I was looking at another thread \"My PC isn\'t calculating anything\" and notice the following comment from Jack: \"This is a lock file used to ensure only one instance is of the cognitive architecture is compiled during the first run. This used primarily for computers that have multiple cores.\"
Is MindModeling working now for Intel multicores?
/ 
Reply
Quote
I was looking at another thread \"My PC isn\'t calculating anything\" and notice the following comment from Jack: \"This is a lock file used to ensure only one instance is of the cognitive architecture is compiled during the first run. This used primarily for computers that have multiple cores.\"
Is MindModeling working now for Intel multicores?
Steve,
It does work for multi-cored computers.
The lockfile is used during inital setup for the very 1st workunit crunched of a set.
It ensures only one instance of the app is trying to compile the program.
If the app is killed in the middle of this process a bad compilation causes things not to run.
The lock file is removed automatically after first compilation (or 10 mins) whichever comes first.
I recommend deleting from your projects directory any directory with the name containing actr6v1.3r563_default.
This will for the next workunit to recompile things.
[If you have 4 instances start and things are all compiled all 4 go immediately otherwise the first instance compiles things and the 3 others wait unitl that it is done (or for 10 min) whichever comes first.
[There is the possibility that 10 is to short in the cause of a older machine.]
Let me know if that helped,
Jack
/ Reply
Quote
I was looking at another thread \"My PC isn\'t calculating anything\" and notice the following comment from Jack: \"This is a lock file used to ensure only one instance is of the cognitive architecture is compiled during the first run. This used primarily for computers that have multiple cores.\"
Is MindModeling working now for Intel multicores?
Steve,
It does work for multi-cored computers.
The lockfile is used during inital setup for the very 1st workunit crunched of a set.
It ensures only one instance of the app is trying to compile the program.
If the app is killed in the middle of this process a bad compilation causes things not to run.
The lock file is removed automatically after first compilation (or 10 mins) whichever comes first.
I recommend deleting from your projects directory any directory with the name containing actr6v1.3r563_default.
This will for the next workunit to recompile things.
[If you have 4 instances start and things are all compiled all 4 go immediately otherwise the first instance compiles things and the 3 others wait unitl that it is done (or for 10 min) whichever comes first.
[There is the possibility that 10 is to short in the cause of a older machine.]
Let me know if that helped,
Jack
There definitely seems to be a race condition happening that the lockfile is not stopping on some of the faster machines. I am adding a new mechanism to prevent this in the next version.
/ Reply
Quote
I was looking at another thread \"My PC isn\'t calculating anything\" and notice the following comment from Jack: \"This is a lock file used to ensure only one instance is of the cognitive architecture is compiled during the first run. This used primarily for computers that have multiple cores.\"
Is MindModeling working now for Intel multicores?
Steve,
It does work for multi-cored computers.
The lockfile is used during inital setup for the very 1st workunit crunched of a set.
It ensures only one instance of the app is trying to compile the program.
If the app is killed in the middle of this process a bad compilation causes things not to run.
The lock file is removed automatically after first compilation (or 10 mins) whichever comes first.
I recommend deleting from your projects directory any directory with the name containing actr6v1.3r563_default.
This will for the next workunit to recompile things.
[If you have 4 instances start and things are all compiled all 4 go immediately otherwise the first instance compiles things and the 3 others wait unitl that it is done (or for 10 min) whichever comes first.
[There is the possibility that 10 is to short in the cause of a older machine.]
Let me know if that helped,
Jack
There definitely seems to be a race condition happening that the lockfile is not stopping on some of the faster machines. I am adding a new mechanism to prevent this in the next version.
And that next version will be available when? :)
And just to be clear, the problem I was asking about concerned the processor usage on multicore machine (not processing @100%). I originally asked about this when running MindModeling on my laptop, which wouldn\'t run at all w/o bombing.
/ Reply
Quote
Edit: (sort of)
That was under Vista that I couldn\'t run MindModeling. Still wondering if that\'s the case.
/ Reply
Quote
I\'ve been crunching on my dual-core for weeks now without any problems at all :)
____________
/ Reply
Quote
My C2D & my 3.4GhzP4W/HT boxes both crunch fine. 100% of the wu\'s crash on my Q9450 Quad Core Box I added Fri nite. Where can I get a copy of the LISP file. Where do I find it in the project folder to replace it? Joe B.
/ Reply
Quote
My C2D & my 3.4GhzP4W/HT boxes both crunch fine. 100% of the wu\'s crash on my Q9450 Quad Core Box I added Fri nite. Where can I get a copy of the LISP file. Where do I find it in the project folder to replace it? Joe B.
Multi-core support is a little different (BETTER) in 3.4+ applications. There is no locking mechanism.
- On start: a instance (core) checks to see if a cached copy of a compiled cognitive architecture is available in the projects directory.
- If not - it makes its own.
- When they have finished a successful WU they cache the data back to the projects directory. (if it is still not there)
This mechanisms is much more robust.
--Jack
____________
MindModeling@Home is fun
/ Reply
Quote
Thanks, Jack :) Should get more processing power on the project now!
/ Reply
Quote