logo separator

[mkgmap-dev] 192GB recommended

From Gerd Petermann gpetermann_muenchen at hotmail.com on Tue Jul 28 21:15:20 BST 2020

Hi Andrzej,

yes, there is an error, see
http://gis.19327.n8.nabble.com/Error-in-memory-calculation-tt5957336.html

I didn't commit my patch because I have no experience with the internals.

Gerd

________________________________________
Von: mkgmap-dev <mkgmap-dev-bounces at lists.mkgmap.org.uk> im Auftrag von Andrzej Popowski <popej at poczta.onet.pl>
Gesendet: Dienstag, 28. Juli 2020 22:02
An: mkgmap-dev at lists.mkgmap.org.uk
Betreff: [mkgmap-dev] 192GB recommended

Hi,

I recently have noticed the following suggestion in mkgmap output:

"To reduce the run time, consider increasing the amnount of memory
available for use by mkgmap by using the Java -Xmx flag to set the
memory to more than 196700 MB, providing this is less than the amount of
physical memory installed."

I run mkgmap to create files for PC installation - overview map, index,
tdb. There is no actual compilation of any map. I guess, that condition
for triggering the message is wrong in my case and maybe the
calculations too. I mean this part of code in Main.java:

if ((taskCount > threadCount + 1) && (maxJobs == 0) && (threadCount <
runtime.availableProcessors())) {
                        System.out.println("To reduce the run time, consider increasing the
amnount of memory available for use by mkgmap by using the Java -Xmx
flag to set the memory to more than " + 100* (1 + ((runtime.maxMemory()
* runtime.availableProcessors()) / (threadCount * 1024 * 1024 * 100))) +
" MB, providing this is less than the amount of physical memory
installed.");
                }

Is "taskCount" correct, when input files are only *.img?

"threadCount" is probably forced to 1, because I see message: "Setting
max-jobs to 1". Then probably there is no need to multiply memory for
all processors.

I'm not sure about "runtime.maxMemory()". It looks like 24GB which I
have allocated with option -Xmx24G. Shouldn't it be a real memory used?

Well, all is not a problem really. I just found the message funny and
investigated a bit. Btw, there is a typo: "amnount".

--
Best regards,
Andrzej
_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev


More information about the mkgmap-dev mailing list