Guest us1111 Posted July 6, 2010 Report Posted July 6, 2010 (edited) You didn't test random writing only sequential and thats not comparable. Did you? So are you 100% sure that's it (I mean the root cause of the stalling..)?? Edited July 6, 2010 by us1111
Guest mobiholic Posted July 6, 2010 Report Posted July 6, 2010 (edited) Is the fast NAND I keep hearing about the 8 / 16GB internal memory or is it something separate? Edited July 6, 2010 by mobiholic
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 MCR r2 has the Lag Fix available in the 'MCR Scripts' app, no more messing with ADB. :) P
Guest JuW Posted July 6, 2010 Report Posted July 6, 2010 MCR r2 has the Lag Fix available in the 'MCR Scripts' app, no more messing with ADB. :) P Wow! What an improvement! So faaast! :D but I have a few noobie questions about this fix. I don`t totally understand, when people say, that if or when the memory gets full, there will be problems... What does this mean getting memory full? Does it ever clear it or it will gather that data to the memory as long as it`s full? How to make it empty again... Sorry, but I`m little bit confused...
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 I'm playing around with repartitioning the internal SD atm. In stock form it's 2GB RFS16 for /data and 6GB RFS32 for /sdcard. I am thinking something like 1GB RFS16 for /data, 1GB EXT3 for /data/data and 6GB RFS32 for /sdcard to see if EXT3 performs better than RFS (in case RFS is the bottleneck?) Not going well so far, device won't boot. :) P
Guest woutf Posted July 6, 2010 Report Posted July 6, 2010 Ouch, hope you can get it running again. What filesystem is /dbdata using btw?
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 Heh, it's alive again now :) /dbdata is RFS too (on the fast NAND ofc)... P
Guest webcrtor Posted July 6, 2010 Report Posted July 6, 2010 Why does it say "cp: not found" when I do the cp command? :s
Guest jmmL Posted July 6, 2010 Report Posted July 6, 2010 Heh, it's alive again now :) /dbdata is RFS too (on the fast NAND ofc)... P Just ordered one of these after seeing all the work you're putting in (and other devs as well) - I was worried that no-one would decide to support it. Just out of curiosity - by RFS are you referring to ReiserFS?
Guest zenkinz Posted July 6, 2010 Report Posted July 6, 2010 Heh, it's alive again now :) /dbdata is RFS too (on the fast NAND ofc)... P still don't get it why is there a extra nand chip ... surely samsung won't be giving us free thing w/o adding to its credit.
Guest soulcrash Posted July 6, 2010 Report Posted July 6, 2010 Hi guys, a hack that brings the galaxy to life :) Can anyone tell me where i can see how much of the 130mb i already use? thx :D
Guest pekkala89 Posted July 6, 2010 Report Posted July 6, 2010 Why does it say "cp: not found" when I do the cp command? :s I had to use busybox in front of cp. Like this: busybox cp ...
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 Just ordered one of these after seeing all the work you're putting in (and other devs as well) - I was worried that no-one would decide to support it. Just out of curiosity - by RFS are you referring to ReiserFS? No, RFS is Samsung variant of FAT for OneNAND. P
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 still don't get it why is there a extra nand chip ... surely samsung won't be giving us free thing w/o adding to its credit. They're not. There isn't. P
Guest Paul Posted July 6, 2010 Report Posted July 6, 2010 Hi guys, a hack that brings the galaxy to life :) Can anyone tell me where i can see how much of the 130mb i already use? thx :D 'adb shell df' or 'adb shell busybox df -h' P
Guest mobiholic Posted July 6, 2010 Report Posted July 6, 2010 They're not. There isn't. P This is confusing but it seems like it should be straight forward. The only NAND in the device is the one moviNAND chip and it is split to allow 2GB for the rom and the rest for the user, is that right so far? Assuming that is correct then it sounds like the speed issues which are contributing to the stalling are due to folder sizes on the NAND and not a result of the speed of the NAND which is not a problem now and is not going to be an issue in the future. Does this cover the problem that we are experiencing when it comes to the stalling or have I got it all upside down and back to front?
Guest Xoneso Posted July 6, 2010 Report Posted July 6, 2010 Paul. I was sitting looking though init.rc and found an line that was commented out. # mount rfs /dev/block/stl7 /data nosuid nodev check=no anyway to check what size /dev/block/stl7 would be and if it's something we would be able to use? Could only find something with stl7 that one time in init.rc
Guest Kilack Posted July 7, 2010 Report Posted July 7, 2010 (edited) Paul. I was sitting looking though init.rc and found an line that was commented out. # mount rfs /dev/block/stl7 /data nosuid nodev check=no anyway to check what size /dev/block/stl7 would be and if it's something we would be able to use? Could only find something with stl7 that one time in init.rc might be tbe missing megs that make up the 512.... :) Edited July 7, 2010 by Kilack
Guest lgkahn Posted July 7, 2010 Report Posted July 7, 2010 mine wouldnt boot either.. thought i bricked it.. had a hell of a time getting it to come up with boot loader.. i tried to replace the mms.apk and mms.odex from the earlier jf3 build.. to get around the messaging not sending bug with jg1... it was more than just alternate keyboards the internal keyboard at least in my case would get send failed if i didnt include a header so i concluded that the bug was in the mms/sms app since hanset and other apps worked fine. the phone didnt like that at all :) boot loop...
Guest Paul Posted July 7, 2010 Report Posted July 7, 2010 Paul. I was sitting looking though init.rc and found an line that was commented out. # mount rfs /dev/block/stl7 /data nosuid nodev check=no anyway to check what size /dev/block/stl7 would be and if it's something we would be able to use? Could only find something with stl7 that one time in init.rc Interesting, stl7 doesn't appear to be being used... not managed to mount it yet tho! P
Guest rasp Posted July 7, 2010 Report Posted July 7, 2010 (edited) Interesting, stl7 doesn't appear to be being used... not managed to mount it yet tho! P if im not wrong.. there is a total of 2GB oneNAND space on this moviMCP chip that Samsung uses.. It has 16GB moviNAND and 16G bit (2GB) of oneNAND for random access.. 1.4GB is accounted for, another 600MB is still unknown.. since this moviMCP is configurable which is confusing but the moviNAND is sequential, thats why is slower.. OneNAND has NOR interface which makes it suitable for random access and can hv 20MB/s transfer speed so the stl blocks should be sitting on it.. Edited July 7, 2010 by rasp
Guest Paul Posted July 7, 2010 Report Posted July 7, 2010 if im not wrong.. there is a total of 2GB oneNAND space on this moviMCP chip that Samsung uses.. It has 16GB moviNAND and 16G bit (2GB) of oneNAND for random access.. 1.4GB is accounted for, another 600MB is still unknown.. since this moviMCP is configurable which is confusing but the moviNAND is sequential, thats why is slower.. OneNAND has NOR interface which makes it suitable for random access and can hv 20MB/s transfer speed so the stl blocks should be sitting on it.. Pity it's not then! P
Guest ssarl Posted July 7, 2010 Report Posted July 7, 2010 Guys, I see that the file system is mounted without noatime option. Could this cause the problem with the delays?
Guest rasp Posted July 7, 2010 Report Posted July 7, 2010 Pity it's not then! P Hope someone can figure this out: http://www.samsung.com/global/business/sem...4_3_support.pdf KLMAG8DEDD http://www.careace.net/wp-content/uploads/...assembly-14.jpg
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now