HDHomeRun (HDHR) Prime + SageDCT

[jbuszkie]

I just discovered this thread and had to wipe the drool off my face a couple of times!

I really want to get the prime but I'm curious to know if I'm going to be able to use it for anything but the locals? I don't suppose there is a way to find out before I purchase it?

I'm thinking I'm ok since I have the firewire tuning working fine on two cable boxes. I checked on the missing remote website and mine isn't listed.

I'd sooo love to dump the cable boxes for cable cards!

Jim

[Kirby]

Originally Posted by jbuszkie

I just discovered this thread and had to wipe the drool off my face a couple of times!

I really want to get the prime but I'm curious to know if I'm going to be able to use it for anything but the locals? I don't suppose there is a way to find out before I purchase it?

I'm thinking I'm ok since I have the firewire tuning working fine on two cable boxes. I checked on the missing remote website and mine isn't listed.

I'd sooo love to dump the cable boxes for cable cards!

Jim

What kind of cable box? Moto? If so, you can go into the service menus of the box and read the CCI status. Not sure about how it correlates to firewire. Back when I did firewire recordings, it was the 5C flag that mattered, not sure if the CCI bit will correlate to 5C?

[jbuszkie]

Originally Posted by Kirby

What kind of cable box? Moto? If so, you can go into the service menus of the box and read the CCI status. Not sure about how it correlates to firewire. Back when I did firewire recordings, it was the 5C flag that mattered, not sure if the CCI bit will correlate to 5C?

Yah.. both are moto.. Where do I find the services menu?
I've had good luck so far with the 5C. I hope it's an indication of the "copy freely" ability.
I'd really want to get away from the cable box all together!

[Kirby]

Originally Posted by jbuszkie

Yah.. both are moto.. Where do I find the services menu?
I've had good luck so far with the 5C. I hope it's an indication of the "copy freely" ability.
I'd really want to get away from the cable box all together!

Going from memory, which is scary in itself, take your STB remote, pick your channel, power off from remote, and hit 'select' a few times quickly. or maybe its 'menu'.

To get out of it, you power off and on.

Whichever one is correct, its like the 5th or 6th item down you navigate into and will then be able to see CCI status I think. Maybe inband something or other? I can tell you for sure tonight.

*EDIT* Its the select button. (or ok)

[doncote0]

Originally Posted by Kirby

"Halt detected in recording". Probably about 200+ since last night.

I am seeing this too. So it could be caused by old NIC drivers? For a while I was considering getting rid of my Norton firewall and going with the Windows firewall. The video sources are listed as functioning: true, but that status will change to false and I am at a loss for how to get it back.

The steps I took earlier seem to be only temporary.

I have had to do manual "finds and replace" in sage.properties, too. I have generated assignments with overwrite and if I chose 6970, 6971 and 6972 for some reason the label of the video source will hold on to the previous source label.
So I can have a SageDCT source at 6970 with a label that makes it sound like it was @ 6976 or similar.

(Yes, I overwrote the mmc portions.)

Auto reset just hides the cause? Should I use it then?

By the way, the Comcast rep activated just one line of the MCard. Do all three need to be individually activated and recorded?

[Kirby]

I actually fixed my halt detected issue simply by resetting the physical channel numbers for all channels that were showing the old QAM mapping. Not having any issues now.

[phelme]

Originally Posted by babgvant

If you have a Realtek NIC and haven't updated the drivers recently, I'd try that. On my HTPC utilization is ~3-4% with 3 HD recordings going.

OK I'll check. Does the Prime do jumbo frames?

[doncote0]

Originally Posted by phelme

Does the Prime do jumbo frames?

Since it is a product designed for end users with various IT knowledge, I find it unlikely. I suppose the Prime could do jumbo frames dynamically, but I doubt it.

Originally Posted by Kirby

I actually fixed my halt detected issue simply by resetting the physical channel numbers for all channels that were showing the old QAM mapping. Not having any issues now.

Thanks, Kirby. I'll look for that.

[Kirby]

Originally Posted by phelme

If I may ask, what does SageDCT do that takes up about 13% of my CPU per tuned tuner? I have a quad-core AMD so it can deal, but I'm just curious.

Might it have something to do with using the motherboard's gigabit Ethernet as opposed to a dedicated card?

For comparison, I am using my motherboard's gigabit, running Intel Q6600 quad core, and with 3 recordings going and a placeshifter set to high quality, and a remote desktop connection into the server I was at 22% cpu usage. stop the placeback on placeshifter and it dropped down to 3-4% cpu. network bandwidth peaked at about 6%.

[phelme]

Originally Posted by doncote0

Since it is a product designed for end users with various IT knowledge, I find it unlikely. I suppose the Prime could do jumbo frames dynamically, but I doubt it.

I'll ask Jason at SD, but I imagine you're right.

Well, the new Realtek driver made no difference (8/11 ,v5.79). I'm at 50+% for SageDCT with three tuners going.

Edit: I turned down logging to "Error" thinking that might help but it didn't.

If I'm in the neighborhood of a nerd store today I'll pick up a new NIC and try that.

[doncote0]

Originally Posted by phelme

I'll ask Jason at SD, but I imagine you're right.

Well, the new Realtek driver made no difference (8/11 ,v5.79). I'm at 50+% for SageDCT with three tuners going.

Edit: I turned down logging to "Error" thinking that might help but it didn't.

If I'm in the neighborhood of a nerd store today I'll pick up a new NIC and try that.

Sounds like you might have a DPC Latency issue. Try turning EIST off in BIOS. Also, take a look at this post from some cool Sage geek.

http://forums.sagetv.com/forums/show...96&postcount=5

[phelme]

Another piece of info. The working set size of SageDCT got to be 1.6 GB with all tuners going after running for about 10 minutes. I eventually got this crash:

2011/09/30 10:29:37.47 : Error ProcessPackets: Exception of type 'System.OutOfMemoryException' was thrown.

There's definitely a memory leak somewhere. I'm running it again and the SageDCT working set steadily rises (up to 500 MB's in 4 minutes). Maybe I have a .NETFramework out of date or the new Realtek driver is the cause.

[phelme]

Updated a few .NetFrameworks. That has brought the CPU load down to the low 20's. An improvement. But the working set still does climb, though now not nearly as quickly.

I'm now having some failure in the tuners where I wasn't before as well (recording failed due to conflicts in the tuners errors too). I guess I'll roll back the Realtek drivers and see what happens.

Oh when I stop the recordings on the tuners, only a small amount of the physical memory is reclaimed.

Edit:
Fell back to the older Realtek and the CPU went back up to 50% but the memory leak still continued. Now back on the latest NIC drivers and CPU load back under 30%.

Am I the only one seeing the leak?

[zoop]

Originally Posted by phelme

Updated a few .NetFrameworks. That has brought the CPU load down to the low 20's. An improvement. But the working set still does climb, though now not nearly as quickly.

I'm now having some failure in the tuners where I wasn't before as well (recording failed due to conflicts in the tuners errors too). I guess I'll roll back the Realtek drivers and see what happens.

Oh when I stop the recordings on the tuners, only a small amount of the physical memory is reclaimed.

i just worked with andy (thx andy!) on this symptom for the past week... are you writing to a UNC path?

in my case memory stability was achieved by updating my network driver, offloading some tcp functions to the NIC, increasing the write cache by 10x, and a few other things.

thx

/jer

[phelme]

Originally Posted by zoop

i just worked with andy (thx andy!) on this symptom for the past week... are you writing to a UNC path?

in my case memory stability was achieved by updating my network driver, offloading some tcp functions to the NIC, increasing the write cache by 10x, and a few other things.

thx

/jer

Yep UNC paths. Have some other network encoders out there that need it.

I think I've already offloaded what I can to the NIC, but I'll double check. I'll also up the write buffer to 10240. Will report back. Thanks!

[jbuszkie]

Originally Posted by phelme

I think I've already offloaded what I can to the NIC,

Stupid question.. but what/how can you set up to off load to the NIC??

Jim

[phelme]

Originally Posted by zoop

i just worked with andy (thx andy!) on this symptom for the past week... are you writing to a UNC path?

in my case memory stability was achieved by updating my network driver, offloading some tcp functions to the NIC, increasing the write cache by 10x, and a few other things.

thx

/jer

Didn't help. Up to a 1.5 GB working set in about 10 minutes even with the larger write buffer.

Also notice I'm getting errors in the SageDCT log:

2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53960 Current: 26486
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53961 Current: 26487
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53962 Current: 26488
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53963 Current: 26489
2011/09/30 13:21:37.64 : RTP Sequence Error Last: 65509 Current: 1119
2011/09/30 13:24:16.71 : ProcessPackets: No data recieved 9/30/2011 1:23:56 PM
2011/09/30 13:24:27.65 : WARN GetSize has not changed: 1422014516 - 9/30/2011 1:24:04 PM

--

Oh and eventually the SageDCT process just dies because it runs out of memory.

What else did you change and how large does your SageDCT process get?

[phelme]

Originally Posted by jbuszkie

Stupid question.. but what/how can you set up to off load to the NIC??

Jim

Look at the configuration of your NIC. It'll be labled something like "Local area connection" Get the properties for it, click on Configure and under the Advanced tab you might see (depends on the NIC) options like Checksum offload, Large Send offload, TCP Offload Engine etc. This indicates to Windows that the NIC can handle some of these tasks rather than the OS.

[babgvant]

Originally Posted by phelme

Yep UNC paths. Have some other network encoders out there that need it.

If you're writing to a local drive via an UNC path you could try the Replace Path feature as long as you only have one recording drive.

[babgvant]

Originally Posted by phelme

Also notice I'm getting errors in the SageDCT log:

2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53960 Current: 26486
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53961 Current: 26487
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53962 Current: 26488
2011/09/30 13:04:40.41 : RTP Sequence Error Last: 53963 Current: 26489
2011/09/30 13:21:37.64 : RTP Sequence Error Last: 65509 Current: 1119

This error indicates that SageDCT is not receiving blocks of data. It is indicative of an overworked network connection or PC.