn't even be working for you though - in src it's currently missing the owOpenCLConstant.h file so you shouldn't be able to compile it.

On 19 December 2013 07:59, Sergey notifications@github.com wrote:

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911167> .

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30911371 .

Best regards, Khayrulin Sergey

vellamike commented 10 years ago

Something is wrong then! The owOpenCLConstant.h file is missing and any

incude "owOpenCLConstant.h" statement (not in the .cl code, in .cpp files)

should throw an error. This might explain why you're not seeing the leaking worm.

On 19 December 2013 08:09, Sergey notifications@github.com wrote:

Hm strange it works fine for me

2013/12/19 Mike Vella notifications@github.com

electrophysiology branch shouldn't even be working for you though - in src it's currently missing the owOpenCLConstant.h file so you shouldn't be able to compile it.

On 19 December 2013 07:59, Sergey notifications@github.com wrote:

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911167> .

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911371> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30911572 .

skhayrulin commented 10 years ago

Ou yes sorry Mike my fault I'm working on my local branch which is child of electrophysiology in there #incude "owOpenCLConstant.h" is uncomment and owOpenCLConstant.h doesn't deleted

2013/12/19 Mike Vella notifications@github.com

Something is wrong then! The owOpenCLConstant.h file is missing and any

incude "owOpenCLConstant.h" statement (not in the .cl code, in .cpp

files) should throw an error. This might explain why you're not seeing the leaking worm.

On 19 December 2013 08:09, Sergey notifications@github.com wrote:

Hm strange it works fine for me

2013/12/19 Mike Vella notifications@github.com

electrophysiology branch shouldn't even be working for you though - in src it's currently missing the owOpenCLConstant.h file so you shouldn't be able to compile it.

On 19 December 2013 07:59, Sergey notifications@github.com wrote:

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911167>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911371> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911572> .

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30911644 .

Best regards, Khayrulin Sergey

vellamike commented 10 years ago

OK - try working with the latest master branch and see if you are able to reproduce the leaking worm I and @Neurophile see.

On 19 December 2013 08:21, Sergey notifications@github.com wrote:

Ou yes sorry Mike my fault I'm working on my local branch which is child of electrophysiology in there #incude "owOpenCLConstant.h" is uncomment and owOpenCLConstant.h doesn't deleted

2013/12/19 Mike Vella notifications@github.com

Something is wrong then! The owOpenCLConstant.h file is missing and any

incude "owOpenCLConstant.h" statement (not in the .cl code, in .cpp

files) should throw an error. This might explain why you're not seeing the leaking worm.

On 19 December 2013 08:09, Sergey notifications@github.com wrote:

Hm strange it works fine for me

2013/12/19 Mike Vella notifications@github.com

electrophysiology branch shouldn't even be working for you though - in src it's currently missing the owOpenCLConstant.h file so you shouldn't be able to compile it.

On 19 December 2013 07:59, Sergey notifications@github.com wrote:

Yes Mike sure I can try but as Neurophile sad the problem is in opencl code file, so we can roll back sphFluid.cl. In electrophysiology branch using old file without new code. I'll try to look on new opencl optimization code a bit later.

Thanks, Sergey

2013/12/19 Mike Vella notifications@github.com

Could you try and reproduce this bug on master? Im not working on electrophysiology On 19 Dec 2013 07:35, "Sergey" notifications@github.com wrote:

I'm working on electrophysiology branch

2013/12/19 Mike Vella notifications@github.com

I'm also working on the master branch sergey. Its very strange that you don't have this leak problem. On 19 Dec 2013 04:33, "Neurophile" notifications@github.com

wrote:

This is definitely in the new opencl code. When I revert to sphFluid.cl from a couple of days ago I get everything as expected. Running on latest master.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30904579>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910233>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910297>

.

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30910403>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub<

https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911167>

.

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911371>

.

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911572> .

— Reply to this email directly or view it on GitHub< https://github.com/openworm/OpenWorm/issues/163#issuecomment-30911644> .

Best regards, Khayrulin Sergey

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/163#issuecomment-30912077 .

Neurophile commented 10 years ago

Found it. Pull request coming.

vellamike commented 10 years ago

Nice catch @Neurophile. Many thanks for that!