1. Excessive disc activity
- Posted by FredRansom Sep 04, 2009
- 1439 views
The following code failed to execute properly with eui.exe. It initiated so much disc activity that I had to reboot to stop it. After rebooting, I ran it through the Watcom compiler. It executed to completion, but slower than it should have due to excexxive disc activity. This is Euphoria 4 beta 2 running on Vista.
sequence res procedure Calc_it () atom sum atom beginning, ending puts (1, "Starting\n") sum = 0.0 beginning = time() for x = 0.0 to 10000.0 do sum = sum + cos(x) for y = 0.0 to 10000.0 do sum += sin(y) end for end for ending = time() res = sprintf("%d %g", {ending - beginning, sum}) puts(1, res) end procedure Calc_it () integer x x = getc(0)
2. Re: Excessive disc activity
- Posted by DerekParnell (admin) Sep 04, 2009
- 1386 views
The following code failed to execute properly with eui.exe ...
It looks like you found a bug. The interpreter is not reusing RAM during the calculations. Instead it is continuing to acquire more RAM for each iteration of the loops to store the temporary results of the calculations in.
However, I found no fault with the translated code.
And by the way, it took 43 seconds on my machine to do 100,010,000 sin/cos calculations, and the resulting sum was 16340.3; what did you get?
Intel Core 2 6600 at 2.4GHz with 2GB RAM, Windows XP SP3.
3. Re: Excessive disc activity
- Posted by mattlewis (admin) Sep 04, 2009
- 1397 views
The following code failed to execute properly with eui.exe ...
It looks like you found a bug. The interpreter is not reusing RAM during the calculations. Instead it is continuing to acquire more RAM for each iteration of the loops to store the temporary results of the calculations in.
However, I found no fault with the translated code.
Yeah, they do things a little differently. It looks like I missed some ops that allocate memory into temps in the interpreter.
Matt
4. Re: Excessive disc activity
- Posted by jeremy (admin) Sep 04, 2009
- 1343 views
With the new fix in the SVN repo, it ran interpreted and completed in 12 seconds on my machine, 2ghz core duo.
You can get the latest eubin at: http://jeremy.cowgar.com/eubins.
Jeremy
5. Re: Excessive disc activity
- Posted by useless Sep 04, 2009
- 1361 views
- Last edited Sep 05, 2009
<moderator: jeremy>
We are not going to get into verbal arguments and assaults. More posts of this nature will be removed as well. If you wish to discuss it you can send a private email.
</moderator>
6. Re: Excessive disc activity
- Posted by FredRansom Sep 05, 2009
- 1320 views
The code translated with EU 4and compiled with Watcom took 17 secs. EU 3 interpreted took 17 secs. Translated and compiled with EU 3 and Watcom took 19 secs. These were on a single core 1.6 mHz laptop.
7. Re: Excessive disc activity
- Posted by useless Sep 05, 2009
- 1263 views
<moderator: jeremy>
We are not going to get into verbal arguments and assaults. More posts of this nature will be removed as well. If you wish to discuss it you can send a private email.
</moderator>
<moderator: Derek>
Parts of this message were not appropriate and have been removed.
</moderator>
There is nothing to discuss. Matt had decided that code in flux, code evolving from a very poor example of how to use Euphoria v4 task scheduler and sockets code <edited was not a proper demonstration of Euphoria code>. I was simply not deleting all the preexisting code with each release, and i stated repeatedly that news*.ex was preferably to be a demonstration, a series of news*.ex files, of how to evolve from the original into a faster and more streamlined application. If none of the news*.ex had any code shared with the previous version number, how would it show the evolution?
useless
8. Re: Excessive disc activity
- Posted by jeremy (admin) Sep 05, 2009
- 1291 views
<moderator: jeremy>
If you wish to discuss it you can send a private email.
</moderator>
There is nothing to discuss....<snip>
If you wish to discuss this matter, please send a private email. Do not hijack this thread. Another post off topic about this issue on this thread will simply be removed.
Jeremy
9. Re: Excessive disc activity
- Posted by alanjohnoxley Sep 06, 2009
- 1307 views
Hi Jeremy, I downloaded the latest SVN windows compiled version as per your link. The file indicates its update 2700, but if you execute eui it says v2670. Small issue I know, but I'm sure you don't want to fix something that isn't broke, due to wrong version being indicated ;)
Cheers Alan
10. Re: Excessive disc activity
- Posted by jeremy (admin) Sep 06, 2009
- 1278 views
I downloaded the latest SVN windows compiled version as per your link. The file indicates its update 2700, but if you execute eui it says v2670. Small issue I know, but I'm sure you don't want to fix something that isn't broke, due to wrong version being indicated ;)
Alan,
I'm sorry about that. Eubins on Windows is semi-manual. I probably goofed something up. I just uploaded rev 2708. It should be correct. Can you give it a try?
Jeremy
11. Re: Excessive disc activity
- Posted by alanjohnoxley Sep 06, 2009
- 1243 views
OK, I fetched v2708, eui is dated 20090905 09:39 pm and has crc32 of D65204DF executing the sample code with this eui gave: CleanUpError malloc failed , here is the ex.err c:\euphoria\demo\testing\test.ex:12 in procedure Calc_it() CleanUpError malloc failed
sum = 10879.78009 beginning = 0.032 ending = <no value> x = 6658 y = 9366
... called from c:\euphoria\demo\testing\test.ex:22
Global & Local Variables
c:\euphoria\demo\testing\test.ex: res = <no value> x = <no value> My dos prompt command line was, from dir c:\euphoria\bin: (where the latest SVN binaries were replaced) eui c:\euphoria\demo\testing\test.ex It took 17 seconds to fail on my WinXp sp3, 2.5Ghz core2 duo. No excessive HDD activity noticed. eui -version still says (r2670) BTW :)
12. Re: Excessive disc activity
- Posted by alanjohnoxley Sep 06, 2009
- 1186 views
OOPS, sorry scrap the previous post. The darn winzip unzipped the binaries to another subdirectory. Same with v2700. Tried again, eui -version says its definitely v2708 this time, but the test says Fatal run-time error: bad file number (-1). Funny.. the test code does not use any explicit files.
13. Re: Excessive disc activity
- Posted by jeremy (admin) Sep 06, 2009
- 1195 views
OOPS, sorry scrap the previous post. The darn winzip unzipped the binaries to another subdirectory. Same with v2700. Tried again, eui -version says its definitely v2708 this time, but the test says Fatal run-time error: bad file number (-1). Funny.. the test code does not use any explicit files.
alan,
I'm sorry. This is a bug I introduced in 2708 and it wasn't caught till another developer was doing some tests. I am away from my computer right now, so I will not be able to publish a corrected binary for another few hours. Sorry. As soon as I get home, however, I'll put up a new, working eubin. Normally eubin releases are not this difficult, you just got in on a bad run
Jeremy
14. Re: Excessive disc activity
- Posted by alanjohnoxley Sep 06, 2009
- 1229 views
Hey, please don't apologise, I'm the dufus here not you! I was just trying to help with testing. No need or rush on my part. BTW, keeping track and testing on different combinations of OS's and svn's must be fun. Do you use VmWare or something similar, to have canned testbeds?
Regards Alan
15. Re: Excessive disc activity
- Posted by jeremy (admin) Sep 06, 2009
- 1226 views
Hey, please don't apologise, I'm the dufus here not you! I was just trying to help with testing. No need or rush on my part. BTW, keeping track and testing on different combinations of OS's and svn's must be fun. Do you use VmWare or something similar, to have canned testbeds?
Ok, 2714 is up and should be good.
We test mainly on Windows and Linux, then other platforms get a work out once in a while with unit tests. Some of us call Windows our development home while others Linux, so those are tested pretty well by the devs on a daily basis.
Jeremy