Page 42 of 55 FirstFirst ... 32373839404142434445464752 ... LastLast
Results 821 to 840 of 1081
Like Tree13Likes
  1. #821  
    Quote Originally Posted by rwhitby View Post
    The Meta-Doctor now has a SWAP_PARTITION_SIZE feature.

    -- Rod
    Hot damn! Thanks, Rod. You're the man. I'm building a new doctor as we speak.
  2. #822  
    That is great news. I knew Carrel had mentioned it but never heard anything again. Meta dr. Always makes life better. Ltm ltm.

    -- Sent from my Palm Pre using Forums
  3. #823  
    Update to the sysctl 'game changer' settings.

    I'm probably going to make this into an .ipk tomorrow.

    Code:
    vm.swappiness = 90
    vm.vfs_cache_pressure = 40
    vm.dirty_expire_centisecs = 1200
    vm.dirty_writeback_centisecs = 500
    vm.dirty_background_ratio = 5
    vm.dirty_ratio = 90
    vm.min_free_kbytes = 1024
    vm.max_map_count = 131072
    vm.overcommit_memory = 1
    vm.overcommit_ratio = 70
    vm.zone_reclaim_mode = 2
    kernel.shmmax = 268435456
    kernel.shmall = 16777216
    sched_compat_yield = 1
    vm.page-cluster = 4
    vm.lowmem_reserve_ratio = 1024






    -- Sent from my Palm Pre using Forums
    Last edited by rmausser; 07/07/2011 at 02:23 PM.
  4. #824  
    Quote Originally Posted by rmausser View Post
    Update to the sysctl 'game changer' settings.

    I'm probably going to make this into an .ipk tomorrow.

    You do know that some of those paramters don't even exist, right? Why would you package up something that causes errors?

    Aside, perhaps using it for more than one day of uptime should be tested first...


    M.
  5. angiest's Avatar
    Posts
    933 Posts
    Global Posts
    952 Global Posts
    #825  
    Quote Originally Posted by rmausser View Post
    Update to the sysctl 'game changer' settings.

    I'm probably going to make this into an .ipk tomorrow.

    Code:
    vm.swappiness = 90
    vm.vfs_cache_pressure = 40
    vm.dirty_expire_centisecs = 1200
    vm.dirty_writeback_centisecs = 500
    vm.dirty_background_ratio = 5
    vm.dirty_ratio = 90
    vm.min_free_kbytes = 1024
    vm.max_map_count = 131072
    vm.overcommit_memory = 1
    vm.overcommit_ratio = 70
    vm.zone_reclaim_mode = 2
    kernel.shmax = 268435456
    kernel.shmall = 16777216
    #kernel.shmmni = 32768
    sched_compat_yield = 1
    vm.page-cluster = 4
    vm.lowmem_reserve_ratio = 1024






    -- Sent from my Palm Pre using Forums
    Is kernel.shmax correct? I have kernel.shmmax (granted, this is a Pre - with 1.4.5.)
  6. #826  
    Quote Originally Posted by angiest View Post
    Is kernel.shmax correct? I have kernel.shmmax (granted, this is a Pre - with 1.4.5.)
    whoops! fixed.
  7. #827  
    Quote Originally Posted by Xanadu73 View Post
    You do know that some of those paramters don't even exist, right? Why would you package up something that causes errors?

    Aside, perhaps using it for more than one day of uptime should be tested first...


    M.
    Which ones do not exist? I thought I weeded those out.
  8. #828  
    Quote Originally Posted by angiest View Post
    Is kernel.shmax correct? I have kernel.shmmax (granted, this is a Pre - with 1.4.5.)
    sorry, but i thought all these tweaks are for os 2.0 and above. Quite a few threads and hard to follow at this point. I thank rma for what he/she is doing, and I hope this turns into a nice wiki one day.
  9. #829  
    Quote Originally Posted by NABRIL15 View Post
    sorry, but i thought all these tweaks are for os 2.0 and above. Quite a few threads and hard to follow at this point. I thank rma for what he/she is doing, and I hope this turns into a nice wiki one day.
    I appreciate the thanks, but I had an operation and am now just a guy, not a he/she.



    Umm, I'll be making a wiki on the weekend.





    -- Sent from my Palm Pre using Forums
  10. angiest's Avatar
    Posts
    933 Posts
    Global Posts
    952 Global Posts
    #830  
    Quote Originally Posted by NABRIL15 View Post
    sorry, but i thought all these tweaks are for os 2.0 and above. Quite a few threads and hard to follow at this point. I thank rma for what he/she is doing, and I hope this turns into a nice wiki one day.
    I don't know what version of kernel is running on webOS 2+, but I doubt there are serious vm system sysctl variable changes between 2.6.24 (from webOS 1) and a generic 2.6.39 kernel, the most recent I am aware of in the 2.6 family. Most of these parameters should work just fine on 1.x, although the shmmax, etc, should probably be reviewed on a Pre - versus a Pre +, Pre 2, etc.
  11. #831  
    Quote Originally Posted by angiest View Post
    I don't know what version of kernel is running on webOS 2+, but I doubt there are serious vm system sysctl variable changes between 2.6.24 (from webOS 1) and a generic 2.6.39 kernel, the most recent I am aware of in the 2.6 family. Most of these parameters should work just fine on 1.x, although the shmmax, etc, should probably be reviewed on a Pre - versus a Pre +, Pre 2, etc.
    I can at least confirm off the top of my head that zone_reclaim is not on the Pixi with 1.4.5, so there are some differences present.

    I also think I found the answer to the previous question on compcache, but I still want to read up more before I post an answer. What I do have a question on though is what version of compcache is included with the Uberkernel releases? Does anyone know off the top of their heads?
  12. ewarhol's Avatar
    Posts
    11 Posts
    Global Posts
    126 Global Posts
    #832  
    I'm new this thread. Is there a wiki page or step by step directions on getting started? I started reading at posting # 411, but it was over my head for getting going.
  13. #833  
    Quote Originally Posted by ewarhol View Post
    I'm new this thread. Is there a wiki page or step by step directions on getting started? I started reading at posting # 411, but it was over my head for getting going.
    I am making a wiki and .ipk over the weekend.
  14. #834  
    Quote Originally Posted by Xanadu73 View Post
    You do know that some of those paramters don't even exist, right? Why would you package up something that causes errors?

    Aside, perhaps using it for more than one day of uptime should be tested first...


    M.
    which ones dont exist? i would like to get them out of there.
    thanks
  15. #835  
    Quote Originally Posted by graffix31 View Post
    which ones dont exist? i would like to get them out of there.
    thanks

    Sorry guys. P|C sometimes doesn't feel like sending me mails. I'm wrong also. I'm not sure what I was thinking on a couple of them (there was the typo as well). It all looks OK, actually.

    Code:
    root@My Pre:/media/internal# for THING in $(cat sysctl.test | awk '{print $1}' | sed -e 's/\#//g' -e 's/vm\.//g') ; do find /proc -type f -iname "*$THING*" -exec ls -l '{}' \; ; done
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/swappiness
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/vfs_cache_pressure
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/dirty_expire_centisecs
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/dirty_writeback_centisecs
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/dirty_background_ratio
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/dirty_ratio
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/min_free_kbytes
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/max_map_count
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/overcommit_memory
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/overcommit_ratio
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/kernel/sched_compat_yield
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/page-cluster
    -rw-r--r--    1 root     root             0 Jul  8 10:29 /proc/sys/vm/lowmem_reserve_ratio
    M.
  16. #836  
    Quote Originally Posted by rmausser View Post
    Update to the sysctl 'game changer' settings.

    I'm probably going to make this into an .ipk tomorrow.

    Code:
    vm.swappiness = 90
    vm.vfs_cache_pressure = 40
    vm.dirty_expire_centisecs = 1200
    vm.dirty_writeback_centisecs = 500
    vm.dirty_background_ratio = 5
    vm.dirty_ratio = 90
    vm.min_free_kbytes = 1024
    vm.max_map_count = 131072
    vm.overcommit_memory = 1
    vm.overcommit_ratio = 70
    vm.zone_reclaim_mode = 2
    kernel.shmmax = 268435456
    kernel.shmall = 16777216
    sched_compat_yield = 1
    vm.page-cluster = 4
    vm.lowmem_reserve_ratio = 1024




    -- Sent from my Palm Pre using Forums


    do you really see an improvement bringing back the page cluster and setting it at 4?
  17. #837  
    All i want to say is these setting so far have been amazing. i have been running well over a day now. running carbon all day which i couldnt do before. eventually after almost a whole day it would make my phone useless. today while using twitter all day, texting a lot, in and out of the browser, preware, checkin govnah i am amazed. only time i really even had a significant lag was when i got to a page with flash. accidentally hit two flash photo slide shows at the same time. locked up for a few seconds. but then both rolls starting playing perfectly. could move around the page and everything. very very slight hesitation when i closed it and then i was back to completely lag free carbon use. my memory was down too 211/97 right after i closed it. i cant even believe how great it has been.
    thanks everyone for all your work to optimize the memory. Great Work.
  18. #838  
    I am curious, I saw a few posts back you upped the screen buffers in luna.conf....what does this do exactly? Does it smooth the UI in that it provides more RAM for UI buffers?
    I have a pixi and its default = 3 ... should I use the 8 you posted?

    ( I am doing all the "game changer" tweaks, w/o 500MB swap but have 65MB compcache and the default 200MB store-swap enabled) seems to be working pretty good.
  19. #839  
    Quote Originally Posted by GenTarkin View Post
    I am curious, I saw a few posts back you upped the screen buffers in luna.conf....what does this do exactly? Does it smooth the UI in that it provides more RAM for UI buffers?
    I have a pixi and its default = 3 ... should I use the 8 you posted?

    ( I am doing all the "game changer" tweaks, w/o 500MB swap but have 65MB compcache and the default 200MB store-swap enabled) seems to be working pretty good.
    Hmm...I made a patch for it, but only for webos 2.1, I am not sure if the luna.conf settings are the same on 1.4.5.

    Could you do me a favski? PM me the luna.con file for 1.4.5 Pixi.

    If they are the same, a patching I will go.

    -- Sent from my Palm Pre using Forums
  20. #840  
    Quote Originally Posted by rmausser View Post
    Hmm...I made a patch for it, but only for webos 2.1, I am not sure if the luna.conf settings are the same on 1.4.5.

    Could you do me a favski? PM me the luna.con file for 1.4.5 Pixi.

    If they are the same, a patching I will go.

    -- Sent from my Palm Pre using Forums
    I am gonna create my own patch of that file =) Well at least for the buffer count. I did the 8px patch from preware to make it more ios smooth for dragging but it doesnt seem to work...LOL the value remains unchanged in luna.conf.
    Ill send you the stock luna.conf so I can test both just buffer change then all the other tweaks too =)

    UPDATE, heck I didnt see an attachment option in the PM creation...so Ill just attach it here...

    In my own experimentation I am changing the buffers to 8 and also, 8px drag radius and messing with the downloads section, put it to 128k I wonder if this is download speed cap.....? hrm....I also set it to max concurrent 4.


    Rename it to .conf =)

    UPDATE: =/
    Well, whatever I did my phone hated it. It wont boot up now, just sits on the first palm logo (the one before it starts fading in and out).

    Time to do some rescue work on its ***....annoying. I wonder which setting freaked it out.

    UPDATE:

    So, I got it back up and running w/ default luna.conf...

    I wonder which setting killed it...would the buffers have done it? Other than changing the buffers to 8, I changed the drag radius to 64, download concurrent to 4 and download max speed 131072....

    Help?
    Attached Files Attached Files
    Last edited by GenTarkin; 07/08/2011 at 04:52 PM.

Posting Permissions