AccelerEyes Forums GPU computing with Jacket for use with MATLAB® & ArrayFire for use with C, C++, Fortran, and Python http://forums.accelereyes.com/forums/smartfeed.php?feed_type=ATOM1.0&limit=1_DAY&sort_by=standard&feed_style=HTML& 2012-05-17T11:33:25+00:00 FeedCreator 1.8.0-dev (info@mypapit.net) Programming & Development with Jacket for use with MATLAB® :: Re: Fatal Matlab Error and Jacket Memory Management Issues :: Reply by pavanky 2012-05-16T20:33:47+00:00 2012-05-16T20:33:47+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=7&t=4022&p=11921#p11921 pavanky
Any chance we can have a look at the code ? I am guessing if you do geval(A); gysnc. before you do double(..) it would fail at gsync.
double(..) is a blocking operation so it waits until all previous gpu operations (related to A) are done. This may mean something else is broken and is being caught very late when doing double.

It would really be helpful if we can have a look at the code (or the smallest piece of code segment that will show this behavior) to figure this out.]]>
Any chance we can have a look at the code ? I am guessing if you do geval(A); gysnc. before you do double(..) it would fail at gsync.
double(..) is a blocking operation so it waits until all previous gpu operations (related to A) are done. This may mean something else is broken and is being caught very late when doing double.

It would really be helpful if we can have a look at the code (or the smallest piece of code segment that will show this behavior) to figure this out.]]> Programming & Development with Jacket for use with MATLAB® :: Re: Fatal Matlab Error and Jacket Memory Management Issues :: Reply by mjs454 2012-05-17T02:52:29+00:00 2012-05-17T02:52:29+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=7&t=4022&p=11923#p11923 mjs454 Thanks,
Mike

x = function1(....); % 3000x3000 gdouble complex
xCPU = double(x);
y = function2(x, ...); % y is 3000x3000x10 gdouble complex]]> Thanks,
Mike

x = function1(....); % 3000x3000 gdouble complex
xCPU = double(x);
y = function2(x, ...); % y is 3000x3000x10 gdouble complex]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: arrayfire 1.1 python malloc error :: Reply by pavanky 2012-05-16T20:05:38+00:00 2012-05-16T20:05:38+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4016&p=11918#p11918 pavanky
can you provide a sample code that is causing this problem ?]]>
can you provide a sample code that is causing this problem ?]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: Using arrays in ArrayFire :: Reply by pavanky 2012-05-16T20:10:28+00:00 2012-05-16T20:10:28+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4014&p=11919#p11919 pavanky
two things.

- ArrayFire is zero based. so 1:15 would be 0:14.
- To get sequences, you qill have to use seq


seq can be used in two ways.

1) seq(start, end) will produce start, start+1, start+2, ..., end.
example: start(3, 5) == 3, 4, 5. seq(0, 4) == 0, 1, 2, 3, 4.
2) seq(elements) will produce 0, 1, 2, 3, 4, .... end - 1;
example seq(10) == 0, 1, 2, 3, 4, 5, ... 9;

so in your sampe code it can be one of the following
- dopmag(span, span, seq(0, zsamples - 1));
- dopmag(span, span, seq(zsamples));]]>
two things.

- ArrayFire is zero based. so 1:15 would be 0:14.
- To get sequences, you qill have to use seq


seq can be used in two ways.

1) seq(start, end) will produce start, start+1, start+2, ..., end.
example: start(3, 5) == 3, 4, 5. seq(0, 4) == 0, 1, 2, 3, 4.
2) seq(elements) will produce 0, 1, 2, 3, 4, .... end - 1;
example seq(10) == 0, 1, 2, 3, 4, 5, ... 9;

so in your sampe code it can be one of the following
- dopmag(span, span, seq(0, zsamples - 1));
- dopmag(span, span, seq(zsamples));]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: How doe interp() work? :: Reply by pavanky 2012-05-16T20:21:06+00:00 2012-05-16T20:21:06+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4020&p=11920#p11920 pavanky
This is a bug in interp. We are fixing this asap. When fixed, interp should work based on indices provided by hAi and not make any assumptions.]]>
This is a bug in interp. We are fixing this asap. When fixed, interp should work based on indices provided by hAi and not make any assumptions.]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: max() function's result is not correct/desired? :: Reply by pavanky 2012-05-16T20:53:26+00:00 2012-05-16T20:53:26+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=3922&p=11922#p11922 pavanky
This is a bug and is being fixed.]]>
This is a bug and is being fixed.]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: How best to initialize an array? :: Reply by pavanky 2012-05-16T19:57:41+00:00 2012-05-16T19:57:41+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4021&p=11917#p11917 pavanky
The shortcut is failing because the compiler can not figure out the data type of {1, 2, 3, 4, 5};

If you specify the data type in the following manner, it should work.

Code: Select all
af::array myArray(1, 5, (float[]){1, 2, 3, 4, 5});
    print(myArray);


myArray =
1.0000 2.0000 3.0000 4.0000 5.0000
]]>
The shortcut is failing because the compiler can not figure out the data type of {1, 2, 3, 4, 5};

If you specify the data type in the following manner, it should work.

Code: Select all
af::array myArray(1, 5, (float[]){1, 2, 3, 4, 5});
    print(myArray);


myArray =
1.0000 2.0000 3.0000 4.0000 5.0000
]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: How best to initialize an array? :: Reply by xiongtx 2012-05-17T13:38:59+00:00 2012-05-17T13:38:59+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4021&p=11927#p11927 xiongtx
Code: Select all
Error: type name not allowed


error.]]>
Code: Select all
Error: type name not allowed


error.]]> Programming & Development with ArrayFire for C, C++, Fortran, and Python :: Re: How best to initialize an array? :: Reply by malcolm 2012-05-17T15:30:10+00:00 2012-05-17T15:30:10+00:00 http://forums.accelereyes.com/forums/viewtopic.php?f=17&t=4021&p=11928#p11928 malcolm
Code: Select all
float ha[] = { 1, 2, 3, 4 };
array a(2, 2, ha);
]]>
Code: Select all
float ha[] = { 1, 2, 3, 4 };
array a(2, 2, ha);
]]>