HD Capture good, until it isn't.

[Hodr]

My problem is that if I record two HD digitial programs at once (which I assume is not re-encoding, but rather a direct stream copy) they will both be fine up until around 45-50 minutes, then they both begin to skip. Sometimes the skipping goes away a short time before the end of the program (55-58 minutes), but usually it does not.

It doesn't appear to be a processor speed issue (using very little CPU as reported in task manager) and if it were a disk / io issue it would show up as soon as I start recording, right?

I disabled comskip to make sure it wasn't running, and I have checked to make sure nothing is running in the background.

This doesn't happen when recording 2 analogue, or 1 of each (to the best of my knowledge).

System:

Vista business (not SP1 as when I upgrade to SP1 my system becomes unstable, perfectly stable without SP1).

MB = FOXCONN A7GM-S AM2+ 780G RT
CPU = AMD|A64 X2 5050E 2.6G AM2 RT
Capture card = Hauppauge 2250

4 Gigs of DDR3 Memory (sorry, don't remember the speed, think its 800 mhz)

1 500 gb SATA (150 I think, not 300).

If it is a disk transfer / seek issue, and I were to add a second disk, is there a way to tie separate tuners to separate disks (so tuner 1, analog or digital records on disk 1, and tuner 2 on disk 2)??? If not, this would be an awesome option.

I would really like to resolve this as I was hoping to add another card so that I could possible record 3 or 4 shows (I only watch a half dozen shows, but they all seem to come on at the same times).

Thank you for reading.

[sub]

I dont really have any solid ideas why it would occur, but maybe try playing the file in VLC to make sure the problem is present there. If the problem is visible there, it will confirm its something happening at record time rather than playback.

[pBS]

it still could be hard drive slowdowns, but only if there's something wrong, 4 digital streams are no problem even on modest systems...
it wouldn't show up immediately always, windows is caching writes and when it runs out of memory it will bog down, and then speed up when it gets caught up...[if it ever does]

a badly fragmented drive will do this, as well as recording to windows drive..
[windows needing to do stuff while recording ]

one note, hard disk slowdowns will show in the recording as frame skips...
but antenna/connection problems will also look similar...
logs may show more..

check task manager while recording looking for cpu hogs and memory usage of each program and overall...

[Hodr]

Just to rule a few things out, the drive is 90+ percent empty, so I don't believe defragmentation is a problem (though I will give it a test). Also, the drive is running at UDMA mode 6, DMA is enabled, and all of the settings to maximize drive speed are checked in the device manager.

I tend to delete the shows after watching, so I don't have any of the problematic ones on hand, so I will check the task manager for usage (CPU / Memory) monday night when I know this situation will crop up again.

And, should the files show the problems under GB-PVR again, I will check them with the VLC player and post back here.

Thanks for the help.,

------------Edit-----------------------> P.S. So, Sub, how about my idea for optionally tieing different recording directories (which could be located on different drives) to each tuner?

[sub]

P.S. So, Sub, how about my idea for optionally tieing different recording directories (which could be located on different drives) to each tuner?

Sorry, but I dont see that happening.

[pBS]

one other thing to check on drives is write caching....need it on for multiple recording..
some sata drives detect as removable, and write caching is off by default for removables..
but sounds like fragmentation isn't a problem...

on that other question, you can point a whole series to another drive with ntfs junctions...
i use it for about a year now and it helps greatly with multiple records...keeps things separate nicely...but only works on a directory level...so whole series only..
http://forums.nextpvr.com/showthread.php?t=38189
on my system i point my timeshift dir, and several series dirs that all record same time, to different drives...that way they don't fragment eachother completely when recording, and effective hard drive throughput is doubled...
[shows recorded at same time to same drive will ALWAYS be fragmented badly..thousands of slices, the packets take turns being written and take forever to defragment]
tho they usually aren't a problem playing back unless it's running low on space..[<1/4 left]

all future shows of same title name are automatically sent to preset junction points...
so only have to setup once..

requires ntfs filesystem tho..[at least one drive must have ntfs, i recommmend having all as ntfs strongly]

it also benefits even if only 1 drive by keeping them separate on partitions...less fragmenting, more control over space, and running out of space...
[if a particular series' drive space runs out, only those shows won't record, the other recordings dirs still have space and will still record..]

and everything looks like normal in pvrx2..looks like it's all in same place..
utils/plugins shouldn't even know the difference..
i made a util to help make the junctions, it's the link above..

[LewE]

Just to rule a few things out, the drive is 90+ percent empty, so I don't believe defragmentation is a problem (though I will give it a test).

You can't tell how much your files are fragmented by the amount of free space. Also, some defraggers report the level of fragmentation by identifying the number of files fragmented as a percentage of all the files on the drive. This is a useless metric. After an evening of recording you may only have five files fragmented, the ones you recorded that evening. They would likely be very fragmented if you recorded more than one program at a time. Picture the operating system writing a second or two's worth of the first program on disk followed by a similar amount of the second program, then back to the first, etc. You will end up with each program file broken into 1-2 second fragments with the two programs interleaving each other.

What I do is every evening I do a defrag only on the files. This does not move everything around to eliminate non-contiguous free space. It takes a lot less time than a full optimization and when you delete the recording you will free up the entire contiguous space that the recording took (1.7GB for a one hour program on my system).