Re: spamming with "Warning: not enough memory, new stack ..."

[Bill Allombert <Bill.Allombert@math.u-bordeaux.fr>]

On Thu, Feb 22, 2018 at 09:00:35AM -0500, Max Alekseyev wrote:
> Hello!
> 
> While I was running a long computation in GP (with rather stable
> allocation of memory), another process has eaten all system memory and
> got killed by the system. That has some weird effect on the running GP
> - it started printing messages like quoted below over and over again
> (long after the incident with the system memory). From time to time
> these messages are interveawed with the messages I expect from the
> computation, so my hope is that the computation itself is not
> affected.
> 
> ...
>   ***   Warning: not enough memory, new stack 4294967296
>   ***   Warning: not enough memory, new stack 2147483648
>   ***   Warning: not enough memory, new stack 1073741824
>   ***   Warning: not enough memory, new stack 4294967296
>   ***   Warning: not enough memory, new stack 2147483648
>   ***   Warning: not enough memory, new stack 1073741824
>   ***   Warning: not enough memory, new stack 4294967296
>   ***   Warning: not enough memory, new stack 2147483648
>   ***   Warning: not enough memory, new stack 1073741824
>   ***   Warning: not enough memory, new stack 4294967296
>   ***   Warning: not enough memory, new stack 2147483648
>   ***   Warning: not enough memory, new stack 1073741824
> 
> Is this an expected behavior? Why the stack goes up and down by
> itself? (my program does not make any stack reallocations)

I assume you are using the pthread version and has set
threadsize (or threadsizemax) to 8GB.
In which case each time GP tries to start a thread, it will
try to allocate a thread stack and display this warning
for each threads.

Cheers,
Bill.