Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[cochinada]

[Edit by JimH -- Solved with a hardware change:  https://yabb.jriver.com/interact/index.php/topic,140752.msg977108.html#msg977108 ]

I have MC 30 running on Debian GNU/Linux 11 (bullseye) on my Quiet HTPC (Pentium Gold G5500T, 8GB RAM DDR4).
I have another PC with Windows 11.
All my movies are stored as MKV files in my QNAP NAS TS-873A in shared folders that are also mapped and mounted to the HTPC and Windows PC.
All these 3 computers are connected to a 10Gbps switch and I have of course also 10Gbps interfaces on each one that were successfully tested for speed and reliability.
My HTPC and Windows PC are connected via HDMI to my TV which in turn is connected also through HDMI to my Processor Anthem AVM 50v 3D for sound decoding and processing.

The issue I've noticed is that for movies with a large throughput like 60Mb/s or bigger (e.g. 4K with HDR), the MC player kind of chokes as there is a noticeable desynchronization between sound and picture as it advances in what seems like slow motion speed.
Like I said, this doesn't happen for movies that have a not so heavy throughput like e.g. 1080p, other "smaller" 4K, etc.

Before you blame my NAS, I made this simple test:
On my Windows 11 PC, I've played the same problematic movies using local installed VLC Player directly using the NAS as a source, without any issues. They all run smoothly!

So, there must be some underlying bottleneck in MC 30, perhaps with the player it uses or something similar?
What is exactly the cause and how can I fix this?

Than in advance.

[mattkhan]

Post is unclear as to which machine had the problem, it is posted to Linux 30 board but Linux 20 and windows 20 are mentioned and exactly how the drives are mapped & whether this is a client being served from the server hence potentially transcoded is also unclear

For a mixed os setup, I strongly recommend an update to https://yabb.jriver.com/interact/index.php/topic,139505.msg974383.html#msg974383 so you can use local file mode everywhere

[cochinada]

I'm sorry if you find it unclear as I tried to give and be as much detailed as possible.

Let me make it simpler:
The problem of course is with my HTPC where I'm running JRiver MC 30 under Linux.

Example of how the shared folders from my NAS are mounted on the HTPC:

joaquim@joaquim-HTPC:~$ cat /etc/fstab
...
//192.168.1.74/Filmes on /mnt/Movies type autofs (rw,relatime,fd=54,pgrp=1,timeout=0,minproto=5,maxproto=5,direct,pipe_ino=12938)

joaquim@joaquim-HTPC:/mnt/Movies/Ficção científica/Venom - Let There Be Carnage (2021)$ ls
'Venom - Let There Be Carnage (2021).jpg'  'Venom - Let There Be Carnage (2021).mkv'  'Venom - Let There Be Carnage (2021)_mkv_JRSidecar.xml'
joaquim@joaquim-HTPC:/mnt/Movies/Ficção científica/Venom - Let There Be Carnage (2021)$



I don't see this you mentioned to be applicable to my use case:
"4) windows/linux interop (e.g. https://yabb.jriver.com/interact/index.php?topic=114800.0 & https://yabb.jriver.com/interact/index.php/topic,135226.msg936774.html#msg936774)
* why? to allow all clients to playback a local file directly without transcoding by the server (which is cpu and network intensive activity in a video setup) in a mixed OS setup"

The "server" here is my QNAP running Quts Hero OS and I don't have any other software interfering with how the "Clients" chose to play the content. In this case the client being JRiver which simply uses the mapped and mounted points to play the movies.
I've given also the example of how my other "client" the Windows PC can play any movie using VLC player over the mapped network drives from the NAS.

EDIT: I've wrote MC 20 by mistake. I actually have MC 30. Sorry.

[mattkhan]

Ok so it's just playing the file itself? Have you monitored the host performance (Cpu, GPU, network) during this time? What do the jrvr stats indicate?

[cochinada]

Ok so it's just playing the file itself? Have you monitored the host performance (Cpu, GPU, network) during this time? What do the jrvr stats indicate?

Yes, that is it.
I didn't monitor as I played the movie in Theater view and it fills the whole screen. How can I do that instead?

[mattkhan]

I guess it will be much the same in windowed mode so try that?

I would try copying the file locally as first step and see if it is ok there, if yes then you know network or the share is the problem

[Awesome Donkey]

If it still happens locally, I'd say it could be the Pentium Gold G5500T being too low in performance to handle "larger" video files.

[cochinada]

I have bad news. I did a different thing easier for me: I've installed VLC for Linux and tried to play the movie remotely and I experienced the same sluggishness.
I wasn't able to collect any network performance stats (tried 'sudo iftop' and 'sudo nload' but all came empty) but I did collect CPU ones:

top - 12:08:42 up  3:05,  2 users,  load average: 4.28, 2.41, 1.48
Tasks: 213 total,   1 running, 212 sleeping,   0 stopped,   0 zombie
%Cpu(s): 95.7 us,  1.2 sy,  0.0 ni,  2.9 id,  0.0 wa,  0.0 hi,  0.2 si,  0.0 st
MiB Mem :   7606.6 total,   3113.1 free,   1553.5 used,   2940.0 buff/cache
MiB Swap:    976.0 total,    976.0 free,      0.0 used.   5105.0 avail Mem

    PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
  39218 joaquim   20   0 3293584 852332 136396 S 378.5  10.9   9:58.52 totem
    804 joaquim    9 -11 1353056  28564  22036 S   3.3   0.4   0:06.02 pulseaudio
  39200 joaquim   20   0  521584  26872  21544 S   2.3   0.3   0:03.72 gvfsd-smb
    957 joaquim   20   0 4965228 275104 158788 S   1.3   3.5   5:49.08 gnome-shell
  39096 joaquim   20   0 1919100 137968 101752 S   1.3   1.8   0:06.09 vlc
     86 root       0 -20       0      0      0 D   0.3   0.0   0:41.01 kworker/u9:0+i915_flip
  38317 root      20   0       0      0      0 I   0.3   0.0   0:00.98 kworker/0:1-events
  39274 joaquim   20   0   10300   4068   3256 R   0.3   0.1   0:00.01 top
      1 root      20   0  164316  10592   7804 S   0.0   0.1   0:02.09 systemd
      2 root      20   0       0      0      0 S   0.0   0.0   0:00.01 kthreadd
      3 root       0 -20       0      0      0 I   0.0   0.0   0:00.00 rcu_gp
      4 root       0 -20       0      0      0 I   0.0   0.0   0:00.00 rcu_par_gp
      6 root       0 -20       0      0      0 I   0.0   0.0   0:00.00 kworker/0:0H-events_highpri
      8 root       0 -20       0      0      0 I   0.0   0.0   0:00.00 mm_percpu_wq
      9 root      20   0       0      0      0 S   0.0   0.0   0:00.00 rcu_tasks_rude_
     10 root      20   0       0      0      0 S   0.0   0.0   0:00.00 rcu_tasks_trace
     11 root      20   0       0      0      0 S   0.0   0.0   0:00.59 ksoftirqd/0
     12 root      20   0       0      0      0 I   0.0   0.0   0:18.51 rcu_sched
     13 root      rt   0       0      0      0 S   0.0   0.0   0:00.04 migration/0
     15 root      20   0       0      0      0 S   0.0   0.0   0:00.00 cpuhp/0

Any idea of what could be wrong?
As I said, I've tested copying files between my NAS and this HTPC and the 10Gbps is proven correct so I don't think this is the issue here...

EDIT: I just saw your hint about my CPU. I guess I'll have to copy the file from my NAS and try using VLC locally this time to eradicate all doubts about it. Stay tuned!

[Awesome Donkey]

My money is on the G5500T. It was released in 2018 and cost $75 at the time, but you get what you pay for performance wise, which to say isn't much unfortunately due to having only 2 cores and 4 threads.

Also try running the benchmark in MC by going to the Help menu > Benchmark and post the results here.

[cochinada]

My money is on the G5500T. It was released in 2018 and cost $75 at the time, but you get what you pay for performance wise, which to say isn't much unfortunately due to having only 2 cores and 4 threads.

Also try running the benchmark in MC by going to the Help menu > Benchmark and post the results here.

Damn! I think you have won the bet. The reason I've chosen the CPU was because of the very low TDP (35W) as this is a Quiet PC without any fans and a low PSU supply of 160W and not so much for economical reasons. Now it seems I must get an alternative one 

I'll try to post MC benchmarks in a while as requested.

[cochinada]

I can't paste the Benchmarks outside of MC as the Paste button is always greyed but I can simply resume the scores I got:

Math: 1541
Image: 3443
Database: 3540
JRMark (version 30.0.96 64 bit): 2841

How bad is this and what is the minimum CPU recommended if there is such thing?
Is it advantageous to have a multithread CPU or does MC only use single thread?

[cochinada]

Well, I have no idea what the best/minimum scores should be but I've ordered a Ryzen 5 8600G so let's see.

[cochinada]

I can't paste the Benchmarks outside of MC as the Paste button is always greyed but I can simply resume the scores I got:

Math: 1541
Image: 3443
Database: 3540
JRMark (version 30.0.96 64 bit): 2841

How bad is this and what is the minimum CPU recommended if there is such thing?
Is it advantageous to have a multithread CPU or does MC only use single thread?

So, the new scores after replacing the HW and upgrading to MC33 are:

Math: 4743
Image: 6900
Database: 5729
JRMark (version 33.0.71 64 bit): 5791

Are these OK now?

[Awesome Donkey]

It's better, yes. The question is, does the sync/choke issues with movies still occur?

[cochinada]

It's better, yes. The question is, does the sync/choke issues with movies still occur?

Gr8 question! That is what I was about to try but unfortunately I can't. 
I've posted on the thread I have with MC33 now:
https://yabb.jriver.com/interact/index.php/topic,140857.msg976996.html#msg976996

[cochinada]

After fixing the issue I was having I can say that the HW upgrade indeed solved the initial reported issue so at the end it was the proper solution and I just hope it will stand the test of time as replacing a motherboard + CPU + RAM and upgrading is no small matter.

For instance, I was hoping I could re-use my OS disk untouched from my previous PC and avoid reinstalling anything but no. The new system didn't recognize it as bootable although it had Debian 11 installed. I guess the issue was because my
previous CPU was an Intel processor and I've replaced it with an AMD...

Another issue I discovered the hard way was that my other M.2 disk was also not recognized on the BIOS. It turned out that this is an M.2 SATA instead of an M.2 NVMe which makes all the difference for those not very aware of these things. I learned this from ASRock support, and I quote:

"For M.2 SSDs, there are 2 types of interfaces.

One is PCIe. This provides high speeds. Most PCIe M.2 SSDs are NVMe SSDs.
The other is SATA. This provides speeds up to 600 MB/s, just like regular 2.5” SATA SSDs. This variant is not so popular, because of the limited speed.
Samsung 860 EVO is a SATA SSD. It is not a PCIe or NVMe SSD.

B850M-X supports only PCIe M.2 SSDs.

SATA M.2 support is quite rare on AM5. Out of 209 currently offered different AM5 boards across all brands, only 12 support SATA M.2 SSDs. 9 from ASRock, and 3 from ASUS."

So I also had to buy an adaptor. As I opted to buy a new chassis as well, this turned out to be a costly upgrade, moreover because I had also to buy a more powerful PSU (and paying custom taxes on top of it) to cope with the 65W of the new CPU as opposed to the more meager 35W of the Pentium.

[zybex]

It may have been better to buy a new M.2 PCIe SSD instead of an adapter, and it would also be much faster

My HTPC is now an AMD 5600G on a Sharkoon QB One case. It's a 65W CPU, but you can limit it to 35 or 45W in the BIOS - I have it at 45W and the performance difference is negligible, and because the temperature is lower the fan is pretty much inaudible. The power consumption is very very low anyway on idle and while playing video - unless you're running games, it's not a concern.

Below is consumption at Idle: less than 3w at idle and 13w while playing a 4K HDR movie (100% EDC is normal at load for Ryzen Master).
Color is off because I captured the screenshots via VNC.

[cochinada]

It may have been cheaper to buy a new M.2 PCIe SSD instead of an adapter, and it would also be much faster

my HTPC is now an AMD 5600G. It's a 65W CPU, but you can limit it to 35 or 45W in the BIOS - I have it at 45W and the performance difference is negligible, and because the temperature is lower the fan is pretty much inaudible. The power consumption is very very low anyway on idle and while playing video - unless you're running games, it's not a concern.

Below is consumption at Idle: less than 3w at idle and 13w while playing a 4K HDR movie (100% EDC is normal at load for Ryzen Master).

I know but I just didn't want to throw out my M.2 that cost me almost 290€ and the adapter just cost me 15€. As for speed it is not an issue for my usage.

That power limitation is an awesome suggestion! I'll definitely look into it but in my case I don't have any fans so my HTPC is dead silent. Even the disks are all SDD. But lowering the temperature and consumption is always a good idea, specially if it doesn't come with noticeable performance drop.

BTW, are you using a dedicated graphics card or your iGPU? If so do you use Balanced mode or Quality mode? That is one doubt I have relative to what setting I should use... 


This is my chassis (without optical drive):
https://streacom.com/products/fc5-fanless-atx-case/

[zybex]

Before I also had a Zotac Zbox and a Gigabyte Brix - nano PCs without a fan. I got tired of some speed and expandability limitations so I upgraded to a more powerful PC with a fan, after testing that it would be inaudible anyway.

I use just the iGPU, no discrete card. It's more than enough for x265 HDR 4K playing at very high bitrates, and I don't sweat over some minor perceived quality bump that you may get with a faster card - at our age we don't see it anyway :-P

I still play with MPC-HC + MadVR, not MC+JRVR (sorry Hendrik) - I'm just more used to the MPC player, and use MC as a catalog/management configured to launch the external MPC player. I believe my MadVR settings are similar to the Balanced mode - on my previous Brix machine there were a couple of settings in Quality mode that would be too slow for usage, but on this Ryzen I can get away with Quality mode with no problem too. I need to re-test that eventually.

[cochinada]

Man!, you are giving me even more great ideas! I had to ask ChatGPT once again what were all those things you mentioned and I think I got it!

🔍 What software/tools are involved:
MPC-HC + MadVR →

MPC-HC = Media Player Classic - Home Cinema, a lightweight video player.
MadVR = a high-quality video renderer that improves image quality dramatically (used for HDR tone mapping, upscaling, etc).
MC + JRVR →

MC = JRiver Media Center (full media management software).
JRVR = JRiver Video Renderer, JRiver’s own in-house video rendering engine (alternative to MadVR).

🔎 What they’re doing:
They are not using JRiver’s built-in video renderer (JRVR) to play movies.
Instead, they prefer to play videos using MPC-HC with MadVR, likely because:
They’re familiar with it.
It may offer better image control or they’ve tuned it just the way they like.
However, they still use JRiver Media Center for organizing and managing the media library (like cover art, playlists, metadata, categories, etc).
JRiver is configured to launch MPC-HC as an external player when playing videos.


I think I will try to do exactly the same as I noticed now and then some instability when I launch a video and hopefully it will be better with the player you are using!

But in this case when you say "on this Ryzen I can get away will Quality mode with no problem too", are you referring to the settings on the MC or on the Media Player Classic?

EDIT: Damn! I just concluded that you are using Windows and not some Linux so I guess I can't use the same setup as you.

I seem to have these options though:

🔁 Summary:
Option                   Works on Linux?   HDR Tone Mapping   Notes
MPC-HC + MadVR   ❌ No                     ✅ Excellent                Windows only
JRiver + JRVR         ✅ Yes                    ✅ Decent                  Native Linux option
MPV                       ✅ Yes                    ✅ Very Good             Lightweight, advanced
VLC                        ✅ Yes                   ⚠️ Limited                   Simpler, but improving
 

[zybex]

Indeed, Windows
I recommend you stick with MC and JRVR, as MadVR is no longer under development and JRVR has already reached feature parity anyway.

Quality mode => I was referring to the MadVR settings equivalent to "Red October MadVR" + "Best Quality" mode in MC.

[cochinada]

Indeed, Windows
I recommend you stick with MC and JRVR, as MadVR is no longer under development and JRVR has already reached feature parity anyway.

Quality mode => I was referring to the MadVR settings equivalent to "Red October MadVR" + "Best Quality" mode in MC.

OK. I better not change too much. I will try "Best Quality" but what about HW acceleration, should I check it as well?

[zybex]

Yes, definitely.
Boa sorte com o PC novo

[cochinada]

Ah!!! I forgot you are a fellow compatriot of mine.
Muito obrigado! Vou precisar 

Back to English so everyone can follow. Can you be so kind to check if these settings are optimised for my specific HW (Ryzen 5 8600G)?



BTW, I came across again of that nasty window bug, even though I'm in X11 and not on Wayland anymore.
This time it displays so small that I couldn't resize it except with the right button and using the Resize menu:


I've been googling for more than an hour and I think there is no tool in Linux to measure power on a desktop computer. It only works if you have a battery, or in other words, on laptops, so I have no way of telling if my tuning will save power or not.

EDIT: I found one way and these were the values I got when playing this movie:

joaquim@HTPC:~$ cat ~/sensor_log_*.txt |grep PPT:
PPT:          29.23 W  (avg =   6.21 W)
PPT:          27.19 W  (avg =   6.19 W)
PPT:          24.08 W  (avg =   6.16 W)
PPT:          16.03 W  (avg =   6.17 W)
PPT:          30.08 W  (avg =   9.06 W)
PPT:          18.05 W  (avg =  10.16 W)
PPT:          16.15 W  (avg =  11.24 W)
PPT:          20.25 W  (avg =  12.12 W)
PPT:          36.20 W  (avg =  24.12 W)
PPT:          45.15 W  (avg =  44.16 W)
PPT:          21.07 W  (avg =  28.23 W)
PPT:          44.08 W  (avg =  29.14 W)
PPT:          40.05 W  (avg =  29.02 W)
PPT:          33.22 W  (avg =  28.20 W)
PPT:          35.16 W  (avg =  28.22 W)
PPT:          50.23 W  (avg =  28.03 W)
PPT:          41.07 W  (avg =  28.11 W)
PPT:          43.25 W  (avg =  29.09 W)
PPT:          19.07 W  (avg =  13.01 W)
joaquim@HTPC:~$

[zybex]

Consumption is a bit high... for Windows there are fixes/power profiles to fix that, not sure about Linux.
Or perhaps your TV resolution is not the same as the video being played, so JRVR is doing upscaling/downscaling. CTRL+J should tell you.

[cochinada]

My TV is also 4K.

I have these options on my BIOS but I'm no expert in overclocking or undervolting:

Performance Preset
Auto
PBO enabled
PBO and TJmax = 85ºC
PBO, TJmax = 85ºC and Curve Optimizer -20mV
PBO, TJmax = 85ºC and Curve Optimizer -30mV
PBO, TJmax = 85ºC and Curve Optimizer -40mV
PBO and TJmax = 75ºC
PBO, TJmax = 75ºC and Curve Optimizer -20mV
PBO, TJmax = 75ºC and Curve Optimizer -30mV
PBO, TJmax = 75ºC and Curve Optimizer -40mV
PBO and TJmax = 65ºC
PBO, TJmax = 65ºC and Curve Optimizer -20mV
PBO, TJmax = 65ºC and Curve Optimizer -30mV
PBO, TJmax = 65ºC and Curve Optimizer -40mV

I've tried these 3:
- STOCK
- PBO, TJmax = 75ºC and Curve Optimizer -20mV
- PBO, TJmax = 75ºC and Curve Optimizer -30mV

Results were...

STOCK
PPT:          50.23 W  (avg =  28.03 W)
edge:         +36.0°C

-20mV
PPT:          44.13 W  (avg =  43.20 W)
edge:         +37.0°C

-30mV
PPT:          43.05 W  (avg =  26.02 W)
edge:         +35.0°C

Note: on the -30mV I noticed some thin horizontal lines during the playing that didn't notice before.

Could it be the case that I should not limit the temperature and use instead e.g., "PBO, TJmax = 85ºC and Curve Optimizer -30mV"?

[zybex]

Personally I wouldn't undervolt/overclock, just leave all those to default including PBO. There's a specific setting to limit the max package power to 35/45 - on my motherboard it was quite hidden. What's your board?

Edit: For Asrock B850M-X there should be some "PPT Limit" settings under Advanced > SMU or Advanced > FCH.

[cochinada]

Personally I wouldn't undervolt/overclock, just leave all those to default including PBO. There's a specific setting to limit the max package power to 35/45 - on my motherboard it was quite hidden. What's your board?

Edit: For Asrock B850M-X there should be some "PPT Limit" settings under Advanced > SMU or Advanced > FCH.

OK. I found these:

Advanced -> AMD CBS -> SMU Common Options:
PPT Control

Here I can select Auto or Manual

When I select Manual, 3 options appear where I can enter any number as mW:
Fast PPT Limit
Slow PPT Limit
Slow PPT Time Constant

I have no clue what to put here...

I have also another interesting setting outside of this called "ECO Mode" with two options:
Auto
H Series OPNs SystemConfig 4 (35W)

What do you reckon?

BTW, those thin lines I saw are always present even with STOCK settings so has nothing to do with these tests... perhaps Quality settings and/or accelaration are too much for my HW or perhaps they are already on the movie (most likely IMO).

[zybex]

That Eco mode sounds good.
Isn't there also an option for "Sustained PPT limit" ?

Google:
"Under Advanced -> AMD CBS -> SMU Common Options, there's an ECO mode that says something like "H Series OPNs System Config 4 (35W)." You can also set 35W, 45W, and 65W under "System Configuration."

[cochinada]

I've tried the Eco mode "H Series OPNs System Config 4 (35W)" but I immediately saw the choke issue coming back as when I was using my old Pentium!

I've tested with...

Fast PPT Limit = 45000
Slow PPT Limit = 40000
Slow PPT Time Constant = 10000
PPT:          43.06 W  (avg =  28.05 W)
edge:         +38.0°C

Then I tried an even more aggressive approach and to my horror and dismay the results were terrible and much worse!!!
Fast PPT Limit = 40000
Slow PPT Limit = 35000
Slow PPT Time Constant = 8000
PPT:          59.11 W  (avg =  37.19 W)
edge:         +41.0°C

"Isn't there also an option for "Sustained PPT limit" ?"

I found it or I think I did under another obscure menu:
AMD CBS -> SMU Common Options -> SmartShift Control
Sustained PowerLimit

I've captured this as well:


"dropped 2 frames" sounds incredibly bad! Does that mean that my CPU is too slow with these settings?