[tform] Crash in incrementing "maximum" $-variables

[tueda]

The following code should print the maximum number of terms that one thread got, but tform often crashes:

Auto S x;
L F = x1+...+x100;  * Ensure well-parallelized.
.sort
#$n = 0;
$n = $n + 1;
ModuleOption maximum $n;
.sort
#message `$n'
.end

TFORM 4.1 (Nov 15 2016, v4.1-20131025-263-g033e880) 64-bits 4 workers  Run: Wed Nov 23 16:19:56 2016
    Auto S x;
    L F = x1+...+x100;  * Ensure well-parallelized.
    .sort

Time =       0.00 sec    Generated terms =        100
               F         Terms in output =        100
                         Bytes used      =       2808
    #$n = 0;
    $n = $n + 1;
    ModuleOption maximum $n;
    .sort
Program terminating in thread 1 at 1.frm Line 6 --> 
  0.00 sec + 0.00 sec: 0.01 sec out of 0.00 sec

Is it not expected to use $n = $n + 1 for the maximum option?

[vermaseren]

Hi

That is where you are supposed to use sum.
Maximum is for setting a fixed value.

Jos

On 23 nov. 2016, at 16:30, Takahiro Ueda notifications@github.com wrote:

The following code should print the maximum number of terms that one thread got, but tform often crashes:

Auto S x;
L F = x1+...+x100; * Ensure well-parallelized.
.sort
#$n = 0;
$n = $n + 1;
ModuleOption maximum $n;
.sort
#message `$n'
.end
TFORM 4.1 (Nov 15 2016, v4.1-20131025-263-g033e880) 64-bits 4 workers Run: Wed Nov 23 16:19:56 2016
Auto S x;
L F = x1+...+x100; * Ensure well-parallelized.
.sort

Time = 0.00 sec Generated terms = 100
F Terms in output = 100
Bytes used = 2808
#$n = 0;
$n = $n + 1;
ModuleOption maximum $n;
.sort
Program terminating in thread 1 at 1.frm Line 6 -->
0.00 sec + 0.00 sec: 0.01 sec out of 0.00 sec
Is it not expected to use $n = $n + 1 for the maximum option?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub #147, or mute the thread https://github.com/notifications/unsubscribe-auth/AFLxEg6jAg9lplY-0rq8wB272wYDyaJKks5rBFwSgaJpZM4K6uNx.

[tueda]

@vermaseren Do you mean the right-hand side of $-variable assignment for maximum/minimum should not depend on the variable itself? For example

$max = max_($max,$n);

is not allowed?

I see an example in the manual like

if ( count(a1,1) > $max ) $max = count_(a1,1);

for maximum. Is it guaranteed that $max has the same value both at the reading and writing? In other words, does the whole process of reading-comparison-writing acquire a lock? If this is not atomic, a data race occurs.

[vermaseren]

Actually you are probably right that the $error = $error+1; should have worked. It did work in convdia.
Maybe there is just something going wrong.

Jos

On 23 nov. 2016, at 17:21, Takahiro Ueda notifications@github.com wrote:

@vermaseren https://github.com/vermaseren Do you mean the right-hand side of $-variable assignment for maximum/minimum should not depend on the variable itself? For example

$max = max_($max,$n);
is not allowed?

I see an example in the manual like

if ( count(a1,1) > $max ) $max = count_(a1,1);
for maximum. Is it guaranteed that $max has the same value both at the reading and writing? In other words, does the whole process of reading-comparison-writing acquire a lock? If this is not atomic, a data race occurs.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub #147 (comment), or mute the thread https://github.com/notifications/unsubscribe-auth/AFLxEknCKxSIU0Wnre6FAnbHKw9zIaJ2ks5rBGgYgaJpZM4K6uNx.

[jodavies]

for maximum. Is it guaranteed that $max has the same value both at the reading and writing? In other words, does the whole process of reading-comparison-writing acquire a lock? If this is not atomic, a data race occurs.

For information: no it is not guaranteed that $max has the same value at the read and the write, but in the current code the write only happens if indeed the RHS value is larger than the current $max, and this part is protected by a lock.