[solved] Many discontinuities when simultaneously recording from 3 tuners (1 Viewer)

[mm1352000]

Thanks again for the suggestions. I ran the DPC latency analysis too l; the absolute maximum latency is around 300us (and usually between 70-130us), so according to this tool 'This machine should be able to handle real-time streaming of audio and/or video data without drop-outs.'.

I should have mentioned it is important to test when trying to use the tuners. Did you do that already?

Any particular settings that I could look at?

No, nothing particular.
I wonder...
On XP in the driver properties I have this interesting dialog which I never noticed before today:

Do you have that?
(I went to device manager, expanded "universal serial bus controllers", selected one of the controllers, right clicked, selected properties, then went to the advanced tab.)

With regards to 'some device to convert the USB 2.0 signal into a USB 3.0 signal' I actually meant something like adding an additional, external USB controller on one of the USB 3.0 ports. Maybe the following picture explains it better:

Ahhh, okay. I had thought that the bandwidth between the USB device and controller was okay, and that the issue between the USB controller and the PCH/chipset was the problem. So, I don't know if that will help at all...

 

[rsenden]

What generation CPU/chipset mac mini are you using? Also a bit curious about seeing the terms MediaPortal and Mac-Mini in the same sentence. What OS are you running?

I'm using a Mac Mini 6,1 (Late 2012) with Core i5 2.5Ghz; see http://www.everymac.com/systems/apple/mac_mini/specs/mac-mini-core-i5-2.5-late-2012-specs.html for more info. The Mac Mini is running Windows 7.

I see the same issues on my HP 8460p laptop though (coincidentally also with Core i5 2.5 Ghz), so it's not a Mac-specific problem.

I should have mentioned it is important to test when trying to use the tuners. Did you do that already?

Yes, I ran the DPC latency tool with all three tuners active (via Manual Control in TV Server Setup).

On XP in the driver properties I have this interesting dialog which I never noticed before today:
View attachment 157370

I looked through the USB device tabs before; I don't remember seeing anything like this for the USB 3.0 controller. On my HP laptop with Windows 7 I see this for the USB 2.0 controllers but not the USB 3.0 controllers.

Ahhh, okay. I had thought that the bandwidth between the USB device and controller was okay, and that the issue between the USB controller and the PCH/chipset was the problem. So, I don't know if that will help at all...

My USB 3.0 HDD performs OK (although I've never done a real speed test), so I am assuming that bandwidth between USB controller and PCH/chipset is OK. It looks like the bandwidth for all USB 2.0 devices combined is limited to 480Mbit/s, but I do not know where that limitation comes from.

 

[RonD]

you could do a quick disk copy test to see basic USB/disk BW. I did copy/paste of a large (6GB) recorded tv video. For win7 use TaskManager > ResourceMonitor and select disk tab. On my Ivybridge i3-cpu laptop I see the following for a WD-elements external USB3 disk drive

USB3-port, RD+WR = 37+37 MB/s = 74 MB/s = 592 mbps, probably 2.5 inch "laptop drive" limited
USB2-port, RD+WR = 14+14 MB/s = 28 MB/s = 224 mbps, mainly usb2 overhead slowdown?

No idea how to really compare disk vs tuner USB performance. What size USB "data packets" do disk/tuner use, driver overhead, USB command latency overheads, etc? But my USB2 port doesn't run 480 mbps even with just disk traffic.

To see USB port connections, you could use DeviceManager > View > Devices by Connection. look under ACPI x64 based PC

 

[rsenden]

Thanks everyone for the suggestions. Last weekend I did some more tests; to summarize these and earlier tests:

• Mac Mini, Windows 7, both tuner devices connected to USB 3.0 ports: Discontinuities when recording from all three tuners
• Mac Mini, Windows 8, both tuner devices connected to USB 3.0 ports: BSOD when trying to timeshift/record
• Mac Mini, Windows 8.1, both tuner devices connected to USB 3.0 ports: BSOD when trying to timeshift/record
• Mac Mini, OpenElec, both tuner devices connected to USB 3.0 ports: Discontinuities when recording from all three tuners
• HP Laptop, Windows 7, both tuner devices connected to USB 3.0 ports: Discontinuities when recording from all three tuners
• HP Laptop, Windows 7, both tuner devices connected to USB 2.0 ports: No discontinuities
  
• Mac Mini, Windows 7, both tuner devices connected to USB 3.0 ports, but without the Intel xHCI drivers installed (which effectively attaches the ports to a USB 2.0 controller): No discontinuities
• I also noticed interference between my USB 3.0 HDD cable and the tuner devices, even when only one tuner is used (discontinuities and strongly varying tuner signal strength when USB 3.0 cable/HDD is near tuner devices). The discontinuity problem when recording from all 3 tuners is however also present when the USB 3.0 HDD is disconnected.

So the issues are definitely related to using the tuners with a USB 3.0 controller, independent of the OS used (so I guess I can rule out driver issues). As a work-around, I now have the following setup:

• Mac Mini, Windows 7, no xHCI drivers installed, both tuner devices (and wireless keyboard/mouse receiver) connected to the USB 3.0 ports, routed to USB 2.0 controller.
• USB 3.0 HDD moved to NAS (connected to Mac Mini via Gbit LAN).
• Recording folder for TV Server moved to the USB HDD on the NAS.

Interestingly, on the Mac Mini I now get much better performance from the USB HDD than when it was connected directly to the Mac Mini (transfer speed of around 100MB/s instead of 30-40MB/s). With the timeshift folder pointing to the NAS, I see no discontinuities in TV Server Setup/Manual Control when simultaneuosly timeshifting on all three tuners.

 

[RonD]

Glad you are up/running, was curious about your results. I don't have USB tuners, just HDHR ATSC ethernet connected tuners. Used a test system, Dell Inspiron "everyday" laptop, Win8.1, Ivybridge i3-3227U 2-core 4-thread, 1.9 GHz. For testing 5 HDHR tuners 100 mbs Ethernet, with 3 MePo Clients, local, 2 remote. Used Win8.1 Task Manager to monitor disk traffic. Disabled Windows Defender realtime virus scan on all systems.

Ran same 5 channels for testing. US ATSC video mbps varies for different channels
KPIX=17 mbps, KBCW=17mbps, KTVU=12.9mbps, timeshift 3 clients
KNTV=13.3mbps, KGO=7mbps, record 2 programs to generate more disk traffic.

With 2 remote clients, ethernet shows RX=95mbps (ATSC tuners), TX=30mbps (2 remote clients). Sort of surprised 100 mbps enet supports 5 hdhrs tuners 95% busy.

********************
With laptop internal sata HDD, disk traffic
WR=10 MB/s, RD=6 MB/s, 3 Clients timeshift -5 minutes behind, RD=0 if 3 clients no timeshift. Without active timeshift, looks like windows uses memory buffered data for MePo Client reads.

No MePo Client video glitches if drive C: is less than 25% busy with MePo Traffic. MePo clients show lots of video glitches when “windows stuff” runs in background and pushes drive C: = 100% busy (OneDrive, livecomm, wwahost). No client video glitches if average drive activity less than 25%. Tried to disable background stuff, but from time to time windows
generates disk traffic. I usually setup MePo with 2nd drive for timeshift/recording.

********************
USB3.0 Port: External USB3.0 WD-Passport-Ultra portable drive 1TB

WR=9 MB/s, RD=6 MB/s, 3 Clients timeshift -5 minutes behind, RD=0 if 3 clients no timeshift
with timeshift, drive utilization = 15-20%

No real MePo client video glitches.

********************
USB2.0 Port: same External USB3.0 WD-Passport-Ultra portable drive 1TB

WR=9 MB/s, RD=6 MB/s, 3 Clients timeshift -5 minutes behind, RD=0 if 3 clients no timeshift
with timeshift, drive utilization = 60-70%, peaks 80+%

Lots of MePo Client video glitches, seems to happen when drive gets busy.

I never saw any crashes for my tests. Maybe USB Tuners interacting with USB external HDD are part of that story. My Windows USB drivers show the following.

Intel(R) USB 3.0 eXtensible Host Controller - 0100 (Microsoft)
Driver = 6.3.9600.17292, 9/2/2014