Jump to content
XPEnology Community

modboxx

Member
  • Posts

    32
  • Joined

  • Last visited

Posts posted by modboxx

  1. On 8/9/2019 at 8:42 AM, real3x said:

     

    I tried too, i removed the i915 folder in /usr/lib/firmware and edit/removed the i915 EXTRA_FIRMWARE in the /etc/rc.modules

     

    All is working, plex HW transcode work as expected. And i can confirm the faster boot, i win like 100 ms, i see the nas login line at [  222.552049], previously it was at [  314.898573], so far so good :)

    This thread got me thinking so I went back and removed the other firmware and modules (See RC.modules) I have a J3455m so don't need most of those modules.  The boot time appears to be a whole lot faster without that stuff.  I also replaced the r8168 module with a newer one from the 24922 update.


    The only ones i haven't touched is OPTIONAL_MODULES="agpgart intel-gtt intel-agp cfbfillrect cfbimgblt cfbcopyarea sysfillrect sysimgblt syscopyarea fb_sys_fops fb" I'm not sure about those have to find some time to look them up and test.  I also tried to replace some of the june path but it didn't work. :(

    extra.lzma

    rc.modules

  2. - Outcome of the update: SUCCESSFUL

    - DSM version prior to update: DSM 6.2.1-23824U6

    - Loader version and model: Jun v1.04b - DS918

    - Using custom extra.lzma: YES, 

    - Installation type: BAREMETAL - ASRock J43455m

    - Additional comments: real3x  extra on update

  3. 9 minutes ago, Polanskiman said:

    Where did you download that extra.lzma? Those .DS_Store come from a MAC OS platform. These files store custom attributes. Are you using MAC?

    Jun added 2 extra.lzma and I am not entirely sure why as only one is normally used. Those .DS_Store files were not there originally so that extra.zma is either not the original one or is coming from you as MAC will automatically add those hidden files.

     

    From this thread, It's made by real3x.  It has the i915.ko and the drm*.ko's removed.  I too have a j3455 so I wanted to play with it.

     

    extra.lzma

    • Like 1
  4. 6 hours ago, real3x said:

    I finally made the jump for the update from 6.2.1-23824 Update 6 to 6.2.2-24922 Update 2, i only use a custom extra.lzma without i915 drivers, and everything work nicely, onboard Lan, HW transcode, i even made some test with geekbench to find out if the cpu burst work and its a big YES ! :)

     

    With the turbo states OFF in bios > https://browser.geekbench.com/v4/cpu/14202331

     

    With the turbo states ON in bios > https://browser.geekbench.com/v4/cpu/14202575

     

    *I made 2 bench of both turbo state to be sure.

     

    For the custom extra you can find it with instruction in this topic > 

     

     

    The Extra file in the link above has an additional extra file in it , a duplicate.  Also question why do you need to keep the i915 firmware if the .ko is removed?

  5. I've been under attack for a couple of days now and what's impressive is that the amount of IP's they have at their disposal(643 currently)

     

    Anyway here's a blacklist I've put together from all the attacking IP's if you want to block them at your gateway.  Additionally, make sure to disable your admin account as that seems to be the only account they are targeting

    Blacklist.txt

     

    Here's how it will look in the auth log.

    2019-07-29T08:11:03-04:00 Hostname synocgid: pam_unix(webui:auth): authentication failure; logname= uid=0 euid=0 tty= ruser= rhost=139.59.84.30  user=admin
    2019-07-29T08:11:48-04:00 Hostname synocgid: pam_unix(webui:auth): authentication failure; logname= uid=0 euid=0 tty= ruser= rhost=52.221.135.26  user=admin
    2019-07-29T08:14:37-04:00 Hostname synocgid: pam_unix(webui:auth): authentication failure; logname= uid=0 euid=0 tty= ruser= rhost=128.199.80.77  user=admin
    2019-07-29T08:16:22-04:00 Hostname synocgid: pam_unix(webui:auth): authentication failure; logname= uid=0 euid=0 tty= ruser= rhost=206.189.119.148  user=admin

     

    PS. I guess if you Synology is not open to port 5000 your probably OK for now.

    • Like 1
  6. - Outcome of the update: FAILED 

    - DSM version prior to update: DSM 6.2.1-23824 Update 6

    - Loader version and model: Jun v1.04b - DS918

    - Using custom extra.lzma: NO

    - Installation type: BAREMETAL - Asrock J3455M (motherboard network card disable) with Intel I350-T4

     - Additional comments: No network at boot

     

    - Outcome of the update: FAILED 

    - DSM version prior to update: DSM 6.2.1-23824 Update 6

    - Loader version and model: Jun v1.04b - DS918

    - Using custom extra.lzma: NO

    - Installation type: BAREMETAL - Asrock J3455M (motherboard network card disable) with Intel NC360T 

     - Additional comments: No network at boot

  7. 15 minutes ago, Jeong said:

    There seems to be a problem with the bios version.
    Upon searching, it appears that the Bios 1.5 and 1.6 versions of J3455-ITX and J4205-ITX have brick issues and some bugs have not been fixed.
    The bios versions of most J3455 and J4205 that are not working properly appear to be 1.4.

    This is just my think.

    I actually think the same, I read about that a while back and most people reported gpu issues after upgrading their bios.  

     

    Try to upgrading again and see if it lets you?  then clear cmos and see what happens maybe you'll get lucky.

  8. 12 minutes ago, Jeong said:

     

    I regret my poor English.

     

    What is your motherboard for?
    Many people using Asrock's J3455 and J4205 are currently experiencing a situation where the /dev/dri folder does not exist and would like to know why you only operate normally.
    1. Is it the main board of Asrock?
    2. What's the bios version of Asrock's main board?
    3. Is there any change to the bios setting you have noticed?
    4. How many HDDs do you use?

    5. Do you think there is a change depending on the type of USB memory?

     

    As you know, I would appreciate it if you could answer these questions.

    Thank you very much for your help.

     

    My Motherboard J3455M

     

    Bios 1.80

     

    I did change my bios settings but I do not believe that has anything to do with it because I've tried almost every combination of settings and never had an issue with HW acceleration.

     

    I used to have one server with six drives (5x 4tb purple + 1 ssd Samsung pro) then I split it into two server, the one from the video only has one drive either way when it was six or one hw acceleration still worked.

     

    I use an old 4gb usb2 stick

     

  9.  

    17 hours ago, ideasman69 said:

    understand that - but hardware acceleration is enabled by default in video station on 6.2.1. toggling the setting does nothing, streaming video works but as soon as you ask it to transcode - nothing happens. i was under the assumption that any hardware acceleration required video station to be installed, hardware accelaration enabled and the codecs activated. Once this is done, the /dev/dri/ device appears to which i can passthrough the plex docker. there is no /dev/dri device on my 1.04b / 6.2.1 build so plex transcoding will not work:

     

    nodri.png.d73916dc9f75fe4979995d35d1e8d18f.png

     

    it's worth noting that i haven't tried your recommended bios settings with 1.03a2 so i might wipe it all and give that a go. i at least had hardware transcoding activated on 1.03a2.

     

    That's odd that the folder is not there?  I can confirm on a clean install that folders is there

    Here's a screenshot from a clean install from this morning with just ssh enabled.  

    PS disk was wiped prior to install.

    tEHdQNH.png

     

    In your bios settings is your graphics set to onboard?

     

    I posted a test video of plex hw acceleration performance with juns new i915 driver, also note this is with performance set eco mode in bios, with sports its even faster.  During playback watch the buffer fill up, it's amazing how good that is with a 4k hevc.  Additionally that video is from a remote storage folder located on another xpenology.  One last note the video was recorded using 10fps so it's a bit choppy but not the reflection of the real experience.

    Test video

  10. 7 hours ago, ideasman69 said:

     

    nope - even with all of @modboxx's advice over here, system still runs slow :(

     

    i've got 3 different models of these boards (J3455-M, J3455B-ITX and J3455-ITX) but they all run slow with either of the 918+ boot images. So im not sure how you are the only one on here with one of these boards and no issues 😕

     

    the activation.conf file doesn't exist with the latest boot loader and 6.2.1, whereas it did with the previous boot loader and 6.2. hardware transcoding in video station doesn't seem to work at all now? 

     

    i have plex pass but it's not even worth touching that until it's working in video station. it'd be awesome if @jun could mention what type of motherboard his testing this stuff on and what his bios suggestions are. might help a lot of us!

     

    Those settings were for 1.03a2  loader that i found to improve my system performance, for 1.04b I actually use the opposite of some of those settings and have a great experience.  I'm not the only one who uses the j3455 platform and has a great experience I recommend you look over these forums on advice for issue your are having.

     

    As for actual performance problems please clarify what exactly is your issue?  Where's the slow down? Is it network? Is it hard drive related?  GPU? 

     

    Activation.conf does not get created when you install DSM6.1/6.2.  So if this was a fresh install don't expect to see it.  The file gets created after you install Videostation, enable hw acceleration, and play a video, until then it will not exist unless you ssh into your system and run the command to create it.  Additionally Plex does not use the file whatsoever, it doesn't care if it's there or not and on top of that Videostation Sucks why even waste your time with it?  The newer version are limited to a few audio codecs even with ffmpeg installed.  

  11. Hi Arkhip, Question?  I'm using the sybdata-proxyalpine build. I tried to start the p2p broadcasting or hybrid broadcasting but nothing happens? Has anyone tried this?  For testing I tried it on the Windows version and it worked fine.  

     

    What do you guys use for IOS?  I've tried a few different apps and they were either bloatware or just didn't work.

     

     

  12. 3 minutes ago, refrain said:

    i dont know. so i can only play the movie by using software transcoding by CPU resources.

    if my HDD or CPU resource has I/O problem, the software transcode function can not be running correctly.

     

    so i am confused.

    No no I'm sorry that was just a copy and paste of the definitions of D and L forget those.  

     

    Explain what exactly happens when you play a movie in Plex with HW transcoding on? forget what you see in process just tell me what actually happens on the client or server?  Does it crash or just says no?

  13. 9 minutes ago, refrain said:

    DI RI process was correct by hw-transcode function.

    my process has marked been "D" means  (Uninterruptible sleep) , almost waitting for the resources like HDD,CPU and etc.

    so my processes was running incorrectly that why i can not do hardware transcoding action.

    "D" uninterruptible sleep (usually IO) + "l" is multi-threaded (using CLONE_THREAD, like NPTL pthreads do)

     

    What exactly is your issue?

     

  14. On 8/14/2018 at 9:46 PM, refrain said:

     

     

    thank you for reply~

     

    the hw-transcode function only can use once for me each reboot.

    the first time play movie by using hw-transcode will fine.

    the next time the movie has been forzen and can not play.

    then i use putty to ssh the DSM machine, then i can find the play process has marked "D" like below:

    
    root     20186  0.0  0.2  82864  9584 ?        D    22:22   0:00 /var/packages/VideoStation/target/bin/ffmpeg -ss 0.000 -vaapi_device /dev/dri/renderD128 -hwaccel vaapi -hwaccel_output_format vaapi -noautorotate -i /volume1/video/AAAA/Just Dance 2018 - Dj Got Us Fallin' In Love Again (Full Fanmade Fitted Gameplay).mp4 -vcodec h264_vaapi -vf format=nv12|vaapi,hwupload,setsar=sar=1,scale_vaapi=w=1280:h=720 -vsync 2 -bf 0 -vb 1000000 -acodec copy -f ssegment -segment_format mpegts -segment_list_type m3u8 -hls_seek_time 0 -segment_time 5 -segment_time_delta 0.000 -segment_start_number 00000 -avoid_negative_ts 0 -break_non_keyframes 1 -map 0:0 -map 0:1 /tmp/VideoStation/HLS/f1d75a150b0ddc68412cdf9d2f5a5c17_bL9Apy9s/slice-%05d.ts
    
    root     18421  0.0  0.2  82864  9356 ?        D    22:20   0:00 /var/packages/VideoStation/target/bin/ffmpeg -ss 0.000 -vaapi_device /dev/dri/renderD128 -hwaccel vaapi -hwaccel_output_format vaapi -noautorotate -i /volume1/video/AAAA/Just Dance 2018 - Dj Got Us Fallin' In Love Again (Full Fanmade Fitted Gameplay).mp4 -vcodec h264_vaapi -vf format=nv12|vaapi,hwupload,setsar=sar=1,scale_vaapi=w=848:h=464 -vsync 2 -bf 0 -vb 500000 -acodec copy -f ssegment -segment_format mpegts -segment_list_type m3u8 -hls_seek_time 0 -segment_time 5 -segment_time_delta 0.000 -segment_start_number 00000 -avoid_negative_ts 0 -break_non_keyframes 1 -map 0:0 -map 0:1 /tmp/VideoStation/HLS/f1d75a150b0ddc68412cdf9d2f5a5c17_z3b6NxWg/slice-%05d.ts

     

    each play time will do this action again then more and more process has been locked. (can not be killed)

     

    i have the plex pass and i have try these apps(Video Station/ EMBY / Plex) and all the same problem has found.

     

    i have change the bios setting like you except CSM setting because i made the loader-img burn to SSD disk. when i disable the CSM that the system can not be started normally.


     


     

    I just checked with Plex and it goes from Dl to Rl to Sl.  Ending the movie exit the process.

    plex     14761 22.4  0.4 498520 67548 ?        Dl 
    plex     14761 23.8  0.4 496448 67684 ?        Rl 
    plex     14761 25.0  0.4 500148 68400 ?        Sl

    Maybe I'm missing something?  but with Plex what i see is that it encodes up to the 60 seconds (Throttle buffer) during this it's in Rl state, once it completes that chunk? the process briefly enters into the Dl state until it's able to send the client those packets? afterwards it goes into the sl state  and exits. Then a new process repeats this.

  15. 22 hours ago, refrain said:

     

    Same to you.

    there will be many transcoding process has been marked "D" by using console(ps aux | grep vavpi) command to show.

    then the hw-transcoding function can not be useful before next roboot.

    the hw-transcoding function only be use once or twice usually.

     

    so confused.

    You guys should look over your bios settings, I have a j3455M with two extra ahci cards, two ports per a card(using 5 of the 6 ports) and I do not have of these issues. 

     

    Bios (1.80 spectre/meltdown)

    CPU config:  

    Intel SpeedStep = disabled - No driver (Slows and crashes system)

    CPU C States = disabled 

    intel vt-d = enabled

    PowerGear = Sport (Buttery smooth Plex)

     

    Chipset Configuration:

    Adjust your memory speed and voltage, don't leave it on auto.

    Share Memory = 64

    Onboard HD Audio = Disabled

     

    ACPI Configuration:

    Suspend to RAM = disabled

    RTC Alarm Power On = disabled (DSM can't read RTC Clock so doubt this would work, best to disable for now)

     

    USB Configuration:

    Legacy USB Support = UEFI only

     

    Trusted Computing:

    Disable all

     

    Security Screen:

    Secure Boot = Disabled

    Intel(R) Platform Trust Technology = Disabled

     

    Boot Screen:

    CSM (Compatibility Support Module) = Disabled

     

    Some additional notes. 

    Goto /etc.defaults/synoinfo.conf

    support_led_brightness_adjustment="no"  that should fix your ahci_led stuff

    support_fan="no"

    support_spectre_meltdown_mitigation="yes" (Check bottom) if your on bios 1.80(j3455m) or which ever version that fixes the spectre/meldown for the j3455*-itx model 

     

    Plex does hw transcoding every time I use it, never had to reboot it, do you guys have plex pass?

     

    • Like 1
  16. 4 hours ago, Sonnyboy666 said:

    Okay, here it is:

     

    [   13.448102] [drm] Initialized drm 1.1.0 20060810
    [   13.517978] [drm] Memory usable by graphics device = 4096M
    [   13.517979] [drm] VT-d active for gfx access
    [   13.526807] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
    [   13.526809] [drm] Driver supports precise vblank timestamp query.
    [   13.540217] [drm] Disabling framebuffer compression (FBC) to prevent screen flicker with VT-d enabled
    [   13.600790] [drm] GuC firmware load skipped
    [   13.604722] [drm] Initialized i915 1.6.0 20160919 for 0000:00:02.0 on minor 0
    [   13.607928] [drm] Cannot find any crtc or sizes - going 1024x768
    [   13.820752] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
    [   14.731309] [drm] RC6 on
     

     

    Thanks!  That's a bad ass CPU for Xpenology.

  17. 6 minutes ago, Olegin said:

     

    I know it, but Sonnyboy666 wrote, that hw transcoding work on his Intel Xeon E3-1225 v5 in VS:

    That's a nice cpu for Plex transcoing :) 

     

    Sonny can you also do below command and post result?

    dmesg |grep drm

     

  18. 36 minutes ago, Olegin said:

    Can you post here the result of this command in telnet/ssh:

    
    cat /usr/syno/etc/codec/activation.conf

     

    This file does not exist by default you need to install Videostation, enable hw transcoding and actually go play a movie before it appears.

     

     

×
×
  • Create New...