Active Topics
-
Porting apps to Leste (32)
to Maemo 7 / Leste by Arno_11 - 1 day, 2 hrs ago -
[ANNOUNCE] CSSU-thumb thread - stable Thumb2 on N900 (2,264)
to Maemo 5 / Fremantle by panjgoori - 2 days, 20 hrs ago -
Full linux distros on Sailfish OS (254)
to SailfishOS by qoh - 3 days, 18 hrs ago - more...
|
2010-10-26
, 04:22
|
|
Posts: 1,746 |
Thanked: 2,100 times |
Joined on Sep 2009
|
#2
|
Originally Posted by champus
Because the CPU has to handle all traffic from USB/Bluetooth to the eMMC as fast as the eMMC can accept it, which ends up consuming a decent amount of CPU power. I suspect it's the same for the SD card.
why is the disk i/o performance that bad?
Are these card-like hard discs known for beeing slow? No matter if yes or no, there should be a way to make the kernel keep some ressources back when writing so that the device does respond as usual.
|
|
2010-10-26
, 06:00
|
|
Posts: 69 |
Thanked: 55 times |
Joined on Nov 2009
|
#3
|
Maybe a BFS enabled kernel could help at least to reduce the UI
lag. Power kernel crew want to try?
lag. Power kernel crew want to try?
|
|
2010-10-26
, 06:59
|
|
Posts: 1,425 |
Thanked: 983 times |
Joined on May 2010
@ Hong Kong
|
#4
|
You experienced lag when you are performing other tasks while transferring files in the background? That is outragous! If N900 could find a way to stop users from performing other tasks while an active task is running, your problem would be solved, and everybody is happy.
One can only dream....
Last edited by 9000; 2010-10-26 at 07:06.
One can only dream....
Last edited by 9000; 2010-10-26 at 07:06.
|
|
2010-10-26
, 07:15
|
|
Posts: 2,829 |
Thanked: 1,459 times |
Joined on Dec 2009
@ Finland
|
#5
|
why it lags?Because you are given option to multitask while those operations are running.Simple fix would be to force single tasking to end user while those operations are running and I believe that most of users would be happy because it would be just designed that way 
|
|
2010-10-26
, 07:40
|
|
Posts: 540 |
Thanked: 288 times |
Joined on Sep 2009
|
#6
|
Also: General flash memory is slow (the fast SSDs that use "flash" use 1. expensive, faster type 2. striping to parallelize access across many memory banks).
10Mbit/s ([min] speed for class 10 SD cards) is considered blazing fast write speed for flash, your basic cheap spinning-platters HD should be able to sustain 30MBytes/s easily (24x faster than the class 10 SD).
10Mbit/s ([min] speed for class 10 SD cards) is considered blazing fast write speed for flash, your basic cheap spinning-platters HD should be able to sustain 30MBytes/s easily (24x faster than the class 10 SD).
__________________
- Live near Helsinki, Finland & interested in electronics ? Check this out.
- Want anti-virus/firewall ? Read this (and follow the links, also: use the search, there are way too many threads asking the same questions over and over and over again).
- I'm experimenting with BitCoins, if you want to tip me send some to: 1CAEy7PYptSasN67TiMYM74ELDVGZS6cCB
|
|
2010-10-26
, 09:15
|
|
Posts: 196 |
Thanked: 224 times |
Joined on Sep 2010
@ Africa
|
#7
|
I wonder if this fix, which is included in 2.6.36 kernel, to address the problem described here, is relevant.
The last kernel I didn't see this issue on was 2.6.27, it was definitely present in 2.6.31, I think it may have been in 2.6.29, so ... is 2.6.28 affected?
(BTW, the topic could be improved, as I think the issue is about the interactivity of the OS during IO, not purely about the disk IO performance)
The last kernel I didn't see this issue on was 2.6.27, it was definitely present in 2.6.31, I think it may have been in 2.6.29, so ... is 2.6.28 affected?
(BTW, the topic could be improved, as I think the issue is about the interactivity of the OS during IO, not purely about the disk IO performance)
why is the disk i/o performance that bad?
Everytime I install something from the repos or I transfer files via USB or bluetooth the whole device lags. And yes, I'm pretty sure there's no special case with my device, I had some others to compare. Can't really believe why it seems like I'm the first in here that is going nuts of this bottleneck? I don't know much about the card the N900 uses as disc (what is it? mmc? ssd? microsd? lsd?) so could someone please explain how this can happen?
Are these card-like hard discs known for beeing slow? No matter if yes or no, there should be a way to make the kernel keep some ressources back when writing so that the device does respond as usual. I'm going to have a look at it, but I'm stil interested in your opinions or hints.
Greets