Active Topics

 


Reply
Thread Tools
Posts: 119 | Thanked: 110 times | Joined on Sep 2009 @ Prague
#21
guys, maemo is linux, you really don't have to worry about how the disk is partitioned - it sure as hell will be 'fixable' one way or the another (if not by repartitioning, then by using some loopback mount etc. etc.)..

I would consider this being a trivial problem - as soon as I have the device in my hands, I'll post some how-to.. But I'm quite sure this will be solved as soon as any other linux-user has the device in his hands ;-)
 
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#22
Oh, we're way beyond the question IF it can be done, the question is what's the easiest/best way to do it
 
chemist's Avatar
Administrator | Posts: 1,036 | Thanked: 2,019 times | Joined on Sep 2009 @ Germany
#23
Originally Posted by andree View Post
guys, maemo is linux, you really don't have to worry about how the disk is partitioned - it sure as hell will be 'fixable' one way or the another (if not by repartitioning, then by using some loopback mount etc. etc.)..

I would consider this being a trivial problem - as soon as I have the device in my hands, I'll post some how-to.. But I'm quite sure this will be solved as soon as any other linux-user has the device in his hands ;-)
and also prototype devices setup might be just that way as its for testing stuff... if id like (just example) to test speeds I would get a bunch of files in 25GB overall size and transfer them, to do that I would need to partition just like it was shown in the reviews, after I setup my testing device I need to prepare the SW for those prototypes (some lazy guy left on my desk this morning) to be shipped to those review guys today, so to save some time (I am lazy as well) I'd just copy the system I just setup for myself to those 20 N900 on my desk to get early out for lunch cause I will meet this adorable blonde if I get there by time! (and maybe the dishes are hot aswell )

do you get the point?
if its that way we need a howto to change it if we like but I honestly think it wont be...
 
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#24
Originally Posted by attila77 View Post
No, I didn't mean the physical flash, but the process - compression does not necessarily slow you down (if you can (de)compress faster than you can read/write, you're actually better off).
at least it eats a lot of CPU which is bad if the CPU time is needed somwhere else. Maybe with N900 cpu being faster this is not such big issue but with 700 and N8x0 the CPU hit is IMO significant. Also CPU usage = battery drain.

Originally Posted by attila77 View Post
When I discussed some loopfile-related projects with Nokia folks it has been suggested that they *did* benchmark it and the internal flash was significantly slower. I would reference it, but it was in an off-list reply so for now you will have to take my word for it unless the original author chimes in
would like to see some data because I don't believe it too. Both with 770 and N8x0 root on SD/MMC is faster than jffs2 in my experience. There are some specific use cases where it is slower (like using sqlite which does fsync after each database write see https://bugs.maemo.org/show_bug.cgi?id=1842#c13 ) but overall it feels faster to me. Even with swapping it shouldn't be so bad. As for benchmarks simple read only or write only tests should be definitely faster with sd/mmc. I never did some concurrent read/write benchmark but can see that this may indeed be a problem. But how often we see concurrent read and write except swapping?
Originally Posted by attila77 View Post
Try starting a larger Qt app on a cloned setup and you'll se the problem - it will read the Qt libs from the card and as it takes up memory start to swap on the same card, resulting in abysmal startup times and several seconds of total unresponsiveness.
swap is slow in any case. I don't think actively using more memory than available is good idea in any case. And BTW one can swap to (or boot from) other SD/MMC card if this is really a bottleneck in real life.

Originally Posted by attila77 View Post
This is also one of the main reasons Qt and Python are not optified in Fremantle - so they would not be on the same device as the swap.
Oh really? Hmm, I know for N900 they configured 756 MB for swap, this also feels very strange to me with only 256MB of real RAM. Planning to use so much swap feels somehow wrong to me. Well we'll see once we have real devices in our hands. I definitely plan to boot from something else than internal NAND.
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.
 

The Following 3 Users Say Thank You to fanoush For This Useful Post:
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#25
Originally Posted by fanoush View Post
would like to see some data because I don't believe it too.
Hey, don't shoot the messenger My sources are Antonio and Kimitake/Kimmo, so you can ask them for specific figures.
 
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#26
Originally Posted by attila77 View Post
Hey, don't shoot the messenger My sources are Antonio and Kimitake/Kimmo, so you can ask them for specific figures.
I'm not shooting anyone and I heard the 'mmc is slower' story in maemo-developers before (without any numbers as a proof). i do believe there are issues with concurrent read/write mmc access, ext3 filesystem and also other significant issues (like firmware flashing) which caused Nokia to stay away from having rootfs on mmc but still it does not mean rootfs on MMC is bad idea or is slower in general.
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.
 
HangLoose's Avatar
Posts: 319 | Thanked: 289 times | Joined on Sep 2009 @ Lisboa, Portugal
#27
There should be a option in the phone of how much memory (between max and min) one wants to allocate to applications. PERIOD.

Maybe some noble soul will develop something as straight forward as a panel with a slider...

IMHO, this is unacceptable that a mobile computer should have this kind of restriction.
 
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#28
I wish I could just post that mail over here to avoid getting anecdotal...
 
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#29
Originally Posted by HangLoose View Post
There should be a option in the phone of how much memory (between max and min) one wants to allocate to applications. PERIOD.
It's not that easy. The space is provided by two (or three) different devices with different underlying technologies, filesystems, etc. ANY choice made will be a compromise. A slider can be made (in fact, there is one, called AppSpace, by yours truly ), but it doesn't change the fact there is no universally superior solution, the question is only what are you willing to sacrifice.
 

The Following User Says Thank You to attila77 For This Useful Post:
Posts: 3,841 | Thanked: 1,079 times | Joined on Nov 2006
#30
Originally Posted by HangLoose View Post
There should be a option in the phone of how much memory (between max and min) one wants to allocate to applications. PERIOD.

Maybe some noble soul will develop something as straight forward as a panel with a slider...
That's something you can do if you have a PDA with memory, as in battery-backed RAM (e.g. windows mobile PDAs). That concept makes no sense at all when you have proper filesystems, shared between different physical parts. On a WinCE device like I described above you have to reduce actual memory (as in RAM, not as in mass storage) to make room for applications.

IMHO, this is unacceptable that a mobile computer should have this kind of restriction.
You haven't read the thread. This is all a red herring. There will be no problem getting enough space for applications.
__________________
N800/OS2007|N900/Maemo5
-- Metalayer-crawler delenda est.
-- Current state: Fed up with everything MeeGo.
 
Reply

Tags
rootfs, space, speedy deletion candidate


 
Forum Jump


All times are GMT. The time now is 08:03.