Mediaportal 1.14Pre: TV-Server always using a new TV-Card when re-use is possible (1 Viewer)

[mm1352000]

Switching off the PCIe L1/L0 States seems to make the situation better - I still have discontinuations, but much lesser.
Before it was nearly impossible to replay a film, now the problem appear not any time _and_ if it happens the amount of errors seems getting much lower (now: 20 drop outs per 90 minutes, before nearly any GOP).

Note that discontinuities are usually caused by one of:

1. Inadequate signal strength/quality.
2. HDD overload (eg. scanned by security software, highly fragmented and almost full).
3. Bad driver(s) (eg. NIC, USB) with high DPC latency.

If you want to eliminate the discontinuities, I suggest you check points 2 and 3.

What I can't understand: I monitored the signal strength and quality of all the cards for a while in "manual control" of TV-Server with the timeshift option.
I never head signals lower than 90% and qualities lower than 80 - 85%.

AFAIK this is normal for TBS tuners.

When I take a look into the logfiles I sometimes see qualities about 50% (most on HD channels).

Unless the reading is consistently low (eg. when you check in TV Server Configuration -> manual control, or MediaPortal tuning details), I would ignore the log.

If I timeshift the same channels or other HD channels on the transporter the window outputs at least 80-85% quality to me

Okay, so as above => ignore the 50%.

Is there a posibility to monitor the quality of the incoming signals during reception?

I don't understand what you mean by "during reception".

About the PCIe compatibility: As both cards draw a MSI option to the system, they should be able to handle PCIe v2(.2). If not, we have found the "compatibility" issue here.

The cards will always be PCIe v1.0a/1.1 compliant. If there are compatibility problems, the problem is with the motherboard's compatibility with PCIe v1.0a/1.1 cards (even though it might be okay for v2 and/or v3).

 

[rue]

Switching off the PCIe L1/L0 States seems to make the situation better - I still have discontinuations, but much lesser.
Before it was nearly impossible to replay a film, now the problem appear not any time _and_ if it happens the amount of errors seems getting much lower (now: 20 drop outs per 90 minutes, before nearly any GOP).

Note that discontinuities are usually caused by one of:

1. Inadequate signal strength/quality.
2. HDD overload (eg. scanned by security software, highly fragmented and almost full).
3. Bad driver(s) (eg. NIC, USB) with high DPC latency.

If you want to eliminate the discontinuities, I suggest you check points 2 and 3.

HDD and latency is no problem (checked it before, 2x SSD Samsung 850 PRO) for system and RAID5 array build on AVAGO 2960i8 PCIe x16 (different computer).
Filesystems NTFS / EXT4

Drivers on Windows 10 Pro@64 Bit freshly loaded from TBS side.
CPU-Load on both systems < 5%. HDDs only for system and video storage.

What I can't understand: I monitored the signal strength and quality of all the cards for a while in "manual control" of TV-Server with the timeshift option.
I never head signals lower than 90% and qualities lower than 80 - 85%.

AFAIK this is normal for TBS tuners.

OK - extra ignoring

When I take a look into the logfiles I sometimes see qualities about 50% (most on HD channels).

Unless the reading is consistently low (eg. when you check in TV Server Configuration -> manual control, or MediaPortal tuning details), I would ignore the log.

OK - extra ignoring

If I timeshift the same channels or other HD channels on the transporter the window outputs at least 80-85% quality to me

Okay, so as above => ignore the 50%.

OK - extra ignoring

Is there a posibility to monitor the quality of the incoming signals during reception?

I don't understand what you mean by "during reception".

Is there a chance to get "life"-data about signal strength and quality from the card - if there is a problem with the DVB-S switch or LNB this would be a chance finding it out.

About the PCIe compatibility: As both cards draw a MSI option to the system, they should be able to handle PCIe v2(.2). If not, we have found the "compatibility" issue here.

The cards will always be PCIe v1.0a/1.1 compliant. If there are compatibility problems, the problem is with the motherboard's compatibility with PCIe v1.0a/1.1 cards (even though it might be okay for v2 and/or v3).

If this is true - and I am absolutely sure you know what you are talking about - the system gets a PCIe 2.2 signalling but has to handle a PCIe 1.1. I assume this is the reason why the board needs the "compatibility switch". Will try what happens when I manually go down to PCIe 1.1 in the boards configuration this evening.

 

[rue]

Hi again - tested some things last night and now I am sure that PCIe board <> PCIe TBS cards

When I returned home the Windows 10 installation completelty stalled (no blue screen, only freezing). I rebooted first and waited for a while - was reproducible.

Than I turned on ASPM again with standard values -> system seems to be stable for a longer time than I saw in the log files.

After that I switched the PCIe sockets with the TBS cards to PCIe v1 -> cards would not be recognized anymore (repeatable)
After that I switched the sockets to PCIe v2 -> system immediatelly hangs at card PCIe init.

Proper startup and function is only possible when PCIe x1 config is turned to "auto" and compatibility mode ... to go deeper would mean to get an analyzer home and gravel into details -> makes no sense to me.

What I saw now in addition: If only one channel is in use it takes reproducible about 40 minutes to get the first discontinuations. Sounds again like incompatibility on PCIe - maybe clock not safely locked.

I double checked disk performance, too:
-> burst transfer rate to NAS for 10 Min: 110 - 120 MByte/s
-> minimum available transfer rate (any time): 50 MByte/s

I think this should be enough for video recording.

Today I ordered a Digital Devices Cine S2 V6.5 for a cross check - will see what happens.

Thanks for your help and best regards

RUE

 

[mm1352000]

maybe clock not safely locked.

Are you overclocking the CPU?

I double checked disk performance, too:
-> burst transfer rate to NAS for 10 Min: 110 - 120 MByte/s
-> minimum available transfer rate (any time): 50 MByte/s

I think this should be enough for video recording.

This bandwidth is more than sufficient. Approximately 16 M. bit/s (= 2 M. Byte/s) would be enough for one HD stream. However, the bandwidth is not the only factor to consider. You should also consider the latency - both absolute latency value, and latency variability. If the latency is consistently high it can be a problem; if there are latency "spikes" (short periods of time when the latency is very high) it can be a problem.

In other words, only looking at the bandwidth is like looking at half the picture. You need to look at the latency as well to see the full picture.

Earlier you said:

HDD and latency is no problem (checked it before, 2x SSD Samsung 850 PRO) for system and RAID5 array build on AVAGO 2960i8 PCIe x16 (different computer).
Filesystems NTFS / EXT4

Drivers on Windows 10 Pro@64 Bit freshly loaded from TBS side.
CPU-Load on both systems < 5%. HDDs only for system and video storage.

It seems like you didn't understand me.

I was talking about DPC latency, not storage access latency.
I was talking about all drivers, not just the tuner driver(s).

Any driver can cause DPC latency issues. Often it seems to be an NIC (wired or wireless) or USB controller driver that causes problems, not a tuner driver.

So, please use LatencyMon to check the DPC latency when the discontinuities are occurring.

 

[rue]

Are you overclocking the CPU?


This bandwidth is more than sufficient. Approximately 16 M. bit/s (= 2 M. Byte/s) would be enough for one HD stream. However, the bandwidth is not the only factor to consider. You should also consider the latency - both absolute latency value, and latency variability. If the latency is consistently high it can be a problem; if there are latency "spikes" (short periods of time when the latency is very high) it can be a problem.

In other words, only looking at the bandwidth is like looking at half the picture. You need to look at the latency as well to see the full picture.

Earlier you said:

It seems like you didn't understand me.

I was talking about DPC latency, not storage access latency.
I was talking about all drivers, not just the tuner driver(s).

Any driver can cause DPC latency issues. Often it seems to be an NIC (wired or wireless) or USB controller driver that causes problems, not a tuner driver.

So, please use LatencyMon to check the DPC latency when the discontinuities are occurring.

In this case I misunderstood you - this is a good idea - I will check it again.

Yes I changed the storage topology to find out if there is an influence if I store it locally or using the NAS.
Because this is very easy in MP (thanks, to) this can be done very fast.
And in the beginning it seems there is a differene - but doing the tests multiple I can deny that now.

Again thanks for the tip - I will try to check the latencies with latencyMon.

best regards

RUE

 

[rue]

Hi and sorry for coming back so late.

Menawhile I played much around with the card but did not found the real issue "in software".
Finaly it happend Win10 is changing it's system ID due to some changes of me - I lost my activation and was unable to redo it.

For me it was not clear till that date, that it is not possible to simply re-use the Win7 Key or the activation itself.
A call to MS made it clear - I had to do a clean install (Win 7 -> upgrade) or buy a new Win10 License.
So this took a couple of time. Finally I had a new system up and running - with the same problems again.

At least I changed the TBS card with the DD Cine S2 v6.5 and lost all trouble.

I plugged the card in, had no longer "compatibility issues" on PCIe (could switch off the setting) and only 3 discontinued paket for the last 3 days.
The feature of losing nearly all information after 40 minutes of card use is disappeared now.

I send this to TBS germany, too. It seem having only one card in the system will cause no problem (except the compatibility issue).
The second card seems to be necessary to bring the large amount of discontinued packets.

I will monitor this and if I receive any information I will post it.

Thanks for your help

RUE