cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Showing results for 
Search instead for 
Did you mean: 

Community Tip - Need to share some code when posting a question or reply? Make sure to use the "Insert code sample" menu option. Learn more! X

Memory leak

ptc-1368288
1-Newbie

Memory leak

Working a day long on work sheets that do "pseudocompile" well enough, only when the box is tired it needs foods ... "Windows is low in memory". Whether the memory is full of information or filled with pointers, better experts could precise. Over the years, I observed that some type of work would starve the machine sooner, but that's not much of a trace evidence. It became a proof while working on that session for Kristjan. It's a two pages sheet doing apparently nothing or so little versus an 80 pages collection, but, the but is there. In 2003, Robert wrote that recursive should be avoided as much as possible... quite true.
What that work sheet does: few peanuts CFFT, ICFFT and DeCasteljau 6 pairs recursive. From a fresh start [not restart], the recursive starts getting visibly slow at about � dozen modifications of the points. Then the hyperfast CFFT/ICFFT start getting slow, very slow. The end of the story is that the "Run time error" is about to display and in few more steps of curve fitting, the all damned box plugs. The CFFT/ICFFT consume little memory at 200 points. By itself, the recursive just builds two more peanut vectors.
All what it says it that this particular recursive "hide and seek" every time in a new place, harder and harder to retrieve at each run up to the point that there is no way to retrieve at all.
Most if not all memory leaks were reported about Mathcad 14. Conclusively, the very Mathcad inner structure is not suitable for recursive. Conclusively again, it could very well be that the attractive recursive has been over implemented in that version. Or in two words: which one make the other suffer ... Mathcad==>Windows or vice versa ?

Let's take it factual, comparing two Mathcad sessions. Working a day long [in fact several days] searching for the Rayleigh model, how many times the Given/Minerr was used ... 100's of times ... Mathcad never complained neither the box crashed. From fresh start, in � dozen fitting the recursive NURBS ... the session ends. The vectors of values are totally negligible but the memory has like vanished in the reported words from collabs ... "memory leak".

I wrote this long observation because no time to write it shorter.

jmG
10 REPLIES 10

I have a fixed but high windows swap file, with this file open I have received a message of "Not enough memory" in mathcad 14 (then mathcad not crashes, only accept to edit but without evaluating nothing, situation that I never found before) and a (hope uncesefull, in a first check) try of windows xp to increase the swap file when using mathcad 11. It's noticeable that this makes crazy my norton antivirus.

So, I can confirm the assertion that this file it's an interesting example for those guys interested in analyze mathcad memory issues -this is, not for me.

Regards. Alvaro.
RichardJ
19-Tanzanite
(To:ptc-1368288)

The memory leak with CFFT is a long standing bug:

http://collab.mathsoft.com/read?50763,11e#50763

I have a vague recollection it's fixed in the latest versions of 14, but I'm not 100% sure of that.

Richard

That file processes fairly quickly for me in Mathcad 14 m030 and recreating it in Mathcad Prime 1.0 Alpha. I didn't run out of memory running both of them at the same time.

Mona

On 4/16/2010 11:01:40 AM, MonaZ wrote:
>That file processes fairly
>quickly for me in Mathcad 14
>m030 and recreating it in
>Mathcad Prime 1.0 Alpha. I
>didn't run out of memory
>running both of them at the
>same time.
>
>Mona
...............................

On 4/16/2010 8:38:27 AM, rijackson wrote:
>The memory leak with CFFT is a
>long standing bug:
>
>http://collab.mathsoft.com/rea
>d?50763,11e#50763
>
>I have a vague recollection
>it's fixed in the latest
>versions of 14, but I'm not
>100% sure of that.
>
>Richard
_____________________________

Thanks Richard & Mona.

I will try to make it "fft" instead .

It is true that rarely CFFT is called in a session of trial/error design.

Jean



RichardJ
19-Tanzanite
(To:RichardJ)

On 4/16/2010 8:38:27 AM, rijackson wrote:

>I have a vague recollection
>it's fixed in the latest
>versions of 14, but I'm not
>100% sure of that.

I finally decided to put in the 2 minutes worth of effort required to check. It's actually fixed even in MC13. I don't know about MC12 though, because I don't have it installed (because even if this is fixed, everything else is broken!).

Richard

I recently had a semi-repeatable memory problem with M14 020. I was importing/pasting tables with 10K entries. The total storage should not have overwhelmed my system - the largest of the final files are on the order of 600K - but several times during the process of building the work sheet Mathcad crashed with memory problems. Never did figure what the pattern was...

Rich
http://www.downeastengineering.com/

Rich,

The best way to work those things is to read file, type = for the display, select/copy selection, paste in a Mathcad table. It looks the way you do, Mathcad/Windows is looking for some piece of information ... filling the memory full of pointers and zap it all.
An excellent transition is ORIGINLAB.

Jean

Dear jean,
This sound similar to what I was doing. I would select Insert>Data>Table, enter a name for the table, copy the data from an editor to the clipboard, r-click the table and select Paste Table. Some combination of editing commands that I can't recall caused Mathcad 14 020 to crash with a Out-of-Memory error.

Rich
http://www.downeastengineering.com/

Don't name the data you import by Table until you are finished. But it won't paste elsewhere. Only thing you can do is hand enter data from your notes on paper. That will stick to your worksheet.

On 4/21/2010 8:57:01 AM, woodwise wrote:
>Dear jean,
>This sound similar to what I
>was doing. I would select
>Insert>Data>Table, enter a
>name for the table, copy the
>data from an editor to the
>clipboard, r-click the table
>and select Paste Table. Some
>combination of editing
>commands that I can't recall
>caused Mathcad 14 020 to crash
>with a Out-of-Memory error.
>
>Rich
>http://www.downeastengineering
>.com/
_______________________________

My dear Rich !
may be mishandling sometimes. Click on the input table [menu], paste table, name the table after... not name the table and paste-table like you seem doing, but that's Mathcad 11. Sometimes I "paste" instead of "paste table" but Mathcad 11 does not crash for that. My data editor and transit is ORIGINLAB 7.5 ... also as a graph producer because of it's editing facilities like minimal arrows ... etc. It used to sell around US $ 650. Not cheap but very useful about filtering and fitting, though the automatic fitting does not work anymore after installing HP Officejet J4580.

I can still fit by manually guessing like it was in ORIGINLAB 6.0

Jean




Announcements

Top Tags